Radio Emission From Cosmic Ray Air Showers

Radio Emission from Cosmic Ray Air Showers
Dissertation
zur
Erlangung des Doktorgrades (Dr. rer. nat.)
der
Mathematisch-Naturwissenschaftlichen Fakult at
der
Rheinischen Friedrich-Wilhelms-Universit at Bonn
vorgelegt von
Tim Huege
aus
Solingen
Bonn 2004
Angefertigt mit Genehmigung der Mathematisch-Naturwissenschaftlichen
Fakult at der Rheinischen Friedrich-Wilhelms-Universit at Bonn
1. Referent: Prof. Dr. Heino Falcke
2. Referent: Prof. Dr. Peter Schneider
Tag der Promotion: 14.12.2004
Diese Dissertation ist auf dem Hochschulschriftenserver der ULB Bonn
http://hss.ulb.uni-bonn.de/diss online elektronisch publiziert.
iii
Summary
Charged particles with energies ranging from a few times 10
6
eV up to
> 10
20
eV continuously impinge on the earth as so-called cosmic rays. Their
study constitutes an important eld of research in both astronomy and parti-
cle physics, and some cosmic ray related eects even have signicance for the
publics every-day life.
Radio emission from cosmic ray air showers oers the opportunity to use
radio observations as an additional powerful observing technique in cosmic
ray research, thereby building a bridge between astroparticle physics and radio
astronomy. As a necessary prerequisite, a detailed theoretical analysis of the
processes responsible for the radio emission has to be performed. In this thesis,
we analyse the emission in the scheme of coherent geosynchrotron radiation
emitted by electron-positron pairs created in the air shower cascade as they are
deected in the earths magnetic eld.
We rst perform an analytic calculation of the emission based on realistic
parametrisations of the particle distributions in the air shower. The analytic
approach allows us to gain a solid understanding of general emission features
and the coherence eects arising from the dierent physical scales present in
the air shower. We compare our predictions with the available historical data
and nd that geosynchrotron radiation can indeed explain the emission.
Afterwards, we conceive and implement a sophisticated Monte Carlo code,
performing the calculation with increased precision and taking into account a
more realistic air shower model. We describe and test the concepts envis-
aged to allow a high-precision modelling of realistic air showers on standard
computer hardware and compare the Monte Carlo results with the analytic cal-
culations to ensure the correctness of our modelling eorts.
In a last step, we use our Monte Carlo code to simulate a large number of
air showers in order to study the eect of important air shower parameters and
geometries on the associated radio emission. Our main result is a parametrisa-
tion formula relating the radio emission characteristics directly to important air
shower and observer parameters such as the air shower geometry, the primary
particle energy, the depth of the shower maximum, the observer distance from
the shower centre, and the observing frequency.
With this analysis, we build the foundation for the interpretation of exper-
imental measurements of radio emission from extensive air showers and thus
make a great step forward in the establishment of radio observations as an
additional observing technique in cosmic ray research.
iv
Contents
1 Introduction 1
1.1 Cosmic Rays . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 Observing techniques . . . . . . . . . . . . . . . . . 2
1.1.2 Cosmic ray induced air showers . . . . . . . . . . . . 4
1.1.3 Cosmic ray measurements . . . . . . . . . . . . . . . 5
1.1.4 Cosmic ray sources . . . . . . . . . . . . . . . . . . 8
1.1.5 Neutrino astronomy . . . . . . . . . . . . . . . . . . 10
1.2 Radio emission from cosmic ray air showers . . . . . . . . . 11
1.2.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2 Historical experiments . . . . . . . . . . . . . . . . . 12
1.2.3 Recent and current experimental eorts . . . . . . . . 13
1.2.4 The LOPES project . . . . . . . . . . . . . . . . . . 14
1.2.5 Historical theory . . . . . . . . . . . . . . . . . . . . 15
1.2.6 Recent theoretical works . . . . . . . . . . . . . . . . 16
1.2.7 This work . . . . . . . . . . . . . . . . . . . . . . . 17
2 Analytic Calculations 19
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 The geosynchrotron approach . . . . . . . . . . . . . . . . . 19
2.2.1 Synchrotron-theory: individual particles . . . . . . . 20
2.2.2 Synchrotron-theory: electron-positron pairs . . . . . . 23
2.3 Observational quantities . . . . . . . . . . . . . . . . . . . . 26
2.3.1 Pulse reconstruction . . . . . . . . . . . . . . . . . . 26
2.3.2 Conversion of |E(R, )| to
. . . . . . . . . . . . . 26
2.3.3 LOPES signal-to-noise . . . . . . . . . . . . . . . . 27
2.4 Extensive air shower properties . . . . . . . . . . . . . . . . 29
v
vi CONTENTS
2.4.1 Longitudinal air shower development . . . . . . . . . 29
2.4.2 Lateral particle distribution . . . . . . . . . . . . . . 30
2.4.3 Particle arrival time distribution . . . . . . . . . . . . 31
2.4.4 Particle energy distribution . . . . . . . . . . . . . . 34
2.5 Coherence: longitudinal eects . . . . . . . . . . . . . . . . 35
2.5.1 Uniform line charge . . . . . . . . . . . . . . . . . . 37
2.5.2 Gaussian line charge . . . . . . . . . . . . . . . . . . 37
2.5.3 Asymmetrical -distribution . . . . . . . . . . . . . . 37
2.5.4 Model calculations . . . . . . . . . . . . . . . . . . . 38
2.6 Coherence: lateral eects . . . . . . . . . . . . . . . . . . . 40
2.6.1 Geometry . . . . . . . . . . . . . . . . . . . . . . . . 41
2.6.2 Approximations . . . . . . . . . . . . . . . . . . . . 41
2.6.3 Integration . . . . . . . . . . . . . . . . . . . . . . . 43
2.6.4 Model Calculations . . . . . . . . . . . . . . . . . . 43
2.7 Flaring disk . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.8 Integration over shower evolution . . . . . . . . . . . . . . . 48
2.9 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.9.1 Theoretical results . . . . . . . . . . . . . . . . . . . 50
2.9.2 Comparison with experimental data . . . . . . . . . . 52
2.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3 Monte Carlo simulations 57
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.2 The Monte Carlo approach . . . . . . . . . . . . . . . . . . . 58
3.2.1 Motivation and objectives . . . . . . . . . . . . . . . 58
3.2.2 General approach . . . . . . . . . . . . . . . . . . . 58
3.3 Implementation details . . . . . . . . . . . . . . . . . . . . . 59
3.3.1 Technical information . . . . . . . . . . . . . . . . . 59
3.3.2 Particle creation and propagation . . . . . . . . . . . 60
3.3.3 Calculating and collecting contributions . . . . . . . 61
3.3.4 Atmosphere model . . . . . . . . . . . . . . . . . . . 62
3.3.5 Random number generation . . . . . . . . . . . . . . 63
3.4 Intelligent concepts . . . . . . . . . . . . . . . . . . . . . . . 64
3.4.1 Cutting o
1
-cones . . . . . . . . . . . . . . . . . 64
3.4.2 Smart trajectory-sampling . . . . . . . . . . . . . . . 64
3.4.3 Intelligent gridding strategy . . . . . . . . . . . . . . 66
3.4.4 Sequentialised and parallelised calculation . . . . . . 68
3.4.5 Automatic ground-bin inactivation . . . . . . . . . . 69
CONTENTS vii
3.5 Data output and reduction . . . . . . . . . . . . . . . . . . . 70
3.5.1 Raw data . . . . . . . . . . . . . . . . . . . . . . . . 70
3.5.2 Spectral ltering . . . . . . . . . . . . . . . . . . . . 71
3.5.3 Further data processing . . . . . . . . . . . . . . . . 72
3.6 Consistency checks . . . . . . . . . . . . . . . . . . . . . . . 73
3.6.1 Individual particle pulses . . . . . . . . . . . . . . . 73
3.6.2 Symmetry N-S and E-W . . . . . . . . . . . . . . . . 76
3.6.3 Gridding algorithms . . . . . . . . . . . . . . . . . . 77
3.6.4 Smart trajectory-sampling . . . . . . . . . . . . . . . 78
3.6.5 Cutting o
1
-cones . . . . . . . . . . . . . . . . . 78
3.6.6 Automatic ground-bin inactivation . . . . . . . . . . 80
3.7 Emission from a shower slice . . . . . . . . . . . . . . . . . 80
3.7.1 Trajectory length eects . . . . . . . . . . . . . . . . 82
3.7.2 Magnetic eld dependence . . . . . . . . . . . . . . 85
3.7.3 Energy distribution eects . . . . . . . . . . . . . . . 87
3.7.4 Comparison with theoretical calculations . . . . . . . 89
3.8 Emission from an integrated shower . . . . . . . . . . . . . . 93
3.8.1 Integration over shower evolution . . . . . . . . . . . 93
3.8.2 Integrated shower results . . . . . . . . . . . . . . . 95
3.8.3 Comparison with theoretical calculations . . . . . . . 100
3.9 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
4 Simulation Results 109
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.2 Simulation parameters . . . . . . . . . . . . . . . . . . . . . 110
4.3 General characteristics . . . . . . . . . . . . . . . . . . . . . 110
4.3.1 Spectral dependence . . . . . . . . . . . . . . . . . . 111
4.3.2 Radial dependence and emission pattern . . . . . . . 113
4.3.3 Wavefront curvature . . . . . . . . . . . . . . . . . . 114
4.3.4 Linear polarisation . . . . . . . . . . . . . . . . . . . 114
4.4 Qualitative dependence on shower parameters . . . . . . . . . 116
4.4.1 Shower zenith angle . . . . . . . . . . . . . . . . . . 116
4.4.2 Shower azimuth angle and polarisation . . . . . . . . 118
4.4.3 Magnetic eld . . . . . . . . . . . . . . . . . . . . . 121
4.4.4 Primary particle energy . . . . . . . . . . . . . . . . 122
4.4.5 Depth of shower maximum . . . . . . . . . . . . . . 124
4.5 Parametrisations for vertical showers . . . . . . . . . . . . . 125
viii CONTENTS
4.5.1 Radial dependence . . . . . . . . . . . . . . . . . . . 128
4.5.3 Polarisation characteristics . . . . . . . . . . . . . . . 129
4.5.4 Combined E
p
and X
max
dependence . . . . . . . . . . 131
4.6 Parametrisations for arbitrary geometry . . . . . . . . . . . . 132
4.6.1 Radial dependence . . . . . . . . . . . . . . . . . . . 132
4.6.3 Dependence of radial scale factor on X
max
. . . . . . 134
4.6.4 Overall parametrisation . . . . . . . . . . . . . . . . 137
4.6.5 Quality and validity of the overall parametrisation . . 139
4.6.6 Comparison with Allan-parametrisation . . . . . . . . 140
4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
4.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5 Conclusions 145
6 Danksagung 161
1
Introduction
Cosmic ray air showers have been known for almost 40 years to emit pulsed
radio emission in the frequency range from a few to a few hundred MHz, an
eect that oers great opportunities for the study of extensive air showers with
upcoming fully digital software radio telescopes such as LOFAR and the en-
hancement of particle detector arrays such as KASCADE-Grande or the Pierre
Auger Observatory. However, there are still a lot of open questions regarding
the strength of the emission as well as the underlying emission mechanism.
Only with a detailed theoretical analysis of the processes leading to the
emission, current and future experimental eorts such as the LOPES project
can make use of their full potential. In this thesis, we build the necessary the-
oretical foundation for the use of radio emission from cosmic ray air showers
as a powerful observing technique in cosmic ray research by modelling the
radio emission from extensive air showers within the scenario of coherent
geosynchrotron radiation.
To motivate the interest in this additional observing technique, which is
largely complementary to the well-established particle detector and air uo-
rescence techniques, we rst give a very short (and by no means complete)
introduction into the eld of cosmic ray physics in general before providing a
quick overview over the merits of the radio technique as well as the historical
and current activities in the eld.
1
2 1.1. Cosmic Rays
1.1 Cosmic Rays
The fact that energetic particle radiation continuously impinges on the earth
was rst discovered by Hess (1912) using an electrometer during balloon
ights. He discovered that the rate with which a static charge on the electrom-
eter diminishes grows as a function of increasing height and inferred that there
had to be ionising radiation coming from space. In 1936, Viktor Hess was
awarded the Nobel Prize for the discovery of these so-called cosmic rays
(CRs).
Cosmic rays mainly consist of protons and ionised nuclei and constitute an
important astronomical window. At the same time, the interaction processes
they undergo fall into the realm of particle physics. Consequently, cosmic rays
form one of the important links between the two elds and indeed are at the
centre of astroparticle physics. Interestingly, although cosmic ray research
has been going on for almost 100 years now, many fundamental questions such
as that of the nature of the CR sources are still unanswered.
Even the public is confronted with cosmic ray-related phenomena in their
every-day life a signicant fraction of the natural radioactivity on earth is
caused by the fragments of cosmic ray induced extensive air showers. Parti-
cle bursts from the sun can knock out important communication infrastructure.
And some scientists even claim to have found evidence for a direct link be-
tween the cosmic ray ux reaching the earth and the global climate (Shaviv
2004).
To appreciate the importance of cosmic ray physics, we give a quick intro-
duction in the following sections.
1.1.1 Observing techniques
Cosmic rays span many orders of magnitude in energy, from low energies of
MeV up to ultra-high energies of 3 10
20
eV (Bird et al. 1995).
Particles at energies below 10
14
eV can be measured directly with par-
ticle detectors mounted on balloons or satellites. Direct measurement of the
primary particles allows detailed analyses, e.g. of the individual particles en-
ergies, their chemical composition, and other properties.
Above this energy threshold, the particle uxes become very low and a di-
rect measurement with balloon or satellite-experiments is no longer economi-
cally feasible, as the eective collecting area is too low. Particles at these ener-
gies, however, can initiate so-called extensive air showers in the atmosphere
(cf. section 1.1.2). The fragments of these air showers arrive at ground-level
1.1. Cosmic Rays 3
and can be measured with ground-based particle detector arrays. One such
experiment for the energy range of 5 10
14
2 10
17
eV is the KASCADE
experiment at the Forschungszentrum Karlsruhe in Germany, which has now
been incorporated into KASCADE-Grande. Another example is the Akeno Gi-
ant Air Shower Array (AGASA) in Japan (The AGASA collaboration 2004),
aimed at ultra-high-energy cosmic rays with energies up to
> 10
20
eV. While
these ground-based arrays have the advantage of huge eective collecting areas
and are thus able to measure even very low cosmic ray uxes, the information
gathered on the primary particles is of only very indirect nature. Reconstruc-
tion of the primary particles energies, let alone their composition, is a very
dicult process requiring state of the art computer simulations of the parti-
cle physics in the air shower development itself at energies far beyond the
regime that can be experimentally tested with particle accelerator experiments.
At yet higher energies of
> 10
18
eV, another technique for the observation
of cosmic rays becomes important: the measurement of uorescence light in
the optical to ultra-violet emitted by atmospheric nitrogen molecules that were
excited during the passage of the air shower. These uorescence measurements
have the advantage of yielding very direct information about the deposition of
energy in the atmosphere and can therefore better reconstruct, e.g., the energy
and composition of the primary particles. The main drawback of this tech-
nique, however, is that it can only be used in very good observing conditions,
i.e., in clear, moonless nights far away from man-made light pollution. The
overall duty cycle therefore is usually only around 10%. Additionally, this
technique requires elaborate modelling of the atmospheric conditions. The
most prominent example for this technique is the High Resolution Flys Eye
(HiRes) experiment in Utah, USA (see, e.g., Sokolsky 2003). Currently, this
technique is adapted for satellite-based monitoring of large fractions of the
earths atmosphere. One such experiment is the EUSO instrument, planned to
be mounted on the International Space Station ISS (Catalano et al. 2003).
The two techniques yield very much complementary information and a
combination into a hybrid approach bears many advantages. Consequently,
(the southern part of) the Pierre Auger Observatory (e.g., Matthews & Pierre
Auger Observatory Collaboration 2004) currently being set up in the Pampa
Amarilla in Argentina will combine a huge array of particle detectors (1600
covering an area of 3100 km
2
) with 24 optical telescopes for uorescence mea-
surements in stereoscopic conguration. The Pierre Auger Observatory will
provide the best statistics so far for ultra-high energy cosmic rays (UHECRs)
> 10
19
eV and will therefore provide the necessary information to address
4 1.1. Cosmic Rays
some of the most important questions regarding UHECRs (cf. 1.1.4).
The radio technique bears many of the advantages of the uorescence tech-
nique, e.g. a more direct and integral nature of the air shower measurements,
but has the additional attractive ability to work 24 hours a day, seven days
a week in almost any environment (Falcke & Gorham 2003). It is therefore
much less demanding regarding the observing site. Additionally, a design such
as that of the LOFAR-based LOPES experiment (Horneer et al. 2004) will al-
low setups with a moderate cost per antenna.
1.1.2 Cosmic ray induced air showers
Auger et al. (1939) discovered that high-energy cosmic rays can initiate exten-
sive air showers (EAS) in the atmosphere. After having traversed an energy-
and composition-dependent atmospheric depth, the primary particle interacts
with a nucleus in the atmosphere and initiates a cascade of particle reactions
producing secondary particles which themselves initiate further sub-cascades.
The overall cascade propagates through the atmosphere with almost the (vac-
uum) speed of light as it evolves to a maximum in particle number and then
diminishes again. A fact important for the coherence of the radio emission
is that the shower remains concentrated in a thin pancake of only a few
metres thickness. The lateral extent of the pancake is mostly due to multi-
ple scattering of the particles and varies for the dierent components in the
air shower: muonic (about 5% of the particles), electromagnetic (electrons,
positrons and photons, about 90% of the particles) and hadronic (pions, kaons,
protons, neutrons and fragments of nuclei, about 1% of the particles). The
cascade fragments arriving at the ground can then be measured by particle de-
tector arrays with great eective collecting areas (cf. section 1.1.1). Figure 1.1
gives a schematic view of an extensive air shower.
To reconstruct information about the primary particle from the fragments
measured on the ground, elaborate Monte Carlo simulation codes such as
CORSIKA (Heck et al. 1998) are necessary. As these simulations use particle
physics models operating in energy regimes far beyond the range experimen-
tally tested with particle accelerators, considerable uncertainty remains. Also,
shower-to-shower uctuations make it very dicult to infer information about
the primary particle of an individual air shower. Fluorescence measurements
have an advantage in this respect as they can measure the energy deposit in
dierent atmospheric depths directly and derive information about the energy
and composition of the primary particle energy from the position and extent of
the so-called shower maximum.
1.1. Cosmic Rays 5
e l . - ma g n .
c o mp o n e n t
hadroni c
component
myonic
component
obser vi ng
l evel
a
t
m
o
s
p
h
e
r
i
c

d
e
p
t
h

1
0
0
0
g
/
c
m
2
zenith angle
primary particle
thickness 1m
shower axis
shower axis
detectors
F 1.1Schematic view of an extensive air shower (Alkhofer 1975).
For a modelling of the radio emission fromcosmic ray air showers, detailed
information about the distribution of electrons and positrons in the air shower
is necessary.
1.1.3 Cosmic ray measurements
In the following, we present a (very minor) fraction of the most interesting
measurements regarding cosmic rays.
As mentioned earlier, the spectrum of cosmic rays spans many orders of
magnitude. Figure 1.2 shows a tentative dierential spectrum of the cosmic
ray ux from energies of 10
11
eV up to
> 10
20
eV as contributed by direct as
well as indirect measurements. The spectrum very closely follows a power-law
of index 2.75 over many decades. It, however, also shows some interesting
features, the most prominent being the so-called knee around 3 10
15
eV
and the so-called ankle near 510
18
10
19
eV. The origin of these features is
the subject of ongoing scientic discussion, and we present some of the most
popular theories in section 1.1.4.
From the direct measurements on satellites and balloons, the composition
6 1.1. Cosmic Rays
kin. Energy E [GeV]
10
2
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
10
11
10
12
]
2

G
e
V
-
1

s
r
-
1

s
-
2

[
m
3

E
d
I
/
d
E
10
5
10
6
10
7
10
2
10
3
10
4
10
5
10
6
10
7
10
8
10
9
10
10
10
11
10
12
KASCADE r a n g e
RUNJOB
JACEE
Sokol-2
Proton-3
EASTOP
Tibet
Hegra
Akeno (1984)
CASA-MIA
CASA-BLANCA
DICE
Akeno (1992)
AGASA
Flys Eye
Haverah Park
HIRES-MIA
gy roradius proton [pc]
10
-4
10
-3
10
-2
10
-1
1 10 10
2
gy roradius iron [pc]
10
-3
10
-2
10
-1
1 10 10
2
10
3
F 1.2Spectrum of the cosmic ray ux, adapted from Maier (2003), for references to
the underlying data see Ulrich (2003). To enhance the features visible in the spectrum, the ux
has been multiplied by E
3
. The gyration radii have been calculated for an interstellar magnetic
eld of 1.4 G.
of cosmic rays at energies up to 10
14
eV (i.e., of Galactic origin, cf. section
1.1.4) is known relatively well. It is very similar to the composition of ele-
ments in the solar system (cf. Fig. 1.3) with some deviations. The increased
abundances for Li, Be and B as well as Sc to Mn in the CRs with regard to the
solar system, e.g., can be explained by spallation of CNO and Fe cosmic rays.
Much attention is currently focused on the ultra-high energy cosmic rays.
Specically, their absolute ux is rather unclear to date. The theoretically mo-
tivated GZK cuto (see section 1.1.4) predicts a diminishing of the cosmic
ray ux at energies above 5 10
19
eV. While the HiRes data indeed seem to
indicate a ux depression in this energy regime, the AGASA data show a con-
tinuation of the spectrum up to energies of > 10
20
eV, as shown in Fig. 1.4. The
statistics, however, are far too bad at these energies to really decide whether
there is a ux depression or not (at energies above 10
20
eV, the estimated ux is
of order one particle per km
2
per century). The Pierre Auger Observatory with
its huge eective collecting area will, however, nally provide the necessary
statistics to address this issue very soon.
1.1. Cosmic Rays 7
10
-5
10
-4
10
-3
10
-2
10
-1
1
10
10
2
10
3
10
4
10
5
10
6
0 5 10 15 20 25 30 35
atomic number Z
r
e
l
a
t
i
v
e

a
b
u
n
d
a
n
c
e
galactic cosmic rays
solar system
Fe
He
C
O
Si
Li
Be
B
Sc
V
Ni
F 1.3Composition of Galactic cosmic rays in comparison with element composition
in the solar system (Wefel 1991). The abundances are normalised to 100 for Si.
F 1.4HiRes data versus AGASA data (Bergman, D.R. for the High Resolution Flys
Eye Collaboration 2003). The t to the HiRes spectra is done by a two-component model taking
into account the GZK eect.
Another feature in the UHECR measurements that is subject to much de-
bate is the presumed small-scale anisotropy of the arrival directions of UHE-
8 1.1. Cosmic Rays
CRs (Takeda et al. 1999) (whereas the large-scale distribution is remarkably
isotropic). There seem to be prominent doublets and triplets of cosmic ray
events arriving from specic directions. These could point back to individ-
ual cosmic ray sources but then again, the underlying propagation models
for cosmic rays in the magnetic elds of the Galaxy themselves are very uncer-
tain. At the highest energies, however, the cosmic rays could indeed point back
directly to their sources, as the gyration radii in the cosmic magnetic elds be-
come very large (cf. Fig. 1.2). One could therefore even do classical astronomy
with these highest-energy particles. The main issue regarding the anisotropy
models to date are, however, again the insucient statistics available so far.
h
12
+60
60
30
+30
0
o
o
o
o
h
h
24
GC
A
G
A
S
A
C
F 1.5Anisotropy above 410
19
eV as presented by The AGASA collaboration (2004).
Red squares and green circles represent cosmic rays with energies of > 10
20
eV, and (4
10) 10
19
eV, respectively. Epoch: Akeno 20 km
2
from Feb. 17, 1990 to Jul. 31, 2002.
Zenith Angles: < 45
. Shaded circles indicate event clustering within 2.5
.
1.1.4 Cosmic ray sources
The sources of cosmic rays are still under much scientic debate. One can,
however, identify dierent regimes in the cosmic ray spectrum: Particles with
energies as low as MeV are of solar origin, and up to energies of 10
10
eV, the
particle ux is strongly modulated by the solar activity. At higher energies, the
sources are presumed to be of Galactic origin, whereas at energies
> 10
18
eV
the particles are no longer conned to the Galaxy by magnetic elds and a
transition to an extragalactic cosmic ray component is suspected.
1.1. Cosmic Rays 9
The classic theory proposes acceleration of charged particles through
diusive shocks in astronomical sources as the origin of cosmic rays (Fermi
1949). One of the most attractive scenarios is acceleration in shock fronts of
supernova remnants (e.g., Lagage & Cesarsky 1983). It has recently gotten
strong support by direct observation of high-energy photons originating from
supernova remnant shells by the HESS -ray telescope (Aharonian et al. 2004).
A number of models involve pulsars (e.g., Harding & Gaisser 1990), some of
them proposing a direct particle acceleration in strong electromagnetic elds
(Cheng et al. 1986).
Most of these models can explain the cosmic ray ux up to energies of
10
15
10
17
eV. This energy regime encompasses the knee feature visible in
the spectrumat energies of 310
15
eV. Although this feature was already dis-
covered by Kulikov & Khristiansen (1959) over 40 years ago, its origin is not
fully understood to date. Since it carries information about the possible sources
of the cosmic rays, many theoretical works concentrate on its interpretation.
One general class of models proposes the knee feature as the superposition
of components dierently accelerated in the source (e.g., Lagage & Cesarsky
1983), whereas other models explain the spectral changes as a consequence of
diusion eects during the cosmic ray propagation in the Galaxy (e.g., Ptuskin
et al. 1993). Erlykin & Wolfendale (1987) propose a single nearby supernova-
remnant as the origin of the knee feature. Furthermore, there is a whole class
of exotic models postulating new particles or changed interaction properties,
explaining the knee, e.g., by changes in the development of the extensive air
showers rather than the cosmic ray ux itself (Nikolsky 1995). Recent results
of the KASCADE experiment (Ulrich 2003; Kampert et al. 2004) favour the
classes of models predicting a rigidity (momentum per unit charge) dependent
scaling of the knee energy position, i.e., the models based on acceleration ef-
fects at the source or diusion/drift eects during propagation in the Galaxy.
Special research eort is currently focused on the UHECRs. The apparent
ux of cosmic rays with energies beyond the so-called GZK cuto visible
in the AGASA data (cf. section 1.1.3) constitutes a mystery. The prediction of
this cuto (which is actually not a hard ux cuto but rather a ux suppression
at the highest energies, the details of which critically depend on the underly-
ing source distribution and particle propagation models) goes back to Greisen
(1966) and Zatsepin & Kuzmin (1966): At energies
> 510
19
eV, protons can
produce pions in interactions with the cosmic microwave background photons.
This eectively limits the path length of UHECRs to
< 100 Mpc (Sigl et al.

1994). Although there are models for acceleration of particles to such high en-
10 1.1. Cosmic Rays
ergies in classical astrophysical sources, e.g., in jets of active galactic nuclei
(Biermann & Strittmatter 1987; Rachen & Biermann 1993), the nature of the
potential UHECR sources remains unclear as there are no obvious candidate
sources in the distance range permitted by the GZK cuto (Elbert & Sommers
1995).
To circumvent the diculties in explaining the presence of UHECRs at
energies beyond the GZK cuto by astronomical sources, a number of more
exotic models have been envisaged. Contrary to the bottom-up scenario of
astronomical sources accelerating the charged particles to ultra-high energies,
these top-down models explain the UHECRs as decay products of supermas-
sive X particles, which themselves are either emitted by topological defects
such as magnetic monopoles created in the early universe or may have been
created directly in the early universe and survived until today (for an overview
see Sigl 2003). An important signature of these processes would be a specic
ratio of cosmic rays to neutrinos as well as TeV -rays. We thus make a very
short excursion to the closely related neutrino astronomy experiments.
1.1.5 Neutrino astronomy
Depending on their origin, the UHECR particles must be accompanied by cer-
tain uxes of high-energy neutrinos and -rays. (We do not discuss the latter
here for the sake of conciseness.) A measurement of the neutrino component
could thus directly dierentiate between the various models, specically the
top-down and bottom-up scenarios. Additionally, neutrinos can be used for di-
rect astronomical observations, because they are uncharged and therefore not
deected in cosmic magnetic elds.
These are, among others, important motivations for many of the experi-
ments trying to measure high-energy neutrinos. The main challenge in these
experiments is a detection in spite of the extremely low cross-section for in-
teraction of neutrinos with matter. Consequently, one needs gigantic detection
volumes as provided, e.g., by the antarctic ice. The AMANDA experiment
(e.g., Halzen 1999) at the south-pole uses photo-multipliers on strings de-
posited in the antarctic ice to look for electromagnetic showers and

Cerenkov
light from the leptons produced by neutrino-induced charged current reactions.
Other experiments such as ANTARES (ANTARES collaboration 1999) and
Baikal (Balkanov et al. 1999) use water as the detection volume.
One experiment very closely related to the measurement of radio emis-
sion from cosmic ray air showers is the RICE project (Kravchenko et al. 2003)
also situated at the AMANDA site in the antarctic. Its goal is to measure ra-
1.2. Radio emission from cosmic ray air showers 11
dio

Cerenkov radiation produced by electromagnetic cascades in the dense ice
arising from the Askaryan-eect (cf. section 1.2.5). The RICE experiment is
therefore the dense media equivalent to the experiments measuring radio emis-
sion from cosmic ray air showers. A related approach is the measurement of
radio emission from the moon arising from neutrino-induced electromagnetic
showers in the lunar regolith (Falcke & Gorham 2003).
In case of horizontal air showers, one can even use the atmosphere as de-
tection volume for neutrinos. Huge air shower experiments such as the Pierre
Auger Observatory can therefore also be used to put limits on the high-energy
neutrino ux.
1.2 Radio emission from cosmic ray air showers
The initial discovery of pulsed radio emission accompanying extensive air
showers was made in the mid-1960ies and triggered intensive research both
on the experimental and the theoretical side. In the late 1970ies, however,
these activities almost ceased completely due to ongoing problems with the
interpretation of the experimental data, technical problems and also the great
success of the alternative observing techniques.
Today, renewed interest in radio emission from cosmic ray air showers
as an additional observing technique for cosmic ray research has arisen. In
the following sections we provide a short overview over the historical as well
as recent developments and motivate why this eld currently experiences its
renaissance. For a more detailed account of the history of radio emission from
cosmic ray air showers, we refer the interested reader to the excellent review
of Allan (1971).
1.2.1 Motivation
Radio measurements of EAS open an entirely new window for the observation
of cosmic rays. The technique has a number of signicant benets. Similar
to the optical uorescence technique, it allows a much more direct view into
the air shower cascade than particle measurements on the ground, yielding
information greatly simplifying the reconstruction of air shower parameters
from particle measurements. The uorescence technique is able to measure
the development of the air shower as it evolves. The radio technique, on the
other hand, mainly measures quantities integrated over the full evolution of
the air shower. (The radio signal thus correlates well with the muon num-
ber measured by particle detector arrays, as almost all muons generated in the
12 1.2. Radio emission from cosmic ray air showers
shower cascade reach ground-level.) A major benet of the radio technique
is that it is not hindered by the need for superb observing conditions (clear,
dark, moonless nights far away from any light pollution) that limits the duty
cycle of optical uorescence detectors to typically less than 10%. For a purely
radio-triggered array with a low number of antennas, radio detection of EAS
should be feasible for energies
> 10
17
eV. With large arrays such as LOFAR
or in combination with external triggering by particle detector arrays such as
KASCADE-Grande (Antoni et al. 2003) or the Pierre Auger Observatory (The
Pierre Auger Collaboration 1996), the study of EAS ranging from 10
15
eV
up to ultra-high energies would be possible (Falcke & Gorham 2003). Conse-
quently, an instrumentation with radio antennas is part of the long-term goals
for the northern hemisphere extension of the Pierre Auger Observatory.
To establish radio observations of cosmic ray air showers as an additional
observing technique, two important prerequisites have to be fullled: First, an
experiment has to demonstrate the feasibility and reliability of measuring radio
emission from cosmic ray air showers in todays environment of high radio-
frequency interference. Second, we need a good theoretical understanding of
the underlying emission mechanisms and a detailed model of the dependences
of the radio emission on the underlying air shower parameters. The former
is one of the goals of the LOPES project (cf. section 1.2.4), while the work
presented here addresses the latter question in the framework of the coher-
ent geosynchrotron emission approach rst proposed by Falcke & Gorham
(2003).
1.2.2 Historical experiments
In the mid-1960s, Jelley et al. (1965) discovered that extensive air showers
(EAS) initiated by high-energy cosmic rays produce strongly pulsed radio
emission at frequencies around 44 MHz. Their measurements were motivated
by the theoretical predictions of Askaryan (Askaryan 1962, 1965). The dis-
covery triggered intensive research, and in the following years a number of
experiments established the presence of radio emission from EAS over the
frequency-range from a few to a few hundred MHz (see Prah 1971 for a very
detailed overview).
The nature of these experiments was simplistic, yet eective. In the early
Haverah Park installation, e.g., as little as two antennas per frequency and
polarisation direction with a receiver system of only a few MHz bandwidth
were connected to oscilloscopes continuously measuring the radio signals. The
Haverah Park particle detector array provided a trigger once an air shower
was registered, causing a still camera to photograph the pulses visible on the
oscilloscopes (Prah 1971). Such simplistic experiments would no longer be
possible today, as the radio frequency interference has grown by orders of
magnitude (in the 1960ies, e.g., the British broadcasting service BBC stopped
its TV transmissions during the nights).
A number of experiments (see, e.g., Allan et al. 1967, 1969; Sun 1975)
soon demonstrated that there is a dependence of the radio emissions polarisa-
tion on the geomagnetic eld, giving emission models postulating a dominance
of the geomagnetic emission mechanism additional weight.
Although a number of independent groups all measured the radio pulses, a
major problem remained: The absolute emission level was very uncertain and
remains unclear even today. While earlier Haverah Park experiments found
emission strengths of order
10 V m
1
MHz
1
(Allan et al. 1971), con-
secutive works reported values of
1 V m
1
MHz
1
or even lower (Allan
et al. 1973, 1975). Other groups inferred yet dierent absolute strengths (e.g.,
Atrashkevich et al. 1975). Some of these discrepancies could be explained by
systematic deviations in the primary particle energy calibration used at that
time. More probably, however, the calibration of the radio receiving systems
themselves is responsible for the discrepancies, especially in case of the dier-
ences between the dierent groups (Atrashkevich et al. 1978). Another prob-
lem associated to the historical data is the sometimes insucient documenta-
tion of details regarding the measured quantities, air shower selection criteria,
and other experimental parameters.
While little doubt remains that radio emission from cosmic ray air showers
has actually been measured in the past, the data on its strength and properties
are still very scarce and uncertain. It is therefore imperative to gather reliable,
well-calibrated data with a new generation of experiments.
1.2.3 Recent and current experimental eorts
Lately, interest in the measurement of radio emission from cosmic ray air
showers has awakened once again. A major driver of this interest have been the
recent advances in digital signal processing. In particular, the advent of digital
radio-interferometers such as LOFAR with their capability to simultaneously
monitor the full sky for transient radio signals promise an eective study of
the radio emission even in todays environments of high radio-frequency in-
terference. The development of a LOFAR prototype station for the dedicated
measurement of extensive air showers consequently constitutes the goal of the
LOPES project described in section 1.2.4. Apart from the LOPES project,
there have been a few other experimental eorts in the recent past.
Green et al. (2003) posted an individual antenna near the CASA/MIA array
in Utah (USA). Their setup, however, was limited by high levels of radio-
frequency interference and was not able to measure radio pulses associated
with cosmic ray air showers. As an upper limit for the emission strength they
found
= 3134 V m
1
MHz
1
.
An ongoing eort is the CODALEMA experiment situated in Nancay,
France (Belletoile et al. 2004). This experiment uses the radio-astronomical
Decametric Array (DAM) in the frequency range from 1100 MHz in con-
junction with high-speed digital oscilloscopes trying to detect radio pulses in
coincidence between multiple antennas. While the CODALEMA experiment
has the advantage of being situated in an extremely radio-quiet environment
with very little radio-frequency interference, it does (at the moment) not have
any independent information on the arrival of cosmic ray air showers. Further-
more, as the experiment only measures one circular polarisation component of
the emission, statistics of the radio pulses alone cannot be used to infer if the
measured radio pulses actually originate from air showers, as will be demon-
strated in the course of this work from the predicted radio emission properties.
The CODALEMA experiment is currently being equipped with particle detec-
tors which will allow to relate the radio pulses to individual air showers. As the
experiment is based on highly sophisticated, expensive, oscilloscope technol-
ogy, it will, however, be very dicult to scale this concept up to great numbers
of antennas for application in giant air shower arrays such as the Pierre Auger
Observatory.
1.2.4 The LOPES project
The aim of the LOPES project (Horneer et al. 2004) is to develop an experi-
ment measuring radio emission from cosmic ray air showers in the framework
of the digital radio-interferometer LOFAR. LOFAR, originally conceived as
a pure radio-astronomical instrument for the low-frequency domain of 10
200 MHz (e.g., R ottgering et al. 2003), provides an ideal basis for observa-
tion of radio emission from extensive air showers as proposed by Falcke &
Gorham (2003). On the one hand, it provides full sky coverage as with a
low-gain antenna design. On the other hand, its signal processing capabil-
ities allow a very ecient noise reduction as in a high-gain antenna design
through digital beam-forming, and additionally facilitate the digital ltering of
radio-frequency interference. The fully digital processing including temporary
buering of the measured radio signals makes it particularly suitable for the
otherwise dicult detection of transient events such as the pulses associated to
air showers. The development of LOPES in the framework of LOFAR allows
the use of many of the LOFAR designs for antennas, readout-electronics, and
other components. As LOFAR itself will consist of tens of thousands of indi-
vidual antennas, one of its major design goals is a low cost per antenna. Con-
sequently, the experience gathered in the development of the LOPES project
will be invaluable for the planned equipment of giant air shower arrays such as
the Pierre Auger Observatory with radio capabilities.
The LOPES experiment currently consists of 30 antennas working as
a phased array in conjunction with the particle detector array KASCADE-
Grande situated at the Forschungszentrum Karlsruhe in Germany. While this
environment suers from high levels of radio-frequency interference espe-
cially from the photo-multipliers of the KASCADE particle detectors them-
selves (Horneer et al. 2004) , operation in conjunction with a particle de-
tector array allows an unambiguous matching of radio signals and air shower
events. This is a fundamental necessity for the study of the radio signal proper-
ties as a function of the air shower parameters, especially in the current phase
of uncertain knowledge of the radio emissions absolute strength.
LOPES has the capability to measure linearly polarised emission, which
is especially useful for the verication of geomagnetic deection of charged
particles as the dominant emission mechanism. The experiment has been tak-
ing quality data since January 2004 and has so far found several dozen radio
pulses that are unambiguously associated to cosmic ray air showers (Horneer,
private communication).
1.2.5 Historical theory
In the early 1960ies, Askaryan (1962; 1965) had already predicted the emission
of radio

Cerenkov radiation fromelectromagnetic cascades in particle showers.
The underlying mechanism is the build-up of a net negative charge excess in
the electromagnetic cascade which then propagates in the ambient medium at
a velocity faster than the corresponding speed of light. These predictions were
one of the motivations for the initial experiments by Jelley et al. (1965).
After the experimental proof for the existence of the radio pulses was
made, a number of authors started working on the theoretical interpretation
of the emission process. Kahn & Lerche (1966) developed an analytical model
taking into account the charge excess as well as geomagnetic emission mech-
anisms. They found that the emission was dominated by a geomagnetic eect
which they described as a continuous charge separation induced by the earths
magnetic eld leading to transverse currents propagating through the atmo-
sphere. However, the air shower model at the basis of their calculations was
severely over-simplied. A number of follow-up works tried to improve on
the Kahn & Lerche model by incorporating more realistic air shower mod-
els. Castagnoli et al. (1969) developed an early Monte Carlo simulation of the
emission in the Kahn & Lerche framework, whereas Fuji & Nishimura (1969)
improved the description of the electromagnetic shower through the use of cas-
cade equations. Other attempts to calculate the radio emission were made by
Colgate (1967) and Allan (as described in Allan 1971).
All these works predicted the geomagnetic mechanism to dominate the ra-
dio emission. (The Askaryan-type charge-excess mechanism, however, still
plays a major role in dense media such as ice and is used e.g. in the RICE
experiment (cf. section 1.1.5) to measure neutrino-induced electromagnetic
showers in ice. It has also been experimentally veried by Saltzberg et al.
(2001) who directly measured the radio emission from electromagnetic cas-
cades in a sand target.)
A major achievement of the historical works was the establishment of the
geomagnetic emission as the dominant source of radio emission from cosmic
ray air showers. None of the historical models, however, reached a level of
sophistication allowing a detailed comparison of theoretical results to experi-
mental data from concrete experiments such as LOPES. While Kahn & Lerche
(1966), e.g., used extremely over-simplied air shower geometries, Castag-
noli et al. (1969) only modelled air showers up to energies of 10
14
eV and the
treatment of Allan (1971) overall remained on a somewhat qualitative, phe-
nomenological level.
Having reached only a limited understanding of the emission mechanism
responsible for the radio emission to date, it is clear that any experimental ef-
fort for the observation of radio emission from cosmic rays must be accompa-
nied by thorough theoretical modelling of the associated emission mechanism.
As many of the details regarding the historical models have never been pub-
lished, taking a fresh approach to this modelling eort seems most promising.
1.2.6 Recent theoretical works
Stimulated by the ideas presented in Falcke & Gorham (2003) and the renewed
experimental activities, a number of authors have recently published theoreti-
cal works on the topic of radio emission from cosmic ray air showers. Apart
from the modelling eorts presented in this thesis, there have been two other
attempts at making predictions about the properties of the radio emission.
Suprun et al. (2003) have calculated the emission from the shower max-
imum of a vertical air shower in the geosynchrotron radiation scenario with
Monte Carlo techniques. A drawback of this work, however, is its missing
integration over the air shower evolution as a whole.
Gousset et al. (2004) have made some qualitative geometrical calculations
to show that inclined air showers produce a much larger radio emission pat-
tern on the ground than vertical showers due to the strongly increased spatial
distance of the shower maximum from the ground.
None of these publications, however, draw a detailed picture of the radio
emission and its dependence on the underlying air shower parameters.
1.2.7 This work
A detailed modelling of the radio emission from cosmic ray air showers as a
function of the underlying air shower characteristics is the goal of this work. It
is specically aimed at providing a theoretical foundation for the interpretation
of the experimental data collected by the LOPES project.
Since many of the historical experimental and theoretical results point to
a geomagnetic process as the dominant emission mechanism for radio emis-
sion from cosmic ray air showers, we concentrate our eort on this eect. We
take the new approach of coherent geosynchrotron radiation as proposed by
Falcke & Gorham (2003), which we feel is an elegant and intuitive description
of the underlying processes. It perceives the radiation as synchrotron pulses
from highly relativistic electron-positron pairs gyrating in the earths magnetic
eld. Radiation emitted at low frequencies is expected to be coherent, as the
emission wavelength is larger than the extent of the air shower pancake (e.g.,
30 metres wavelength vs. a few metres pancake thickness at 10 MHz). Other
mechanisms such as the Askaryan-type

Cerenkov emission can then be in-
cluded in our model at a later time. We develop our model in a number of
separate steps:
In chapter 2, published as Huege & Falcke (2003), we perform an analytic
calculation of the geosynchrotron emission from a vertical air shower. The air
shower properties are modelled realistically with analytic parametrisations of
the important characteristics such as the lateral and longitudinal particle distri-
butions, the particle energy distribution and the air shower evolution as a func-
tion of atmospheric depth. The emission from the individual particles itself
is modelled with well-known synchrotron theory. While a number of neces-
sary approximations limit its accuracy, this analytic analysis provides us with
a solid understanding of the important coherence eects shaping the emission.
With this fundamental understanding of the relevant eects we conceive
and implement a detailed Monte Carlo simulation of the air shower emission.
To retain comparability with the analytic calculations, we model the air shower
properties themselves with the same analytic parametrisations as in the an-
alytic calculations. The individual particle emission, however, is calculated
without the use of any approximations, retaining the full polarisation informa-
tion. Chapter 3, published as Huege & Falcke (2004a), describes the detailed
strategies used in the Monte Carlo approach and demonstrates the consistency
and robustness of the simulations. It nishes with a direct comparison of the
Monte Carlo and analytic results for a vertical air shower.
Having demonstrated the correctness of the Monte Carlo calculations, we
use our Monte Carlo code for an elaborate analysis of the emissions depen-
dence on specic air shower parameters in chapter 4, to be published as Huege
& Falcke (2004b). We achieve this goal by performing detailed simulations
of air showers with various congurations as well as geometries and analysing
the corresponding emission characteristics. To facilitate the comparison of our
simulation results with experimental data, we then parametrise the inferred
dependences into a simple overall formula describing the dependences of the
emission on the main air shower parameters.
The conclusions given in chapter 5 then summarise the achievements of
our modelling eorts so far and give an outlook on the future evolution of our
emission model.
2
Analytic Calculations
2.1 Introduction
In this chapter, we perform an analytical analysis of radio emission from cos-
mic ray air showers based on the new approach of interpreting the emission
process as coherent synchrotron emission from electron-positron pairs de-
ected in the earths magnetic eld (coherent geosynchrotron radiation), as
proposed by Falcke & Gorham (2003); see also Huege & Falcke (2002). Other
than Suprun et al. (2003), who recently simulated geosynchrotron emission
from EAS with Monte Carlo techniques, we pursue an analytical approach to
get a solid understanding of the eects governing the emission. The develop-
ment of a sophisticated Monte Carlo code will then be the second step in our
modelling eorts.
We describe the basis of our approach in some detail in Sec. 2.2 and derive
some observationally relevant quantities in Section 2.3. Sec. 2.4 summarises
the characteristics of the air shower development that are needed for a realistic
modelling of the emission process. In Sec. 2.52.8 we develop our model for
the radio emission from EAS step by step with increasingly realistic geome-
tries, which helps in understanding the coherence eects that play a role in
shaping the emission spectrum and spatial distribution. After a short discus-
sion of the results in Sec. 2.9 we conclude our work in Section 2.10.
2.2 The geosynchrotron approach
Two main emission mechanisms have been proposed in the past for radio emis-
sion from EAS:

Cerenkov radiation from a charge excess moving with a ve-
19
20 2.2. The geosynchrotron approach
locity higher than the speed of light in the traversed medium (the so-called
Askaryan mechanism motivated by Askaryan 1962; Askaryan 1965) and ac-
celeration of charged particles in the earths magnetic eld. While the former
is dominant in case of dense media (Buniy & Ralston 2002; Zas et al. 1992;
Alvarez-Mu niz et al. 2000), polarisation measurements in a number of exper-
iments subsequently supported the dominance of the geomagnetic emission
mechanism for radio emission from EAS in air (e.g., Allan et al. 1967, 1969).
It also seems unavoidable in principle for highly relativistic charged particles
moving in the earths magnetic eld.
Coherent geosynchrotron emission from highly relativistic electron-
positron pairs gyrating in the earths magnetic eld represents an equivalent
scenario to that of the transverse currents of Kahn & Lerche (1966) (and other
geomagnetic mechanisms) but is particularly appealing because it has the ad-
vantage of being based on well-studied and well-understood synchrotron the-
ory, an excellent starting point for the development of our emission model. In
the case of radio emission from cosmic ray air showers, however, coherence
eects as well as non-periodic trajectories that are usually not considered for
synchrotron radiation have to be taken into account.
In order to assess the coherence eects arising from the intrinsic air shower
structure, we rst analyse the emission from a specic point during the air
shower evolution, namely the point of maximum shower development. Only
in the last step we integrate over the shower evolution as a whole, which is
eectively compressed into the radio pulse that the observer receives since
the particles have velocities v c.
At this stage, we do not take into account the Askaryan-type

Cerenkov
radiation. In other words, we set the refractive index of the atmosphere to
unity.
2.2.1 Synchrotron-theory: individual particles
We base our calculations on the formalism developed in Jackson (1975). Any
acceleration of a charge gives rise to electromagnetic radiation. The emission
due to acceleration in the direction of the instantaneous velocity vector is, how-
ever, insignicant compared to that caused by the perpendicular acceleration
(Jackson 1975). As a consequence, any arbitrary particle motion, including the
helical motion of a charged particle in a homogeneous magnetic eld, can be
approximated as an instantaneous circular trajectory with adequate curvature
radius.
Retardation eects caused by the nite speed of light give rise to strong
2.2. The geosynchrotron approach 21
F 2.1Geometry of single particle synchrotron radiation for an observer with line-of-
sight vector n enclosing a minimum angle to the instantaneous particle velocity vector v. The
equivalent curvature radius is given by , and the emission can be conveniently divided into the
components e
and e
. The particle trajectory lies in the x-y plane.

beaming eects for highly relativistic particles. For particles with Lorentz
factor the original dipole emission pattern is beamed into a narrow emission
cone of order
1
semi-opening angle which sweeps over the observer in a
very short time, leading to strongly pulsed emission dominated by frequency
components signicantly higher than the particle gyration frequency.
The geometry of the problem corresponds to Fig. 2.1 if one chooses the
origin of the coordinate system to lie in the point on the particle trajectory
where the angle between instantaneous particle velocity vector v and line of
sight vector n reaches its minimum .
Calculation in the frequency domain circumvents problems arising from
the retardation eects. Jackson denes the quantity A(R, ) as a measure of
the frequency component of the electric eld normalised to unit solid an-
gle . In the far-eld limit (distance R to the observer large compared to the
extent of the particle trajectory, i.e. use of Fraunhofer-approximation is possi-
ble), A(R, ) can be approximated and conveniently divided into the two per-
pendicular components e
and e
dened in Figure 2.1. Retaining the phase

information, A(R, ) can then be written as
A(R, ) =
e
8c
e
i(
R
c
2
)
_
e
() e
()
_
, (2.1)
where the plus-sign is to be used for electrons and the minus-sign for positrons,
e denoting their unit charge. Furthermore
A
() = i
2
3c
_
1
2
+
2
_
K
2/3
(), (2.2)
A
() =
2
3c
_
1
2
+
2
_
1/2
K
1/3
() (2.3)
with
=
3c
_
1
2
+
2
_
3/2
, (2.4)
where = 2 denotes the angular frequency corresponding to the observ-
ing frequency , K
a
denotes the modied Bessel-function of order a, and the
curvature radius of the instantaneous circular orbit is given by
=
vm
e
c
eBsin
(2.5)
with magnetic eld strength B and pitch angle between the particle trajectory
and the magnetic eld direction.
Apart from the adopted far-eld approximations, the derivation of this re-
sult incorporates an integration over a highly oscillatory integrand only part of
which contributes signicantly. This integration is usually conducted using the
so-called method of steepest descents also known as method of stationary
phase (Watson 1944). Jacksons derivation, although somewhat simplied,
is correct as long as the observing frequency is high compared to the gyra-
tion frequency of the particles in the magnetic eld. As the latter is around a
few kHz and we are only interested in observing frequencies > 10 MHz, the
Jackson result is well suited as the basis for our calculations. It also correctly
takes into account that the observer sees only one ash of radiation from each
particle and not the periodic repetition that is associated with synchrotron ra-
diation in the classical sense, since the mean free path length of the particles
of 450 m (at a height of 4 km) is very small compared with the length of a
full gyration cycle of 20 km.
The energy spectrum per unit solid angle of a single gyrating particle, cor-
respondingly, is given by (Jackson 1975)
d
2
I
dd
= 2 | A(R, )|
2
=
e
2
3
2
c
_
c
_
2
_
1
2
+
2
_
2
F 2.2Misalignment between the electron and the positron in an electron-positron pair
no longer allows coherent addition of the individual emissions.
_
K
2
2/3
() +
2
+
2
K
2
1/3
()
_
. (2.6)
Since the energy spectrum is | A(R, )|
2
it grows as N
2
with particle number
N if one assumes fully coherent emission. Given a specic distance to the
observer R the frequency component of the E-eld can be calculated as
E(R, ) =
_
4
c
_
1/2
1
R
A(R, ). (2.7)
For a given (observer-frame) distribution of gyrating particles, the correspond-
ing E(R, ) can then be superposed to calculate the total emission.
2.2.2 Synchrotron-theory: electron-positron pairs
In the air shower, electrons and positrons are created in pairs. The symmetry
arising from the opposite curvature of electron and positron trajectories can
lead to a signicant simplication of the calculation: For an electron-positron
pair with perfectly symmetric trajectories with regard to the observer, the A
contributions from the two particles add up to 2A
, whereas the A
contribu-
tions completely cancel each other.
This is, however, an overly special case which does not adequately repre-
sent the problem we are facing. Depending on the direction from which the
observer sees the particle pair, the cancellation of the A
contributions as well
as the summation of the A
contributions are only partial. Furthermore, as the

pulses emitted by the relativistic particles are very short, there is an inherent
coherence length associated to the emissions of the individual particles. If
there is considerable misalignment between the particles, the resulting phase
dierences destroy the coherence as illustrated in Figure 2.2. Overall, one
would therefore have to quantify the coherence losses and incomplete summa-
tion/cancellation arising from the pairing through a form factor.
A more detailed look at the numbers and characteristics of the particle
distributions in the shower, however, reveals that we may indeed assume full
summation and cancellation of A
and A
for an eective electron-positron

pair without introducing too large an error. This approximation works well
if we no longer look at electron-positron pairs that actually form together but
rather group pairs of positrons and electrons together such that their trajectories
overlap symmetrically as seen by the observer and if we can accomplish this
pairing for the vast majority of particles.
For coherent addition of the positron and electron emission to be possible,
a signicant portion of those parts of the particle trajectories from which the
observer actually receives radiation must overlap. (That part has a length of
110 m for = 60, given by the length over which the instantaneous velocity
vector encloses an angle
< with the observers line of sight.) In a typical

10
17
eV air shower the shower pancake, even somewhat before and after the
showers maximum development, consists of 10
8
particles at any time. Even
if the particles were distributed homogeneously in the shower pancake, this
would lead to a particle density of 1000 m
3
. For the realistic distributions
described in section 2.4, the densities in the dominating centre region are a lot
higher. This illustrates that each particle (except in the unimportant outskirts
of the shower pancake) will a priori have a high number of particles in its
direct vicinity with which it can be paired. The probability that there is a
signicant overlap between the paired particles trajectories is high because
the particles mean free path length of 450 m is considerably larger than the
aforementioned 110 m of the trajectory from which the observer receives
radiation.
Whether a consequent pairing with symmetric trajectories is possible,
however, depends critically on the direction distribution of the particles in-
stantaneous velocity vectors. Throughout this work we assume a -distribution
of the particle velocity directions at any given point in the shower shell, as we
choose the initial velocity vectors to point radially away from the spherical
0 2 4 6 8
0
1
2
3
4
PSfrag replacements
[
1
]
F 2.3Comparison of |A

| (solid) and |A
| (short-dashed) for = 100 MHz, = 60 and

B = 0.3 G. Absolute scale is arbitrary.
shower surface. In this situation, the pairing of particles with symmetric tra-
jectories becomes simple as long as one allows pairing between positrons and
electrons from generations of particles with a certain net oset in generation
time.
In this scenario of high particle density and -distribution of velocity di-
rections, the emission from an eective electron-positron pair can therefore
be approximated as that from a pair with perfectly symmetric trajectories:
E
p
(R, )
_
4
c
_
1/2
1
R
2e
8c
e
i(
R
c
2
)
_
e
_
A
(). (2.8)
The fact that |A
| > |A
|, especially for small where most of the radiation

is emitted (Fig. 2.3), furthermore demonstrates that A
is not dominating the

emission, anyway, and therefore gives further condence in the approximation.
Eectively, this result allows us to drop the dierentiation between
positrons and electrons and to consider only generic particles hereafter. The
spectrum emitted by such an individual particle then corresponds to:
E(R, ) =
_
4
c
_
1/2
1
R
e
8c
e
i(
R
c
2
)
_
e
_
A
(). (2.9)
Superposition of these spectra for all particles in the shower, correctly taking
26 2.3. Observational quantities
into account the phase dierences arising from their relative positions, then
yields the emission from the air shower as a whole.
2.3 Observational quantities
We present a number of relations of the previous results to observational quan-
tities:
2.3.1 Pulse reconstruction
The time-dependence of the electromagnetic pulse corresponding to a given
spectrum E(R, ) can easily be reconstructed for a specic receiver bandwidth
by an inverse Fourier-transform of the remaining spectrum. Hence, if the fre-
quency characteristic of the receiver is given by b(), the time-dependence of
the electric eld E(R, t) can be calculated as
E(R, t) =
1
2
_

b()E(R, ) e
it
d, (2.10)
where E(R, ) is given by the complex conjugate of E(R, ).
2.3.2 Conversion of |E(R, )| to
In the works of the 1960ies and 1970ies, the strength of the measured radio
emission was usually denoted with a quantity
in units of V m
1
MHz
1
,
which was dened as the peak electric eld strength during the pulse divided
by the eective bandwidth of the receiver system . In practice, the total
pulse amplitude (in V) at a given observing frequency was derived from the
photographed oscilloscope traces of the two polarisation directions and then
converted to an electric eld strength (in V/m) taking into account the receiver
and antenna gain. This eld strength, representing the projection of the electric
eld vector on the horizontal plane, was then back-projected to yield the eld
strength in the plane perpendicular to the shower axis and the magnetic eld,
in which the electric eld vector lies (see Eq. (2.55) for = 0). Division of the
resulting eld strength by the eective bandwidth of the receiver system
then yielded
.
To compare our theoretical values of |E(R, )| to the experimental results
for
, we analytically reconstruct the time-dependence of the electromagnetic

eld pulse E(t) for the case of an idealised rectangle lter of bandwidth ,
over which |E(R, )| is assumed to be constant, and which is centred on the
2.3. Observational quantities 27
observing frequency . After time-averaging over the high-frequency oscilla-
tions,
then directly follows from the peak eld amplitude divided by and
is given by
=
_
128
|E(R, )| 6.4 |E(R, )| . (2.11)

2.3.3 LOPES signal-to-noise
In order to compare our predictions to the abilities of LOPES (or any other
generic dipole array), we rst estimate the expected signal-to-noise ratio
(SNR) for a receiving system consisting of an individual inverted V shape
dipole antenna with gain G = 1.9 and a receiver incorporating a lter with
bandwidth centred on the observing frequency , a square-lawdetector (i.e.,
a detector measuring the received power) and an integrator which averages the
signal over a time .
The noise level of the receiving system, in our case dominated by Galactic
noise, can be characterised by the noise temperature T
sys
T
sky
(). Compari-
son with the temperature increase T corresponding to the power of the pulse
intercepted by the antenna then yields
SNR =
2
T
T
sys
, (2.12)
where the rst factor takes into account the increase of the SNR due to the
number of independent samples accumulated in case of bandwidth and av-
eraging time as determined by the Nyquist theorem. The energy ux of an
electromagnetic wave propagating through the (vacuum-like) atmosphere is
given by the Poynting vector, in SI-units and omitting the argument R for the
elds being dened as
S(t) = E(t) H(t) =
1
0
E(t) B(t), (2.13)
where
0
= 4 10
7
Vs/Am. As E B, it follows that
|S(t)| =
1
c
0
|E(t)|
2
1
120
|E(t)|
2
. (2.14)
For a point-like source, the eective area of a dipole antenna is given by
(Rohlfs & Wilson 1996) A
e
= G
2
/4 = Gc
2
/4
2
, so that it receives the
28 2.3. Observational quantities
power
P(t) =
1
2
A
e
|S(t)| =
Gc
8
2
0
|E(t)|
2
, (2.15)
where the factor 1/2 is introduced for an antenna measuring only one polari-
sation direction of unpolarised radiation. Averaging over the time then leads
to
<P>
=
Gc
8
2
_

0
|E(t)|
2
dt
Gc
8
2
_

h
E(
2
d
, (2.16)
where
h/l
= 2( 1/2) and the last step follows from Parsevals theorem
as long as the bulk of the pulse is sampled in the averaging time . Assuming
that the spectrumis at over the observing bandwidth with a value |E(
)|
|E(2)| = const. and using Eq. (2.11), we can write
<P>

Gc
8
2
|E(2)|
2
2
=
Gc
4
2
128
. (2.17)
This averaged power is then directly related to T via the Boltzmann-constant
k
B
by
T =
<P>
k
B
, (2.18)
so that from Eq. (2.12) follows
SNR
Gc
256

2
2
0
k
B
_
|
2
T
sys
. (2.19)
Setting to a sensible value of 2
1
, we get
SNR 0.5
_
G
1.9
_ _

60 MHz
_
2
_
T
sky
()
4000 K
_
1
_

35 MHz
_ _

1 V m
1
MHz
1
_
2
(2.20)
for a typical LOPES antenna with an observing bandwidth of 35 MHz cen-
2.4. Extensive air shower properties 29
tred on the observing frequency 60 MHz and an estimate of T
sky
(60 MHz)
4, 000 K (Falcke & Gorham 2003).
For a complete LOPES array consisting of N
ant
antennas, the SNR is then
increased by an additional factor

1/2N
ant
(N
ant
1)
1/2N
ant
for large
N
ant
.
2.4 Extensive air shower properties
Extensive air showers can be initiated by primary particles with strongly dier-
ing energy and composition and at variable inclination angles. Consequently,
their properties such as the position of their maximum development and the
longitudinal and lateral distributions of secondary particles can vary signi-
cantly.
Additionally, the simulation of air showers consisting of > 10
8
particles
with energies in the MeV range created in a cascade initiated by primary par-
ticles of energies as high as 10
20
eV, is in itself a very dicult process. There
are elaborate Monte Carlo simulations such as CORSIKA (Heck et al. 1998)
and AIRES (Sciutto 1999) which themselves incorporate a number of dierent
models for the underlying particle interactions. But although these codes are
very sophisticated, uncertainties remain, especially at the very highest ener-
gies that are out of the reach of accelerator experiments (see, e.g., Knapp et al.
2003).
At this stage, however, we do not incorporate the results of elaborate air
shower simulations. We rather revert to the widely used analytical parametri-
sations for the longitudinal development and lateral particle distributions dat-
ing back to Greisen (1956), Kamata & Nishimura (1958) and Greisen (1960)
which are admittedly crude, but as a rst step seem adequate to describe the
properties relevant to our calculations that an average air shower would have
in case of vertical inclination. (For an overview see, e.g., Gaisser 1990.)
2.4.1 Longitudinal air shower development
The longitudinal air shower development can be parametrised by the so-called
shower age s as a function of atmospheric depth X:
s(X) =
3X/X
0
(X/X
0
) + 2 ln(E
p
/E
crit
)
=
3X
X + 2X
m
(2.21)
where X
0
= 36.7 g cm
2
denotes the electron radiation length in air,
E
crit
= 86 MeV corresponds to the threshold energy where ionisation losses
30 2.4. Extensive air shower properties
equal radiation losses for electrons moving in air, and X
m
= X
0
ln(E
p
/E
crit
)
marks the theoretical value for the depth of the shower maximum in this
parametrisation. The shower commences at s = 0, builds to its maximum de-
velopment at s = 1 and then declines over the range s = 13. Although orig-
inally developed for purely electromagnetic showers, the formula is suitable
to qualitatively describe the average development of the clumpier hadronic
showers as well. The theoretical X
m
value does then, however, not correspond
to the actual position of the shower maximum. For purely electromagnetic
showers, the development of the total number of charged particles (almost
purely electrons and positrons) can be described by
N(s) =
0.31 exp
_
(X/X
0
)(1
3
2
ln s)
_
_
ln(E
p
/E
crit
)
(2.22)
as a function of shower age. The predicted value for N in the shower maximum
(s = 1) is very close to the Allan (1971) rule of thumb N
max
= E
p
/GeV =
10
8
for a 10
17
eV shower. For the position of the shower maximum X
max
we
refer to the measurements and simulations presented in Knapp et al. (2003) that
suggest a value of X
max
630 g cm
2
which corresponds to R
0
4 km for a
10
17
eV air shower and to the works of Pryke (2001) as well as Abu-Zayyad
et al. (2001).
2.4.2 Lateral particle distribution
The lateral particle density can be described with the NKG (Nishimura-
Kamata-Greisen) parametrisation, which without normalisation corresponds
to
NKG
(r) =
1
r
2
M
(4.5 s)
2(s)(4.5 2s)
_
r
r
M
_
s2
_
1 +
r
r
M
_
s4.5
. (2.23)
To avoid the unphysical singularity of the NKG prole at the shower centre we
cut o the distribution with a constant value at radii smaller than 0.1 m. The
normalisation for the integration is chosen correspondingly (see Sec. 2.6.3).
The parameters relevant to the NKGdistribution, shower age s and Moli ` ere
radius r
M
, show a high degree of degeneracy. The increase in s during the
shower propagation broadens the lateral distribution, but at the same time the
decrease of r
M
with increasing atmospheric density tends to narrow it. One
can therefore often parametrise a given lateral particle distribution with a wide
range of dierent values for s and r
M
, where r
M
in fact need not be close to
the theoretical Moli` ere radius at all (Antoni et al. 2001). We here stick to the
theoretically motivated values of s = 1 for the shower maximum and set r
M
to
the Moli` ere radius at the atmospheric height of the maximum derived from the
atmospheric density as (Dova et al. 2003)
r
M
(h) = r
M
(h
0
)
atm
(h
0
)
atm
(h)
= 9.6
g cm
2
atm
(h)
. (2.24)
According to the US standard atmosphere of 1977 as implemented in COR-
SIKA (Ulrich 1997) the atmospheric density at a height of 4 km corresponds
to
atm
= 0.82 mg cm
3
, which in turn yields a Moli` ere radius of r
M
117 m.
2.4.3 Particle arrival time distribution
Knowledge of the arrival time distributions of particles in the air shower is
necessary to parametrise the curvature and thickness of the shower front as
a function of radial distance to the core. Unfortunately, the development of
the particle arrival time distributions during the shower evolution is not well
established. Agnetta et al. (1997) have analysed Haverah Park data of more
than 450,000 air shower events. These lie in the adequate energy range of
10
17
eV, but were measured at an altitude of 220 m and cannot dierentiate
between e
and
. They, however, still trace the distribution of e
correctly
because the number of e
by far exceeds the number of
. An earlier analysis
of Volcano Ranch data by Linsley (1986) reects the particle distribution at an
altitude of 1,800 m, but is based on a very low number of events (especially
at the radial distances up to a few hundred metres relevant to our model) and
only determines the shower thickness and not the functional form of the arrival
time distributions. We therefore base our model on the Agnetta et al. (1997)
data and use the Linsley (1986) data only for comparison.
In Agnetta et al. (1997) the measured arrival time distribution at a given ra-
dial distance is tted with a -probability distribution function (-pdf) dened
as
f (t) = A t
B
exp(Ct), (2.25)
the form of which (cf. Fig. 2.4) arises from multiple scattering events during
the shower propagation. While A only comprises a normalisation factor, the t
0 10 20 30 40 50
0
0.02
0.04
0.06
0.08
PSfrag replacements
t [ns]
F 2.4-pdf determining the arrival time distribution of particles as measured by Ag-
netta et al. (1997). Solid: in the shower centre, short-dashed: 50 m from shower centre, long-
dashed: 100 m from shower centre.
parameters B and C of the -pdf are directly related to the mean arrival time
< t > and the standard deviation
t
of the measured arrival time distributions
(Bury 1975),
B =
_
<t >
t
_
2
1 and C =
<t >
2
t
. (2.26)
The dependence of < t > and
t
on the radial distance to the shower core is
modeled by a generalised paraboloid of the form
<t >(r),
t
(r) = F + G (r/r
0
)
H
(2.27)
where r
0
is set to the Moli` ere radius at ground level of 79 m. The parameter
sets for <t >(r) and
t
(r) are listed as
F
t
= (8.039 0.068) ns
G
t
= (5.508 0.095) ns (2.28)
H
t
= 1.710 0.059
and
F
= (5.386 0.025) ns
G
= (5.307 0.032) ns (2.29)

0 50 100 150 200
0
10
20
30
40
50
PSfrag replacements
radial distance from shower core [m]
t
[
n
s
]
F 2.5Radial dependence of the particle arrival time distribution. Solid: shower cur-
vature as given by a spherical surface with K = 2, 300 m, short-dashed: < t > (r) as given by
Agnetta et al. (1997), long-dashed:
t
(r) as given by Agnetta et al. (1997), dash-dotted:
t,L
(r)
as given by Linsley (1986).
H
= 1.586 0.020.
Fitting the arrival time distribution with a -pdf partially cuts o the long tail
of particles arriving with very high delay. Since the radio emission is, how-
ever, dominated by the bulk of the particles, the eect is negligible for our
calculations.
The thickness of the shower pancake is directly determined by
t
. The
eective curvature of the shower front is governed by two factors. On the
one hand, there is a delay of the rst particles of the -pdf arriving at distance
r from the shower core with respect to the rst particles arriving in the shower
centre. This eect is not included in the Agnetta et al. (1997) data. Here we
assume that the delay is negligible for the shower distances
< 100 m that we

are interested in (the rst particles can, with good agreement, be assumed to lie
on a at surface). On the other hand, the mean particle delay rises as one goes
to greater distances from the shower core, a fact represented by the increase of
<t> (r). The shower curvature determined by <t> (r) can be expressed very
well with a spherical surface of curvature radius K = 2, 300 m, as can be seen
in Figure 2.5.
For comparison, we also examine the Linsley (1986) parametrisation for
t
, which is dened as
t,L
= G
L
(1 + r/r
L
)
H
L
, (2.30)
where, for a 10
17
eV vertical air shower, we have G
L
= 1.6 ns, r
L
= 30 m
and H
L
= 1.68 0.14. As Linsley (1986) does not specify the functional form
of the arrival time distribution, we assume that it also corresponds to a -pdf.
However, since < t > and
t
are not independent in this parametrisation, we
have to assume a <t > (r) that ts the
t
(r)-distribution given by Linsley. From
the fact that
<t >=
1 + B
t
(2.31)
and the relative constancy of B(r) in the Agnetta et al. (1997) data, a natural
choice for the distribution is given by
<t >
L
(r) =
<t >
A
(r)
t,A
(r)

t,L
(r), (2.32)
where the index A refers to the Agnetta parametrisations and L refers to the
Linsley parametrisations.
2.4.4 Particle energy distribution
The average energy of the electrons and positrons in an air shower corresponds
to 30 MeV, i.e. 60 (Allan 1971). As a very crude approximation, one
can therefore adopt a mono-energetic conguration of particles with 60.
To illustrate the eects induced by a more realistic particle energy distribution,
we compare this with a (spatially uniform) broken power-law distribution of
particle energies where dN/d rises linearly with , peaks at
0
= 60 and then
declines as
2
:
p() =
_

1
_
u
_
1 e
(/
1
)
wu
_
, (2.33)
where we set u = 1, w = 2 and
1
= 74.2 which corresponds to a peak of
the distribution at
0
= 60. One can then calculate the emission of an energy
averaged particle through
E
(R, ) = p
0
_

max
min
p() E(R, ) d, (2.34)
2.5. Coherence: longitudinal eects 35
where the normalisation constant p
0
is
p
0
=
1
_

max
min
p() d
. (2.35)
This energy integration leaves the number of particles unchanged. Note, how-
ever, that the total amount of energy in the particles varies with changing
min
and
max
. Additionally,
min
must not be chosen too small as our deriva-
tions include approximations that are only valid in the ultra-relativistic case.
In general, the presence of high-energy particles amplies the emission near
the shower centre, whereas low-energy particles enhance the radiation at high
distances due to their wider beaming cone.
We will continue to compare results with both mono-energetic and broken
power-law particle distributions and dierentiate the two cases through the ab-
sence or presence of an additional index . Any result calculated for a broken
power-law distribution, indicated through an index , also applies to the mono-
energetic case if one substitutes the energy-averaged E
(R, ) by the original

E(R, ).
2.5 Coherence: longitudinal eects
Having established the emission from individual particles and the spatial dis-
tribution of particles in the air shower, we can now calculate the emission
from the shower maximum, taking into account the inherent (observer-frame)
shower structure. The phase dierences between the pulses from the individual
particles lead to strong coherence eects that signicantly change the spectrum
of the received emission from that of a fully coherent synchrotron pulse. (For
a general discussion of coherence eects regarding synchrotron radiation see
also Aloisio & Blasi 2002.)
To get a better understanding of the spectral features, we start with a
strongly simplied conguration that neglects any lateral structure: We re-
duce the shower to its core. In this approximation, the air shower pancake
of thickness d is collapsed to a one-dimensional line of length d. The charges
distributed along the line are adopted to move synchronously, i.e. the momen-
tum distribution of the particles at a given time corresponds to a -function.
The emission from a particle at distance R = nR from the observer is given by
E(R, ) as dened in Eq. (2.9). For a particle oset by a distance x from the
36 2.5. Coherence: longitudinal eects
line-centre (located at R
0
) along the shower core direction

l, we therefore have
E(R
0
+ x
l, )
1
|R
0
+ x
l|
exp
_
i|R
0
+ x
l|
c
_
A
()
1
|R
0
|
exp
_
i|R
0
+ x n|
c
_
A
()
=
1
R
0
exp
_
i(R
0
+ x)
c
_
A
(). (2.36)
Here we keep the distance of the particle at the constant value R
0
for the rst
factor, which only introduces negligible errors since d R
0
. The approxima-
tion of

l n in the second factor is justied since A
only gives signicant

contributions if the directions of

l and n enclose angles of order
1
or smaller.
In other words, projection eects do not play a signicant role because only in
the regime where they are very small, we have signicant emission from the
particles.
Dening the particle distribution function f (x) such that
_
+
f (x) dx = 1,
and taking into account the particle energy distribution, the integrated emission
from N particles is then given by
E
N
,l
(R, ) =
_
+
N f (x) E
(R
0
+ x
l, ) dx
NE
(R
0
, )
_
+
f (x) e
i
x
c
dx
= NE
(R
0
, ) S (). (2.37)
Note that this basically corresponds to a Fourier transformation, i.e. the func-
tion S () modulating the eld strength spectrum is given by the Fourier trans-
formof the particle distribution function, as in standard diraction theory. (The
energy spectrum is then modulated by |S ()|
2
.)
We will now compare a number of dierent distributions of particles along
the line to better understand the coherence eects that arise from longitudinal
distributions of particles.
2.5.1 Uniform line charge
The easiest case of a line charge is a uniform distribution of particles along a
line of length d,
f (x) =
_
1/d for |x| d/2
0 for |x| > d/2
. (2.38)
Integration over x then leads to the well-known (sin z/z)
2
modulation of the
energy spectrum that corresponds to the diraction pattern of a rectangular
opening,
S () =
_
+
d
2
d
2
1
d
e
i
x
c
dx =
sin (d/2c)
d/2c
. (2.39)
2.5.2 Gaussian line charge
A more realistic case is that of a Gaussian distribution of particles along the
line. The width of the distribution is set by the standard deviation of the Gaus-
sian (the FWHM then corresponds to

4 ln 4 2.35 ), with the distri-
bution being dened as
f (x) =
1
2
exp
_
1
2
x
2
2
_
. (2.40)
The coherence function then equals
S () = exp
_
1
2
2
c
2

2
_
, (2.41)
i.e. a Gaussian as well, which is clear from the fact that the Fourier transform
of a Gaussian is again a Gaussian.
2.5.3 Asymmetrical -distribution
A realistic longitudinal particle distribution is given by an arrival-time distri-
bution as specied by Eq. (2.25) with the substitution x = ct,
f (x) =
_
_
A
_
x
c
_
B
exp
_
C
x
c
_
for x > 0
0 for x 0
, (2.42)
where from the normalisation of f (x) follows
A =
_
C
(1+B)
(1 + B)
_
1
. (2.43)
38 2.5. Coherence: longitudinal eects
The corresponding coherence function S () is then given by
S () =
_
1 +
2
C
2
_
1
2
(1+B)
exp
_
i(1 + B) arctan
_
C
__
exp (i <t >
A
) , (2.44)
where the last phase factor is needed to centre the asymmetrical distribution
on the curved shower surface to make it comparable to the symmetrical uni-
form and Gaussian distribution for the later calculations taking into account
lateral structure. (The origins of the -pdfs then again lie on a at surface as
discussed in Section 2.4.3.)
2.5.4 Model calculations
The results derived so far allow us to perform a number of model calculations
in order to analyse the eects of longitudinal particle distributions on the ob-
served spectra as well as the dependence of the emission on the observers ra-
dial distance from the shower core. Where no analytical result was presented,
integrations and other calculations are done numerically. We model the max-
imum of a vertical air shower with primary particle energy E
p
= 10
17
eV and
therefore N = 10
8
particles at a height of R
0
= 4 km. This is a realistic value as
outlined in Section 2.4.1. The earths magnetic eld is adopted with a strength
of B = 0.3 G and, for simplicity and symmetry reasons, assumed to be perpen-
dicular to the shower core and thus parallel to the earths surface (a realistic
value for central Europe would be B = 0.5 G with declinations around 70
).
The thickness of the air shower pancake is determined by the standard
deviation
t
as parametrised in Sect. 2.4.3. To ensure an equivalent width of
the -pdf, the uniform line charge and the Gaussian line charge, we normalise
the latter distributions such that they have a standard deviation of c
t
. For the
Gaussian distribution, this corresponds to = c
t
. The uniform line charge
must be set to a total length of d = 2
3c
t
. Evaluated in the shower core,
c
t
(0) = 1.61 m, which results in d = 5.6 m.
Fig. 2.6 compares the spectral modulations arising from the dierent lon-
gitudinal particle distributions. If the particles radiated fully coherently i.e.
moved as one particle on the exact same trajectory the eld strength spec-
trum of the emission would simply be a synchrotron spectrum enhanced by a
factor N. The coherence eects modulate this spectrum by the coherence func-
tion S (). In the case of the uniform line charge, we see the rst coherence
minimum at 54 MHz, which corresponds to c/d. The Gaussian line charge
10 20 50 100 200 500 1000
0.01
0.1
1
PSfrag replacements
[MHz]
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.6 |E(R, 2)|-spectrum in the centre of the area illuminated by the maximum of
a 10
17
eV air shower with R
0
= 4 km and 60. Solid: full coherence, short-dashed: uni-
form 5.6 m line charge, long-dashed: Gaussian line charge with = 1.61 m, dash-dotted:
asymmetrical -distribution with c
t
= 1.61 m
spectrum does not exhibit such a sharp minimum, but is strongly attenuated at
higher frequencies. The asymmetrical -pdf lies between these two simplied
models.
Obviously the longitudinal eects very strongly modulate the emitted spec-
trum at high frequencies (
> 50 MHz) and therefore are an important limiting

factor for the choice of a suitable observing frequency. The thickness of the
air shower pancake has a very profound and direct inuence on the emitted
radiation and could therefore be probed directly through observations of radio
emission from EAS at frequencies > 50 MHz.
Another interesting characteristic of the radiation is its radial dependence
at a given frequency as illustrated by Fig. 2.7 for the case of full coherence
without any longitudinal distribution, i.e. for particles concentrated in a point
source. In this case, the associated emission pattern is purely governed by
the inherent emission pattern of the synchrotron pulses. The extent of the
illuminated area on the ground is an important characteristic that ultimately
limits the probability to detect very scarce ultra-high energy cosmic ray air
showers with a given collecting area.
As expected, adoption of the broken power-law distribution of particle en-
ergies inuences the radial emission pattern. Going from the mono-energetic
case to the broken power-lawenergy distribution mainly amplies the emission
40 2.6. Coherence: lateral eects
200 400 600 800
0.01
0.1
1
PSfrag replacements
distance from shower centre [m]
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.7Radial dependence of |E(R, 2 50 MHz)| for the maximum of a 10
17
eV point
source shower with R
0
= 4 kmfor the 60 case (solid) and for broken power-lawdistributions
from = 5120 (short-dashed), = 51000 (long-dashed) and = 510000 (dash-dotted).
in the centre region due to the presence of high energy particles that radiate
more strongly, but into a smaller beaming cone. At the same time, the low-
energy particles amplify the emission at very high distances due to their wider
emission pattern. The drop in the number of medium-energy particles is cor-
respondingly reected in a drop of the emission on medium scales. Obviously,
there is only negligible dierence when increasing the upper limit
max
from
a value of 1000 to higher values such as 10000. For the remaining calcula-
tions, we therefore adopt a distribution with in the range 51000 to minimise
computation time.
2.6 Coherence: lateral eects
After having analysed coherence eects arising from longitudinal distributions
of particles, we now take a look at the inuence of the lateral structure of the
air shower on the radio emission. This we accomplish by smearing out the
line charge considered so far over a segment of a spherical surface with (for
the moment) constant thickness d. Inside this shell we continue to consider
the types of longitudinal particle distributions introduced in Section 2.5.4.
2.6. Coherence: lateral eects 41
F 2.8The geometry for the air shower maximum.
2.6.1 Geometry
The geometry of the air shower maximum is dened as in Fig. 2.8 and char-
acterised by the curvature radius of the shower surface K, the shower shell
thickness d and the shower inclination angle . The observer is positioned on
the x-axis at a minimum distance R
0
from the shower surface, with an inclina-
tion angle
0
to the shower core. The magnetic eld strength B, inclination
B
and azimuthal direction
B
determine the conguration of the earths magnetic
eld.
To derive the total emission from the air shower maximum, we now have
to integrate over the shell and hence must relate E(R, ) and consequently the
quantities going into E(R, ) to the position on the surface as given by and
. We refer the reader to the appendix for the details of these calculations.
2.6.2 Approximations
In order to facilitate the integration over the shower shell we apply a number
of approximations. First, we sum the contributions from the dierent regions
of the air shower maximum in a scalar way, i.e. we do not evaluate Eq. (2.55)
but set e
(, ) e
(0, 0), which is justied due to the minute changes in

the direction of e
(, ) over the shower surface. The general polarisation di-

rection of the radiation then is perpendicular to both the shower axis and the
magnetic eld direction. Second, as pointed out in the appendix, we assume
that the instantaneous velocity vectors of the particles in the shower shell are
perpendicular to the shower surface at the moment corresponding to the ori-
gin of Figure 2.1. This again corresponds to a -function for the distribution
of the particle momenta, and in this strict sense, the minimum angle to the
line-of-sight (, ) is given by Eq. (2.54).
Adoption of this , however, yields a very conservative estimate for the
emission. (, ) as calculated by Eq. (2.54) overestimates the minimum an-
gle to the line-of-sight as dened in Fig. 2.1 in case of a particle trajectory
bending towards or away from the observer, where a signicantly reduced
is reached during a later or earlier position on the particle trajectory. The
amount of compensation in attainable by this eect is considerable since
the ratio of mean free path length to curvature radius of = 60 electrons is
450 m/3400 m 8
1
. is therefore eectively reduced to its irreducible
component given by
sin (, ) =

B n(, ). (2.45)
Adoption of this value for (, ) yields a more realistic estimate of the emis-
sion from the air shower shell, and at the same time takes into account the
asymmetry of the emission pattern in that arises from the magnetic eld
conguration. Without a more precise criterion for the maximum compensa-
tion achievable, however, the radial dependence of the emission pattern at very
high distances is obviously no longer valid. We therefore continue to work
with both the conservative approach using Eq. (2.54) and the reduced def-
inition in order to compare the two cases.
The change in R associated to the adoption of the optimum position on
the particle trajectory is negligible because of the following reasons:
the compensated angles are small, therefore the changes in R are small
additional attenuation/amplication through the 1/R-term is thus negli-
gible
there is no signicant change of phase since the particle velocity v c
and the trajectory is only mildly curved.
2.6. Coherence: lateral eects 43
2.6.3 Integration
Using these geometrical relations, the integrated spectrum of the emission
from the air shower maximum with N particles can be calculated as
E
N
,S
() =
0
_
2
0
d
_
r
M
/K
0
d K
2
sin

NKG
(r(, )) E
(R(, ), ) (2.46)
with the normalisation factor
0
= N
__
2
0
d
_
r
M
/K
0
d K
2
sin
NKG
(r(, ))
_
1
. (2.47)
Cutting o the integration at = r
M
/K signicantly reduces computation time
while giving acceptable accuracy as
> 80 % of the particles are included in this

region. The remaining particles are redistributed over the integration region by
the normalisation according to the NKG-prole, which might lead to a slight
overestimation of the emission strength near the shower centre.
2.6.4 Model Calculations
We again examine the frequency and radial dependence of the emission to
study the eects introduced through the lateral particle distribution. The ba-
sic parameters adopted are the same as in Sec. 2.5.4, but we assume a broken
power-law particle energy distribution from = 51000 for all calculations.
The curvature radius of the shell is adopted as K = 2, 300 m and the Moli ` ere
radius set to r
M
= 117 m as discussed in Sects. 2.4.3 and 2.4.2, correspond-
ingly.
In Fig. 2.9 we plot the spectrum received by an observer in the centre of
the area illuminated by the air shower maximum, considering the same set of
longitudinal particle distributions as before. The spectra look very similar to
those of a line charge, but are attenuated additionally at high frequencies.
A more interesting result is illustrated by Fig. 2.10 which demonstrates the
eect of a purely lateral particle distribution on the radial dependence of the
emission in the conservative scenario, completely ignoring any longitudi-
nal eects. The lateral structure introduces a modulation of the radial depen-
dence, caused by interference of emission from opposite ends of the shower
disk. For higher frequencies and, correspondingly, shorter wavelengths, the
interference minima move to smaller radial distances.
10 20 50 100 200 500 1000
0.01
0.1
1
PSfrag replacements
[MHz]
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.9 |E(R, 2)|-spectrum at the centre of the area illuminated by the maximum of a
10
17
eV air shower with realistic lateral distribution, R
0
= 4 km and a broken power-law energy
distribution from = 51000. Solid: full longitudinal coherence, short-dashed: uniform 5.6 m
longitudinal distribution, long-dashed: Gaussian longitudinal distribution with = 1.61 m,
dash-dotted: longitudinal -distribution with c
t
= 1.61 m. For comparison: fully coherent
case without lateral distribution (dotted)
0 100 200 300 400 500
0.01
0.1
1
PSfrag replacements
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.10Radial dependence of |E(R, 2)| for the maximum of a 10
17
eV air shower
with full longitudinal coherence, realistic lateral structure, conservative approach, R
0
=
4 km and a broken power-law energy distribution from = 51000. Solid: = 50 Mhz,
short-dashed: = 75 Mhz, long-dashed: = 100 Mhz
2.7. Flaring disk 45
0 100 200 300 400 500
0.01
0.1
1
PSfrag replacements
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.11Same as Fig. 2.10 but with reduced and magnetic eld parallel to the
direction of the observer.
In comparison, Figs. 2.11 and 2.12 show the radial dependence in case of
the reduced calculations. The interference eects are somewhat washed
out and the overall emission level is higher. As expected, there is a drastic
asymmetry between the emission pattern in the directions parallel and perpen-
dicular to the magnetic eld. In case of Fig. 2.12, where the observer is posi-
tioned in a direction perpendicular to the magnetic eld, is basically reduced
to zero even for distances of a few hundred metres (
< 8
1
as explained
in Section 2.6.2). Correspondingly, the emission pattern is only very slightly
attenuated even at high distances.
2.7 Flaring disk
We now combine the results derived so far to obtain a more realistic model of
the emission from the maximum of an extensive air shower: a aring disk. In
other words, we adopt the same geometry as specied in Sec. 2.6, but now vary
the thickness of the disk as a function of position (, ) on the shower surface
in the form of the varying asymmetric -pdfs parametrised as in Section 2.4.3.
This geometry therefore correctly takes into account the curvature, the lateral
and the longitudinal structure of the air shower maximum.
Fig. 2.13 again shows the spectrum emitted by the air shower maximum
as a realistically aring disk according to the Agnetta et al. (1997) and Lins-
ley (1986) parametrisations. As expected, the spectrum emitted by the Linsley
46 2.7. Flaring disk
0 100 200 300 400 500
0.01
0.1
1
PSfrag replacements
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.12Same as Fig. 2.10 but with reduced and magnetic eld perpendicular to
the direction of the observer.
aring disk extends to higher frequencies than the one generated by the Ag-
netta aring disk because of the lower thickness in the shower centre where
most of the particles reside (cf. Fig. 2.5).
The radial dependence at dierent frequencies is once again shown in Fig-
ure 2.14. Comparison with the corresponding diagrams for the purely lateral
distribution shown in Figs. 2.102.12 shows that the overall emission level
drops as the observing frequency is increased due to the dampening of higher
frequencies by the longitudinal particle distribution. Additionally, one can
again observe a smearing out of the interference minima. As a consequence,
the conservative and the reduced with observer parallel to the magnetic
eld calculations yield almost identical results.
In Fig. 2.15 we have reconstructed the pulse generated by the aring Ag-
netta disk as it would be measured by a receiver with a given bandwidth using
Eq. (2.10). The pulse amplitude drops noticeably when the observer moves
from the centre of the illuminated area on the ground to a distance of 100 m,
and is already quite diminished at a distance of 250 m, as expected for the
conservative approach. The pulse length of 8 ns is a result of the lter
bandwidth of 120 MHz, i.e. the pulse is bandwidth-limited.
2.7. Flaring disk 47
10 20 50 100 200 500 1000
0.01
0.1
1
PSfrag replacements
[MHz]
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.13 |E(R, 2)|-spectrum at the centre of the area illuminated by the maximum of
a 10
17
eV air shower with aring -pdf, R
0
= 4 km and a broken power-law energy distribution
from = 51000. Solid: aring Agnetta et al. (1997) lateral distribution, short-dashed: aring
Linsley (1986) lateral distribution
0 100 200 300 400 500
0.01
0.1
1
PSfrag replacements
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 2.14Radial dependence of |E(R, 2)| for the maximum of a 10
17
eV air shower
with aring Agnetta et al. (1997) -pdf, R
0
= 4 km and a broken power-law energy distribution
from = 51000. Solid: = 50 Mhz, short-dashed: = 75 Mhz, long-dashed: = 100 Mhz,
upper/lower curves for reduced perpendicular/parallel to magnetic eld direction
48 2.8. Integration over shower evolution
-10 0 10 20 30
-150
-100
-50
0
50
100
150
PSfrag replacements
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 2.15Reconstructed pulses emitted by the maximumof a 10
17
eVshower with aring
Agnetta et al. (1997) -pdf, broken power-law energy distribution from = 51000 and R
0
=
4 km, using an idealised rectangle lter spanning 40160 MHz and conservative scenario.
Solid: centre of illuminated area, short-dashed: 100 m from centre, dash-dotted: 250 m from
centre
2.8 Integration over shower evolution
The last step in modelling the total air shower emission is to integrate over the
shower evolution as a whole. This can be done in a very simplied fashion
by approximating the shower evolution with a number of discrete steps. The
characteristic scale for these steps is given by the radiation length of the
electromagnetic cascades in air, X
0
= 36.7 g cm
2
, corresponding to 450 m
at a height of 4 km. One can therefore discretise the shower evolution into
slices of thickness X
0
, assuming these contain independent generations of
particles and therefore radiate independently.
The emission from each of these slices is calculated as that from a aring
disk, taking into account changes of s, R
0
,
0
, r
M
and N correctly through the
relations given in Section 2.4 and reverting to the conservative denition
to be able to correctly calculate the emission at great angles. Superposition of
the individual slice emissions, correctly taking into account the phases arising
from arrival time dierences, then leads to the total emission of the shower.
Slices far away from the observer are attenuated both due to the high dis-
tance and the decreasing number of particles N. The concrete number of far-
away slices taken into account is therefore uncritical. The situation is dierent
for the slices close to the observer. In their case, the attenuation through the
2.8. Integration over shower evolution 49
10 100 1000
0.01
0.1
1
10
0.1
1
10
100
PSfrag replacements
[MHz]
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
V
m
1
M
H
z
1
]
F 2.16 |E(R, 2)|-spectrum of a full 10
17
eV air shower with aring Agnetta et al.
(1997) -pdf, conservative approach, R
0
= 4 kmand a broken power-lawenergy distribution
from = 51000. Solid: centre of illuminated area, short-dashed: 100 m from centre, long-
dashed: 250 m from centre, black points: rescaled Spencer (1969) data as presented by Allan
(1971), grey points: rescaled Prah (1971) data
decreasing number of particles N is more than compensated by the decreasing
distance to the observer. In fact, the slices closest to the observer yield the
highest contributions of radiation, and the total result depends considerably on
the number of nearby slices taken into account. However, at the same time,
the illuminated area on the ground, governed by the intrinsic beaming cone,
becomes very small for the slices very close to the observer, especially for the
high frequencies where the radiation mainly originates from high-energy par-
ticles with even smaller beaming cones. Except for low frequency emission in
the centre region of the illuminated area, the result for the total emission can
therefore be considered robust.
For our vertical 10
17
eV air shower at a height of R
0
= 4 km we add the
emission from eight slices above and eight slices below the shower maximum
to the emission from the maximum itself. The closest slice then lies at R
0
=
950 m from the observer, a distance we do not want to fall below because of
approximations contained in our calculations that are only valid in the far-eld.
The main eect of the integration over the shower evolution is a boosting
of the total emission because of the increased total number of particles taken
into account, as can be seen in the spectra shown in Fig. 2.16. For frequencies
of 40 MHz and radial distances of 100 m, the amplication factor corre-
50 2.9. Discussion
0 100 200 300 400 500 600
0.01
0.1
1
10
0.01
0.1
1
10
PSfrag replacements
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
V
m
1
M
H
z
1
]
F 2.17Radial dependence of |E(R, 2 55 MHz)| for a full 10
17
eV air shower with
aring Agnetta et al. (1997) -pdf, conservative approach, R
0
= 4 km and a broken power-
law energy distribution from = 51000, data from Allan et al. (1970), horizontal lines from
top to bottom: emission strength needed for a 3-detection with an individual LOPES antenna
or an array of 10 or 100 LOPES antennas
sponds to 10. Apart from the overall amplication, the radial dependence
is signicantly steepened because the important nearby slices only contribute
at low radial distances as discussed earlier. This can be seen when comparing
Fig. 2.17 with the earlier results for the conservative case.
2.9 Discussion
The calculations presented here represent only a few illustrative examples of
possible congurations of EAS that could be calculated with our model. These
examples, however, already demonstrate the most important dependences be-
tween shower structure and emission spectrum as well as radial emission pat-
tern.
2.9.1 Theoretical results
As expected, the thickness of the air shower pancake, and correspondingly in
our model the width of the longitudinal particle arrival time distributions, is
the main factor determining the position of the high-frequency cut-o in the
spectrum. Typical longitudinal scales of a few metres lead to frequency cut-
os in the 100 MHz regime, which supports a choice of observing frequency
2.9. Discussion 51
well below 100 MHz. Due to the strong dependence of the spectral cut-o
on the shower thickness, radio emission from EAS could be used very eec-
tively to probe the longitudinal structure of air showers during their evolution,
a quantity that is not well known at the moment.
The radial emission pattern is mainly governed by the inherent emission
pattern of the synchrotron pulses and the superposition of the beamed emis-
sion from dierent parts of the air shower evolution as a whole. Additionally,
the lateral extent of the air shower slightly inuences the size of the illumi-
nated area on the ground through the resulting coherence minima. A profound
change in the radial emission pattern is visible when one adopts the reduced
approach, which predicts signicant radio emission up to higher distances
depending on the relative orientation of observer and magnetic eld. This is an
important prerequisite for the detection of ultra-high energy EAS with an array
of aordable collecting area in combination with particle detector arrays such
as KASCADE Grande or the Pierre Auger Observatory and will be veriable
by LOPES.
The emitted total power in the coherent regime at low frequencies scales as
the number of particles squared, which could therefore be probed directly by
radio measurements of EAS, yielding information about the primary particle
energy.
We have not explicitly presented how variations of other parameters inu-
ence the radio emission, but most of the associated eects are fairly straight-
forward to foresee: The emitted power scales linearly with B-eld strength.
The declination of the B-eld in central Europe eectively decreases the value
of B and introduces an asymmetric pattern to the radial dependence. An in-
crease of the primary particle energy will boost the radiation because higher-
energy showers will have their maximum closer to the observer. At the same
time, the number of particles increases linearly with primary particle energy
and the power emitted at low frequencies increases as number of particles
squared, which more than compensates the shrinking of the illuminated area
on the ground. Inclined air showers will cause an asymmetric emission pattern
and an attenuation of the emitted power because they reach their development
at higher altitudes. A stronger curvature of the shower front will shift the in-
terference minima to smaller radial distances and thus slightly decrease the
eective size of the illuminated area on the ground.
52 2.9. Discussion
2.9.2 Comparison with experimental data
A number of experiments have clearly established the presence of radio emis-
sion from cosmic ray air showers in the past. A dependence of the polarisation
of the emitted radiation on the earths magnetic eld direction was also con-
rmed by a number of experiments (e.g., Allan et al. 1967, 1969), supporting
the case for the geomagnetic emission mechanism. The actual strength of the
emission, however, is still largely unknown at present state. The analysis of
Allan (1971) led to a widely used formula summarising the presumed depen-
dences:
= 20 Vm
1
MHz
1
_
E
p
10
17
eV
_
sin cos exp
_
r
r
0
(, )
_
, (2.48)
where the scale factor r
0
corresponds to (11010) m at = 55 MHz and for
< 35
. Later works (e.g., Sun 1975; Prah 1971 and references therein),
however, yielded values as low as 15 V m
1
MHz
1
. A recent experiment in
conjunction with the CASA/MIA array conducted by Green et al. (2003) was
only able to place upper limits of
= 3134 V m
1
MHz
1
on the emission
strength.
Part of these discrepancies could be explained by uncertainties in the pri-
mary particle energy calibration at the time the experiments were made. A
number of authors involved in the past works suspect the calibration of the
radio measurements to be the major source of uncertainty (Atrashkevich et al.
1978). Additionally, the documentation of the available data is not always
totally precise regarding the included energy ranges of primary particles, the
selection of allowed zenith angles, the radial distance to the shower axis or the
back-projection of the electric eld vector in the plane normal to the shower
axis and earths magnetic eld, which further complicates the issue.
Extremely low values of
of only 1 V m
1
MHz
1
or even lower are,
however, disfavoured by the fact alone that air showers actually have been
measured by experiments with only a few antennas (e.g., two per frequency
and polarisation direction in case of Prah 1971) with receivers of only a few
MHz bandwidth in the early experiments.
In this dicult situation, we choose to revert to the well documented data
of Allan et al. (1970) as the basis of our analysis. A comparison of these
data with our predicted radial dependence of the emission is shown in Figure
2.10. Conclusions 53
2.17. While we clearly overpredict the emission strength in the centre, the
general radial dependence ts relatively well. Regarding the spectral depen-
dence, we make use of the Spencer (1969) data as presented, converted and
complemented in Allan (1971) as well as the Prah (1971) data. These data
sets, again, yield considerably lower values of
, and we manually scale them

up to make them consistent with the Allan et al. (1970) radial data. While the
absolute values presented in Fig. 2.16 therefore are somewhat arbitrary, the
trend in the dependence actually does correspond to the spectral dependence
that we predict near the shower core.
All in all, we overpredict even the most optimistic past data by a factor 2,
which is, however, not too surprising considering the very simplied integra-
tion over the shower evolution as a whole and the problems involved especially
in the centre region. Additionally, the cuto of the spatial integration as stated
in Sec. 2.6.3 redistributes further emission to the centre region.
We feel that having achieved a result which is consistent with past experi-
mental data within a factor of a few using such approximate descriptions of
the shower characteristics and a mainly analytical approach incorporating ma-
jor approximations is a very encouraging outcome. In addition, our result fur-
ther supports the geomagnetic emission mechanism as the dominant source of
radio emission from EAS. A huge improvement of our model will be achieved
once we revert to elaborate computer simulations that use fewer approxima-
tions and more realistic particle distributions. Consequently, this will be the
next step in our modelling eorts.
2.10 Conclusions
We have analysed properties of radio emission from EAS in the scenario of co-
herent geosynchrotron emission. Our step-by-step analysis has helped to dis-
entangle the coherence eects arising from the dierent physical scales present
in the air shower and to get a good feeling for the relative importance of these
eects. While the spectral cuto is directly governed by the longitudinal ex-
tent of the air shower, the radial dependence arises from the intrinsic beaming
of the synchrotron radiation and its superposition over the shower evolution as
a whole.
The emitted radio power is of the expected order of magnitude, which is the
strongest constraint we can make at the moment due to the large uncertainties
associated with the available experimental data. Hence, in light of the data
available to date, coherent geosynchrotron emission is able to explain the bulk
of the radio emission from EAS.
54 2.10. Conclusions
Our calculations show that LOPES should be able to easily detect the radio
emission from a typical 10
17
eV air shower and will be a very useful tool for
the study of EAS properties, especially the longitudinal structure of the particle
distribution in the shower.
Having achieved a solid understanding of the important eects shaping the
radio emission from cosmic ray air showers, our next step will be to conceive
and implement an elaborate Monte Carlo simulation calculating the emission
with increased precision.
Appendix: Geometry
We adopt the instantaneous velocity vectors of the generated particle pairs as
radially pointing away from the centre of the sphere. For a particle at position
(, ) on the shell, its direction is therefore given by
v(, ) =
_
_
cos sin cos + sin cos
sin sin
sin sin cos cos cos
_
_
, (2.49)
whereas the direction of the B-eld is given by
B =
_
_
sin
B
cos
B
sin
B
sin
B
cos
B
_
_
. (2.50)
Furthermore, the line-of-sight vector R from the particle to the observer is
given by
R(, ) = (R
0
+ K)
_
_
sin( +
0
)
0
cos( +
0
)
_
_
Kv(, ) (2.51)
The direction of R is then calculated as
n(, ) =
R(, )
|R(, )|
(2.52)
and the pitch angle and angle to the line-of-sight correspond to
cos (, ) = v(, )
B (2.53)
cos (, ) = v(, ) n(, ). (2.54)
The direction of the dominating emission component then changes as follows
with (, ):
e
(, ) =
B v(, )
| sin (, )|
. (2.55)
These are all of the geometrical relations that are needed to execute the inte-
gration.
3
Monte Carlo simulations
3.1 Introduction
In chapter 2, we presented calculations of radio emission from extensive air
showers within the scheme of coherent geosynchrotron radiation rst pro-
posed by Falcke & Gorham (2003). These calculations were based on an ana-
lytic approach and were specically aimed at gaining a solid understanding of
the coherence eects that shape the radiation emitted by an air shower. Build-
ing on this foundation, we now continue to develop and enhance our model
with elaborate Monte Carlo (MC) simulations. The MC technique allows us
to infer the emission characteristics with much higher precision by taking into
account more realistic and complex shower properties and applying fewer ap-
proximations than in the analytic calculations. At the same time, it provides
an independent means to verify our previous calculations due to the totally
dierent technique employed. The approach we take is similar to the MC sim-
ulations done by Suprun et al. (2003), yet our simulation is developed to a
much higher level of complexity.
The layout of this chapter is as follows: In sections 3.2 and 3.3 we moti-
vate and explain the application of the MC technique and provide details about
its implementation. In section 3.4, we explain the intelligent concepts that
we have explored to make simulations with a high number of particles and
observer bins feasible on standard personal computer hardware. After a short
description of the raw output of our program and the associated data reduction
in section 3.5, we demonstrate the robustness and consistency of the imple-
mented algorithms in detail in section 3.6. Similar to our earlier theoretical
57
58 3.2. The Monte Carlo approach
calculations, we then rst concentrate on the emission from a single slice of
particles in the air shower and compare the results with the analytical results
in section 3.7. Next, we perform the integration over the shower evolution as
a whole and present the results in comparison with our analytical work and
historic data in section 3.8 before concluding with a discussion of the results
and our conclusions in sections 3.9 and 3.10, respectively.
3.2 The Monte Carlo approach
After having established the general dependences of the radio emission on a
number of air shower properties in chapter 2, a continuation of the calculations
with MC techniques oers a number of advantages.
3.2.1 Motivation and objectives
First, MC techniques allow an independent verication of the analytic calcu-
lations by adopting the exact same shower properties (e.g., distributions of
particles in space and energy, choice of magnetic eld and shower geometry),
but doing the calculations in a completely dierent way. In particular, the MC
calculations are carried out by summing up the individual particles pulses in
the time-domain, whereas the analytic calculations were done in the frequency
domain.
Second, it is relatively easy to include highly complex (and thus realistic)
shower characteristics in the MC simulations, thereby increasing the model
precision signicantly over the previous results. Additionally, it is not neces-
sary to make approximations such as adopting a far-eld limit, which reduces
the accuracy of the analytic calculations.
Overall, MC simulations therefore constitute the logical next step in the
development of our model.
3.2.2 General approach
The general idea of a MC simulation of radio emission from cosmic ray air
showers is simple:
1. model the radiation emitted by an individual particle as precisely as pos-
sible,
2. distribute particles randomly in a simulated shower according to the de-
sired shower characteristics (e.g., spatial and energy distributions),
3.3. Implementation details 59
3. superpose the radiation received from the shower particles at the given
observing positions, taking into account retardation eects.
In fact, the microphysics of an individual particles geosynchrotron emission
(step 1) can be described analytically without the need of any approximations
as long as one does not simulate particle interactions explicitly but only takes
them into account via statistically distributed track lengths. The strategy for
the implementation of steps 2 and 3 then is as follows:
generate shower particles according to the desired distributions,
for each particle:
for each observing position (ground-bin):
establish an adequate sampling of the particle trajectory,
calculate and retard the emission contributions emanating
fromthe sampled points on the trajectory, building up the elec-
tric eld time-series that the observer sees,
incorporate the contributions into the ground-bins pre-
existing time-series data.
This strategy is fairly simple. What makes the problem dicult is the huge
computational eort of a simulation with a large number of particles and
ground-bins, as was already discussed by Dova et al. (1999) who considered
a similar approach. A number of intelligent concepts has to be applied to ac-
tually develop a working simulation out of this simple recipe. We discuss
these concepts in depth in section 3.4.
3.3 Implementation details
In this section we describe a number of relevant implementation details of our
MC code.
3.3.1 Technical information
We decided to program the MC simulation in C++. This modern, industry-
standard programming language oers a number of advantages over others:
it is highly portable, even among Windows and UNIX/Linux,
its stringent typing mechanisms force the programmer to write clean
code,
60 3.3. Implementation details
the wide range of routines collected and standardised in the standard
template library (STL) allows an ecient and less error-prone way of
programming,
object-orientation allows to create a modular, exible and easy to main-
tain program,
powerful compilers and debugging tools are freely available,
the performance is adequate.
The program has been developed under Linux using gcc 2.9.5, gcc 3.3.1,
and the Intel C++ compiler version 8.0.5 which produces much faster code.
No third-party libraries were used in order to maximise portability and min-
imise dependency on external factors. The program source-code will be made
available at a later time.
3.3.2 Particle creation and propagation
The particles in the shower are created with random properties distributed ac-
cording to analytic parametrisations. If not explicitly stated otherwise, the
parametrisations are chosen exactly as in chapter 2. While these parametri-
sations are admittedly crude and do not take into account some air shower
properties such as a realistic particle pitch-angle distribution, this approach al-
lows a direct comparison of the MC results with our analytic calculations. A
more realistic modelling of the air shower will be achieved once our code is
interfaced with the air shower simulation code CORSIKA. The particles are
created with the following properties chosen randomly:
the shower age at which a particle is created (longitudinal development
according to Greisen (1960) function), which directly yields
the position along the shower axis
the creation time
the lateral shift from the shower axis (NKG-distribution dating back to
Kamata & Nishimura (1958) and Greisen (1960))
the longitudinal shift along the shower axis as function of the lateral shift
(asymmetrical -PDF)
the azimuth angle for the lateral shift (isotropic)
the particle gamma factor (broken power-law distribution or xed
60)
the particle track length (exponential probability distribution or xed
40 g cm
2
)
To take into account the pair-wise creation of particles, one electron and one
positron are always generated with the same properties. At the moment, no
random spread in the particle pitch-angle is introduced, i.e., the initial particle
momenta radially point away from a spherical surface with 2,300 m radius,
exactly as in the analytical calculations, motivated by the data from Agnetta
et al. (1997). The fact that the initial velocity direction is shared by both the
electron and positron introduces only minor error as the transverse momentum
arising from the pair production is minimal a fact that is intuitively illus-
trated by the still very dense core of the lateral distribution function even after
many generations of particle creation.
As each of the particles is created at a specic position at a given time
with a given initial velocity, the trajectory r(t) resulting from the deection
in the geomagnetic eld is a well-dened helix which can easily be described
analytically by equations (3.27), (3.31) and (3.33) as derived in the appendix.
3.3.3 Calculating and collecting contributions
Once the trajectory (and its time-derivatives) are known analytically, the radi-
ation an observer at position x receives at time t can be calculated (cf. Jackson
1975 equation 14.14) immediately by
E(x, t) = e
_
n
2
(1 n)
3
R
2
_
ret
+
e
c
_
n {(n )

}
(1 n)
3
R
_
ret
, (3.1)
where e denotes the particle charge, (t) = v(t)/c is directly given by the par-
ticle velocity, R(t) refers to the vector between particle and observer position,
R(t) = |R(t)| and n(t) = R(t)/R(t) denotes the line-of-sight direction between
particle and observer.
The index ret points out that the quantities in the brackets have to be
evaluated at the retarded time
t
ret
= t R(t
ret
)/c (3.2)
62 3.3. Implementation details
rather than at the time t in order to accommodate the nite light-travel time.
This recursive retardation relation imposes signicant problems for an an-
alytical calculation in the time-domain. In case of a MC simulation, on the
other hand, it is absolutely straight forward to take the retardation into account
by simply delaying the emitted signal appropriately before collecting it in the
ground-bins.
The rst term in equation (3.1) constitutes the static term that falls o
with R
2
in eld strength. It is usually neglected, and was not taken into ac-
count in the analytical calculations of chapter 2 either. While its contributions
are indeed negligible, we still include it in our MC simulation as to not make
any unnecessary approximations.
The second term is the usual radiation term which drops as R
1
in eld
strength and therefore dominates very quickly over the static term as one goes
to higher distances. In analytic approaches, it is usually necessary to apply
approximations such as the Fraunhofer-approximation for the far-eld limit to
this term, which naturally limits the precision of the results. Again, for a MC
simulation, it is not necessary to make any approximation for the radiation
formula.
Beaming eects are naturally taken into account in this formula through
the (1 n)
3
terms in the denominator that lead to very high eld strengths
for particle velocities close to c and small angles to the line-of-sight. As soon
as one takes into account the refractive index of a medium rather than vacuum,
the denominator actually becomes zero at the

Cerenkov angle. The arising
singularity, in combination with the modied retardation relation, then leads
to

Cerenkov radiation. The analysis of these eects, however, is beyond the
scope of this work and will be carried out in a later paper.
For a given particle, the trajectory is then sampled in a suciently high
number of points, and the retarded contribution emanating from each of these
points is calculated for each of the observing ground-bins with the full preci-
sion of equation (3.1). Thus, the retarded electric eld time-series produced
by the particle is inferred for each of the ground-bins and can then be incorpo-
rated into their pre-existing time-series data. The details and subtleties of this
procedure are explained in sections 3.4.2 and 3.4.3.
3.3.4 Atmosphere model
We use the US standard atmosphere of 1977 as implemented in CORSIKA
(Ulrich 1997). It splits the atmosphere in a number of layers, in each of which
the path depth in g cm
2
(as counted vertically from the outer edge of the
Layer Height [km] a
i
[g cm
2
] b
i
[g cm
2
] c
i
[cm]
1 0 4 -186.56 1222.66 994186.38
2 4 10 -94.92 1144.91 878153.55
3 10 40 0.61 1305.59 636143.04
4 40 100 0.00 540.18 772170.16
T 3.1Parameters for the parametrisation of the atmospheric layers.
atmosphere) is parametrised by an exponential dependence on height
X(h) = a
i
+ b
i
exp
_
h
c
i
_
. (3.3)
The layer boundaries and corresponding parameters a
i
, b
i
and c
i
are given
in table 3.1. (The outermost layer in the US standard atmosphere in which
X(h) scales linearly with height has been omitted as we are not interested in
processes above 100 km height.)
For the sake of completeness, the height as a function of atmospheric depth
is then given by
h(X) = c
i
log
X a
i
b
i
, (3.4)
and the density at a given atmospheric depth is inferred as
(X) =
(X a
i
)
c
i
. (3.5)
Finally, the Moli` ere radius r
M
is parametrised as (Dova et al. 2003)
r
M
(X) =
9.6
(X a
i
)
c
i
. (3.6)
3.3.5 Random number generation
An important centre-piece of any MC simulation is the random number gen-
erator at its heart. We revert to the well-known Mersenne-Twister random-
number generator of Matsumoto & Nishimura (1998) in the C++ implementa-
tion by Fog (2003). Only one instance of the random number generator is used
throughout the whole program to prevent potential problems with interfering
multiple instances of random number generators (see section 3.6.2 for details).
If possible, the generators for non-uniform probability distributions were
64 3.4. Intelligent concepts
implemented using analytic inversions. If not, we reverted to rejection meth-
ods.
3.4 Intelligent concepts
As mentioned earlier, a brute-force approach as sketched in the recipe
given in section 3.2.2 involves such a high computational eort (both regard-
ing CPU time and memory) that it simply is not feasible on standard PCs.
One therefore has to employ a number of intelligent concepts to minimise the
computing eort. The main ideas that we have explored to reach this goal are
described in the following subsections.
3.4.1 Cutting o
1
-cones
The radiation pattern emitted by an individual highly relativistic particle is
heavily beamed in the forward direction (see, e.g., Jackson 1975). Most of
the emission is radiated into a cone with opening angle of order
1
. This
directly leads to a very simple, but eective idea of how to cut down on com-
puting time:
For an individual particle ying on its given trajectory, the
1
emission
cone sweeps over a relatively small region on the ground. It is only in this
ground-region of a few times
1
width, which we call the ground-trace
of the particle trajectory, that receives considerable contributions of radiation
from that particle. Thus, we can select only the ground-bins inside the ground-
trace of this specic particle for evaluation as sketched in gure 3.1, hugely
cutting down on computing time.
It turns out, however, that the discrete cutting o after a certain angular
distance introduces errors in the calculation, which are although only at a
percent-level signicant when one is interested in the emission strength at
distances a few hundred metres from the shower core. We will discuss the
details in section 3.6.5.
3.4.2 Smart trajectory-sampling
In general, there are two approaches of how to calculate the time-dependence
of the emission that a specic ground-bin receives from a given particle:
The rst approach is to take the trajectory of a given particle, sample
it in a sucient number of points and calculate the corresponding contribu-
tions for the given ground-bin using (3.1). If the particle trajectory is sam-
pled in equidistant time-steps, the retardation eects involved lead to a non-
3.4. Intelligent concepts 65
F 3.1Cutting o
1
-cones: only bins in the ground-trace of a particles trajectory
are selected for evaluation.
equidistantly sampled time-series for the ground-bin. The heavily-peaked
pulse shape is, however, automatically sampled with high precision in this
approach.
As the time-series data collected by the ground-bin have to be gridded
eventually, it would be easier if one could take another approach: taking the
ground-bins pre-dened equidistant time-grid and sampling the particle tra-
jectory in the points corresponding to this grid. This, however, is not easily
possible because of the recursive retardation relation (3.2). One would have
to do an iterative search for the corresponding points on the particle trajec-
tory, which would be at least as time-consuming as interpolating or binning
a non-gridded time-series derived with the rst approach to the ground-bins
time-grid. At the same time, this approach bears the risk of missing highly
peaked contributions in case of a too widely spaced time-grid.
The rst approach, therefore, is the better choice for our calculations. It
is, however, possible to improve on the case of equidistant sampling of the
particle trajectory by taking advantage of the strong beaming of the emission
once again. The peaks in the time-series result when the denominator (1 n)
3
in (3.1) gets small as the angle between and n gets small. It therefore makes
sense to densely sample the region of small angle to the line-of-sight, and
increase the distance between the sampled points (up to a maximum value)
as the angle gets larger. This minimises the number of points used while it
guarantees high-precision sampling of the pulse shape as can be seen in gure
3.2.
Our algorithm for the smart trajectory sampling is implemented as follows:
The trajectory of length l (in g cm
2
) is mapped onto the interval d [0, 1]
with user-dened equidistant step-size d, defaulting to d = l
1
g cm
2
. The
minimum and maximum step-sizes are set to d
min
= (1/3)d and d
max
=
(20/3)d, respectively. When (d) denotes the angle between line-of-sight
vector n(d) and instantaneous particle velocity vector v(d), and the particle
gamma factor is , the dynamic step-size d(d) is then set to
d(d) = (d)
2
2
d
min
, (3.7)
but no less than d
min
and no more than d
max
.
3.4.3 Intelligent gridding strategy
The individual particles pulses are very short, of order 10
11
s for = 60
particles. Usually, however, one is only interested in the result on scales of
0
100000
200000
300000
400000
500000
-1 0 1
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 3.2Smart trajectory sampling: the particle trajectory is sampled densely in the peak-
region and only sparsely further outside.
several nanoseconds as dened by the interesting frequency range of tens to
hundreds of MHz. On the other hand, it would be useful for diagnostic and
verication purposes to have a means to record the events in the very high
time-resolutions of the individual pulses. We have therefore implemented both
possibilities into the simulation program. The dierent demands are met by
the use of two very dierent gridding strategies.
The more ecient and thus favoured strategy for low time resolutions is
the use of a simple grid. It is based on an equidistant grid of user-dened
resolution (typically 1 ns). The contributions from a particles time-series
data are then binned and thus time-averaged onto the grid. The gridding is
dynamic in the sense that data points are inserted automatically when needed.
This saves memory in comparison with an ordinary equidistant grid. Still, this
strategy does not scale well to high time-resolutions.
To counter the high memory demands for high time resolutions, we have
implemented an economic grid which describes the pulse-shapes using only
a minimum number of points. It is based on an underlying equidistant time-
grid which limits the time-resolution to a pre-dened maximum value. The
time-series of an individual particle is interpolated to the underlying grid po-
sitions and then incorporated into the pre-existing time-series, correctly inter-
polating any contributions that were already present. Points are, however, only
F 3.3The economic gridding mechanism: when new contributions are registered onto
existing contributions, points are inserted and interpolated only as needed.
inserted in the grid where necessary. See Fig. 3.3 for an illustration of the
algorithm.
As another major advantage, the availability of these two very dierent
gridding strategies allows an independent cross-check of their implementation.
3.4.4 Sequentialised and parallelised calculation
The program is designed such that calculations can easily be sequentialised or
parallelised to cut down on memory requirements or take advantage of multiple
computers, respectively.
Memory usage increases with the number of ground-bins to be calculated.
To facilitate the use of low-memory machines, the calculation of ground-bins
can be sequentialised: the emission from the complete shower is calculated for
only a subgroup of ground-bins, and only after the calculation has nished and
the results have been written to disk, the next group of ground-bins is evalu-
ated. This eciently decreases the memory-usage as compared to concurrent
evaluation of all ground-bins. The overhead introduced due to the necessary
multiple creation of particle distributions is negligible for most combinations
of parameters. Manually specifying the random seed value for the random
number generator guarantees identical particle distributions in the dierent cal-
culation segments. Allowing for dierent seed values in the calculations, on
the other hand, provides a consistency check based on the underlying symme-
tries in the emission pattern as the dierent ground-bins are calculated based
on independent sets of random numbers.
Similarly, dierent subgroups of the desired ground-area can be calculated
in parallel on dierent computers, yielding up to a linear decrease of net com-
putation time.
3.4.5 Automatic ground-bin inactivation
As the simple discrete cutting o of regions outside the
1
ground-trace
region described in section 3.4.1 introduces errors that are too big if one is
interested in regions of a few hundred metres distance to the shower core, we
developed a more sophisticated means of cutting down on computation time.
Since most of the particles are distributed in the innermost centre-region of
the shower, the ground-bins close to the centre-region receive a high number
of strong contributions, whereas the far-away regions only receive a smaller
number of not-so-strong contributions. While the centre-regions might already
have reached sucient precision after a certain number of particles, the outer
regions might still not have reached the desired statistical precision, aording
a calculation with an even higher number of particles.
The computing time can be distributed in a much more ecient way by an
on-the-y inactivation of ground-bins that have reached the desired precision.
To accomplish this, the program evaluates the shower emission in steps of
(typically) 10,000 particles at a time. After each of these steps it compares the
(smoothed) time-series derived for a specic ground-bin up to that stage with
the results for that ground-bin at the previous step. Once the relative changes
fall under a pre-dened limit for a user-dened number of steps in a row, the
corresponding ground-bin is marked as inactive and is not evaluated any
further. The computing time is thus eectively redistributed to the outer bins.
The fact that the inactivation sequence should propagate from the inner to
the outer regions is used as a consistency check for the procedure. Another
advantage is that the user can directly specify a desired precision rather than
70 3.5. Data output and reduction
-20
0
20
40
60
80
100
120
140
160
180
200
0 20 40 60 80 100 120
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 3.4Time-dependence of the raw pulses originating from the shower maximum as
observed by an observer at (from left to right) 20 m, 140 m and 460 m to the north from the
shower centre.
having to estimate the number of particles needed to reach adequate results.
3.5 Data output and reduction
To facilitate the understanding of the following discussions, we give a short
overview of the raw data that the simulations produce and the kind of data
reduction that we apply to visualise the results.
3.5.1 Raw data
The Monte Carlo code tracks the individual particles and calculates the as-
sociated (vectorial) electromagnetic pulses a specic observer (i.e., a specic
ground-bin) receives. The individual pulses are extremely short, of order a
few 10
11
s (cf. Fig. 3.2). Superposition of all the individual pulses yields
the raw output of the program: one data le per ground-bin stating the time-
dependence of the north-south, east-west and vertical polarisation components
of the electric eld. Fig. 3.4 shows the total eld strength of the raw pulses at
dierent distances to the north from the centre of a shower slice consisting of
10
8
particles in the shower maximum. Due to the particles spread in position
and time, the pulses are considerably broader than the individual pulses, of
order tens of nanoseconds. A helper application is then used to process these
3.5. Data output and reduction 71
0.001
0.01
0.1
1
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.5Spectra of pulses originating from the shower maximum for observers at (from
top to bottom) 20 m, 140 m, 340 m and 740 m distance to the north of the shower centre.
raw data and reduce them to the desired physical quantities.
3.5.2 Spectral ltering
In a rst step, the time-series data of the electric eld vector is Fourier-
transformed, yielding the associated spectrum depicted in Fig. 3.5. Due to
coherence losses caused by interference eects, the spectra fall o steeply to-
wards high frequencies. At a certain frequency, dependent on the distance
from the shower centre, the eld strength reaches a rst interference minimum
followed by a rapid series of alternating maxima and minima in the incoherent
regime. Insucient sampling of these extrema yields the unphysically seem-
ing features seen here at high frequencies. In the emission from a real air
shower such features are unlikely to exist as the inhomogeneities present in an
air shower, but not taken into account in the analytic parametrisations, destroy
the pronounced extrema. Calculation of the emission in this region therefore
requires a more detailed air shower model, e.g. by interfacing of our code to
CORSIKA.
Any concrete experiment will have a nite frequency bandwidth. Thus, we
lter the spectra to infer the actual pulses that the experiment will register. As
is obvious from the spectra, most of the power resides at the low frequencies.
For this reason the amplitude drops signicantly when ltering frequencies
72 3.5. Data output and reduction
-100
-50
0
50
100
150
200
-20 0 20 40 60 80 100 120
time [ns]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
E
E
W
(
t
)
[
V
m
1
]
F 3.6Comparison of the east-west component of a raw pulse (solid), a pulse smoothed
with a 40160 MHz idealised rectangle lter (long dashed) and a 42.577.5 MHz lter as used
in LOPES (short dashed) for emission from the shower maximum.
below, e.g., 40 MHz as seen in Fig. 3.6.
Here we have used the same idealised 40160 MHz lter that we applied
in the theoretical calculations of chapter 2 as well as the actual 42.577.5 MHz
lter that is used in LOPES. The acausality of the pulse ltered with the ide-
alised rectangle lter illustrates that such a lter is an unphysical concept. A
physical lter does not show acausal behaviour despite its steep edges as it
delays the signal appropriately. This is well visible for the LOPES lter.
3.5.3 Further data processing
To analyse the radial dependence of the emission strength, we then determine
the maximum amplitude of the ltered pulse in each ground-bin. The depen-
dence of this ltered pulse maximum amplitude can then be visualised in a
number of ways, e.g., as surface plots, contour plots or cuts in specic direc-
tions (see the following sections for examples). The absolute time associated
to the maximum pulse amplitude additionally yields information about the cur-
vature of the radio wave front.
One subtlety involved with this procedure is the noise levels present at high
radial distances. As can be seen in Fig. 3.4, the pulses get signicantly broader
as one goes to higher distances. (This eect gets even stronger for fully in-
3.6. Consistency checks 73
tegrated showers.) At distances of several hundred metres, the pulses are so
broad that a lter clipping frequencies below 40 MHz actually resolves the
pulses out. Consequently, the pulse amplitude drops to very low values compa-
rable to those introduced by the higher-frequency numerical noise associated
with the very short individual particle pulses. Taking the maximum amplitude
as a measure for the emission strength then might no longer constitute a useful
procedure. Calculating the time-integral of the eld strength or the received
power would be a better approach in these cases. We are, however, mostly
interested in the emission strengths up to radial distances of 500 m, where
the ltered pulse amplitude gives adequate results which are directly related to
the experimentally relevant signal-to-noise levels.
For the diagnostics of the employed algorithms in section 3.6 and the anal-
ysis of track length eects in section 3.7.1 we therefore adopt 800 m as a cuto
distance. For the calculation of emission strengths from a shower slice and an
integrated shower, we limit the plots to distances of 550 m and 400 m,
respectively, to insure that the ltered pulse amplitudes give adequate results
which are directly related to the experimentally relevant signal-to-noise levels.
3.6 Consistency checks
We have studied the output of our MC code very carefully to make sure that the
calculations are correct. In particular, we have made the following consistency
checks:
3.6.1 Individual particle pulses
Figure 3.7 shows the comparison between an analytical calculation and our
MC code for a pulse created by a point-source consisting of 10
8
particles with
60 at 4 km height, comparable to a pulse that would originate from the
maximum of a vertical 10
17
eV air shower concentrated into a point, as seen
from an observer in the shower centre. The magnetic eld is adopted as hori-
zontal with a strength of 0.3 Gauss. The good agreement between the MC and
analytic results demonstrates that the calculation of the particle trajectories and
emission contributions is implemented correctly.
It should be noted that the pulse length usually stated in textbooks such as
Rybicki & Lightman (1979) (equation 6.10a)
t
A
=
2
B
sin
_
1
v
c
_
(3.8)
74 3.6. Consistency checks
0
100000
200000
300000
400000
500000
-0.04 -0.02 0 0.02 0.04
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 3.7Total eld strength of a pulse originating from a point source of 10
8
= 60
particles in 4 km height, observed in the shower centre. Solid: analytic calculation, points:
calculated with MC code.
is in itself already an approximation which yields values that are signicantly
too low. This results from an implicit approximation sin() in the deriva-
tion of the formula. The correct result is
t
A
=
2
B
sin
_
1
v
c
sin
1
_
. (3.9)
Figures 3.8 and 3.9 demonstrate howthe pulses fromthe same point-source
change when the observer moves outwards from the shower centre to the north
and east, respectively. While the pulse amplitude drops quickly when one goes
to the north, it stays fairly constant as one goes to the east. This eect was
already discussed under the term reduced in chapter 2: The particle trajec-
tory bends towards an observer in the east or west, so that he or she still sees
the particle with a very small angle to the line of sight just during a dier-
ent part of the trajectory. For the specic geometry chosen in this example,
i.e. a particle pair starting o vertically downwards in the shower centre, the
pulses even get broader as one goes outwards to the east, since one only sees
an (asymmetric) half pulse in the centre and can see the full (symmetric) pulse
only at considerable distance. All in all the behaviour is exactly as expected.
Another important consistency check is the dependence of the individual
0
100000
200000
300000
400000
500000
0 1 2 3 4 5
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
E
E
W
(
t
)
[
V
m
1
]
F 3.8East-west polarisation component of a point-source with = 60 particles at
4 km height at increasing distance to the north from the shower axis. From left to right: 5 m,
305 m, 505 m, 705 m, 855 m and 995 m.
0
100000
200000
300000
400000
500000
0 1 2 3 4 5
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
E
E
W
(
t
)
[
V
m
1
]
F 3.9Same as gure 3.8 at increasing distance to the east from the shower axis.
0
100000
200000
300000
400000
500000
600000
700000
800000
900000
0 0.02 0.04 0.06 0.08 0.1
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 3.10Individual particle pulses from a point-source shower as measured by an ob-
server situated 995 m east of the shower centre. Solid: B = 0.3 Gauss, dashed: B = 0.5 Gauss.
The time-integral over the pulses is constant.
pulses on the magnetic eld strength. Figure 3.10 shows the changes arising
when the magnetic eld is enhanced from 0.3 Gauss to 0.5 Gauss (both with
0
inclination, i.e. horizontal). The pulse gets stronger, yet at the same time
shorter so that the integral over E dt stays constant (while the power an ob-
server receives from an individual particle, i.e. the integral over E
2
dt, thus
scales linearly with B), exactly as inferred from analytic calculations. This
directly leads to an important result: The overall radio emission from an air
shower cannot depend strongly on the magnetic eld strength, as it is the sum
of a great number of individual pulses, the integral of each of which is inde-
pendent of the value of B. Likewise, asymmetries between north and south that
are introduced by a realistically inclined magnetic eld cannot be very strong.
The fact that the 0.5 Gauss pulse arrives earlier than the 0.3 Gauss pulse
for an observer 995 m to the east of the centre is explained by the stronger
curvature of the particle trajectory.
3.6.2 Symmetry N-S and E-W
For a vertical air shower, the emitted radiation pattern must have a number of
inherent symmetries. In particular, the east and west directions are completely
equivalent (as long as particle track lengths for electrons and positrons are
adopted as identical), so that the emission pattern must be symmetric in east
and west. For a horizontal magnetic eld, north and south must also be equal.
Exploiting these fundamental symmetries, we discovered a subtle bug in
an early version of the code. It turned out that the pulses in the east were
actually dierent from the ones in the west, as well as north was dierent from
south. This led us to the conclusion that the random number generation itself
was awed in the sense that it introduced an articial correlation between the
longitudinal and the azimuthal distribution. In fact, the seed values for the
multiple instances of random number generators that were used at this stage
in the code were chosen by an unsuitable routine that set all seeds to the same
value (time in seconds since a given date). After having changed the program
such that only one instance of the RNG is used for the entire simulation, the
intrinsic symmetries were fullled as expected.
Needless to say that one can also use the intrinsic symmetries to save com-
putation time by calculating only a half- or even quarter-plane on the ground
and then mirroring the results accordingly.
3.6.3 Gridding algorithms
As explained in section 3.4.3, we have implemented a simple grid for low
time-resolution calculations as well as an economic grid for cases in which
the user is interested in the full time-resolution associated to the individual
particle pulses.
The independence of the two algorithms allows a cross-check of the rou-
tines. Fig. 3.11 shows the raw pulse as calculated with the dierent gridding
strategies at dierent time resolutions. The result is consistent between all
three cases, which demonstrates that both algorithms work well. The simple
grid proves to be very ecient at low time-resolutions (typically 1 ns). It
is especially robust in the sense that it remains stable regardless of the spe-
cic resolution used. The economic grid on the other hand has to be set to
a time-resolution high enough to resolve the individual particle pulses. Es-
pecially when particle energy distributions are switched on, introducing very
high-energy particles with up to = 1000, this strategy quickly becomes inef-
cient.
Additionally, low time-resolution simple grid data yields smoother
pulses due to the time-averaging over the individual particle pulses, which
allows precise calculations with fewer particles.
0
50
100
150
200
0 5 10 15 20 25 30 35 40
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
t [ns]
|
E
(
t
)
|
[
V
m
1
]
F 3.11Raw pulse from a shower slice calculated with dierent gridding strategies
and resolutions. Solid with points: simple grid 10
9
s, solid black: simple grid 10
10
s, light
coloured: economic grid 10
12
s.
3.6.4 Smart trajectory-sampling
Figure 3.12 demonstrates that the errors introduced by the smart sampling
algorithm are only slight. At the same time, the algorithm allows a huge
cut-down on computation time. The algorithm can, however, optionally be
switched o for a more precise calculation.
3.6.5 Cutting o
1
-cones
As mentioned earlier, the discrete cutting o of radiation contributions out-
side the ground-trace of a few
1
-cones width can decrease the computation
time enormously. Figure 3.13, however, demonstrates that this strategy is not
suitable if one needs precision at the percent-level to be able to describe the
emission pattern out to distances of several hundred metres from the shower
centre.
The discrete cutting introduces breaks in the radial emission pattern ex-
actly at the positions corresponding to the cuto, i.e. at 530 m in case of
8
1
-cones and 270 m in case of 4
1
-cones for = 60 particles at 4 km
height. The eect is less strong in the east-west direction, but overall this
strategy is disqualied for high-precision calculations.
The problems are resolved by the more sophisticated on-the-y inactiva-
0.1
1
10
100
0 100 200 300 400 500 600 700 800
distance from centre [m]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.12Changes introduced by the smart sampling algorithm in the radial emission
pattern of the rectangle- ltered maximum pulse amplitude for emission from the shower max-
imum. Thin lines: dense equidistant sampling, thick lines: smart sampling; solid: to the north,
dashed: to the west.
0.01
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.13Changes introduced by the cutting o of regions outside a few
1
-cones in the
radial emission pattern of the frequency- ltered maximum pulse amplitude for emission from
the shower maximum. Distance from the shower centre is to the north. Solid: no cutting, long
dashed: cutting after 8
1
, short dashed: cutting after 4
1
80 3.7. Emission from a shower slice
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.14Changes introduced to the radial emission pattern of the frequency- ltered
maximum pulse amplitude for emission from the shower maximum by the automatic bin inac-
tivation algorithm. Thin lines: no automatic bin inactivation, thick lines: automatic bin inacti-
vation; solid: to the north, dashed: to the west.
tion of ground-bins.
3.6.6 Automatic ground-bin inactivation
Fig. 3.14 demonstrates the stability of the automatic ground-bin inactivation
algorithm. The calculations are as precise as those with a xed high number of
particles for all ground-bins, yet at the same time allow a much more ecient
use of computing time and the denition of a user-specied precision goal. The
deactivation sequence propagates from the inside to the outside as expected
(see Fig. 3.15). It is also no surprise that the bins in the far east and west require
the most computing time, as these are the regions that are inuenced most by
edge eects from trajectory cutos, dierent trajectory curvatures and the like.
The automatic ground-bin inactivation strategy therefore turns out to be a very
powerful and self-consistent technique to conduct extensive simulations with
high precision.
3.7 Emission from a shower slice
Similar to the analytical calculations described in chapter 2, we rst take a look
at the emission from a single slice of the air shower. Throughout this section,
3.7. Emission from a shower slice 81
N S
W
E
F 3.15Automatic ground-bin inactivation sequence. Darker bins are set inactive later
than lighter bins. The sequence propagates from the inside to the outside. The pattern is east-
west and north-south symmetric as expected for a vertical shower and a horizontal magnetic
eld.
we consider only the maximum of a vertical air shower induced by a 10
17
eV
primary particle, consisting of 10
8
charged particles at a height of 4 km. In a
step by step analysis, we increase the complexity of the particle distributions
and evaluate the changes introduced in the simulation results.
3.7.1 Trajectory length eects
In a rst step, we adopt the geomagnetic eld parallel to the ground with a
strength of 0.3 Gauss, as it is present at the equator. This is the same con-
guration that we used in the theoretical calculations. For the moment, we
consider the simplied case of monoenergetic 60 particles.
The theoretical calculations carried out in chapter 2 were based on an ana-
lytical derivation of the spectra of individual particles on circular trajectories.
This derivation makes the implicit assumption that the trajectory is always
symmetric with respect to the point in which the minimum angle to the ob-
servers line of sight is reached. In other words, edge eects arising from the
cutting o of the nite trajectories are not taken into account.
These edge eects, however, turn out to signicantly shape the radial emis-
sion pattern of the radiation. Let us rst consider the case of trajectories which
are long enough to not produce signicant edge eects for observers far away
from the shower centre by adopting a trajectory length of 100 g cm
2
. The
result is depicted in Fig. 3.16. As expected, it is very similar to the theoretical
prediction for the reduced case presented in Fig. 2.14. In this scenario, one
would expect a signicant asymmetry between the north-south and east-west
direction, which would obviously be a very useful observable.
What happens, however, if one adopts more realistic particle track lengths?
Let us rst consider a constant trajectory length of 40 g cm
2
, which is approx-
imately the free path length (equal to one radiation length) of electrons and
positrons in air (Allan 1971). In this scenario, edge eects strongly shape the
emission at high distances in the east-west direction as shown in Fig. 3.17. An
unusual kink appears at 540 m. This is not a numerical glitch, but an inter-
ference eect arising when the observer stops to see the main (positive) peak
of the individual particle pulses due to the edge eects. He then only receives
the initial (negative) contribution of the electric eld pulse. This eectively
causes a polarity change in the raw pulses as shown in Fig. 3.18, accompanied
by a temporary drop in the ltered pulse amplitude.
Finally, we change to a realistic exponential distribution of track lengths
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.16Radial dependence of the maximum rectangle- ltered pulse amplitude for
emission from the shower maximum in the north (solid) and west (dashed) direction in case
of constant and long particle trajectories ( 100 g cm
2
, no edge eects at high distances
from the shower centre).
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.17Same as Fig. 3.16 for 40 g cm
2
. See text for explanation of the kinkat
540 m.
-100
-50
0
50
100
50 100 150 200
time [ns]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
E
E
W
(
t
)
[
V
m
1
]
F 3.18Time-dependence of the raw pulses originating from the shower maximum as
observed by an observer at (from left to right) 460 m, 500 m, 540 m, 580 m and 620 m to the
west from the shower centre. See text for explanation of the polarity change.
with a mean of = 40 g cm
2
,
p(X) = p
0
exp
_
_
. (3.10)
As can be seen in Fig. 3.19, the asymmetry between north-south and east-west
direction is now washed out up to high distances. (In fact, it will be washed
out almost completely once the integration over the shower evolution is taken
into account.)
Apart from the regions far from the shower centre, edge eects also occur
in the centre region due to the instantaneous starting of the trajectories. These
eect are discussed in section 3.7.4.
The signicance of the trajectory length eects already illustrates the im-
portance of adopting as realistic properties for the particle distributions as pos-
sible, a goal that could not be reached with analytic calculations alone. We
retain the realistic statistical distribution of track lengths for the following dis-
cussions of the magnetic eld and energy distribution eects.
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.19North (solid) vs. west (dashed) asymmetry in the maximum ltered pulse
amplitude for emission from the shower maximum in case of statistically distributed track-
lengths. The asymmetry is washed out up to high distances.
3.7.2 Magnetic eld dependence
We now make the change from an equatorial 0.3 Gauss horizontal magnetic
eld to the 0.5 Gauss 70
inclined magnetic eld present in central Europe.

As explained in Sec. 3.6.1, the inuence of the magnetic eld on the inte-
grated shower pulse should not be very signicant. This is conrmed by Fig.
3.20. Although the projected magnetic eld strength drops from 0.3 Gauss
to 0.17 Gauss, the emission pattern only changes very slightly. In particular,
the amplitude level stays almost constant. Interestingly, the north-south vs.
east-west asymmetry is washed out even further.
On the other hand, the inclination of the magnetic eld breaks the intrinsic
north-south symmetry of the emission pattern. This is visible in the north-
south and east-west polarisation components of the contour plots shown in
Fig. 3.21. Similarly, the symmetry of the ground-bin inactivation sequence is
broken as shown in Fig. 3.22. The eect is, however, only weak and there is
still no asymmetry if one considers the total electric eld strength rather than
a specic polarisation direction.
Note that there is no signicant emission in the centre region of the north-
south polarisation component (second column of Fig. 3.21). This demonstrates
that the numerical cancellation of north-south polarised radiation components
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.20Changes to the north (solid) and west (dashed) radial emission patterns for
emission from the shower maximum when going from a 0.3 Gauss horizontal magnetic eld
(thin lines) to a 70
inclined 0.5 Gauss magnetic eld (thick lines).

-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-600
-400
-200
0
200
400
600
-600 -400 -200 0 200 400 600
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
F 3.21Contour plots of the 40160 MHz rectangle- ltered maximum pulse amplitude
for emission from the shower maximum in case of a horizontal 0.3 Gauss magnetic eld (upper
panel) and a 70
inclined 0.5 Gauss magnetic eld (lower panel) and 60 particles. Contour
levels are 5 Vm
1
apart. Fromleft to right: total electric eld strength, north-south polarisation
component, east-west polarisation component, vertical polarisation component.
N S
W
E
F 3.22Automatic ground-bin inactivation sequence in case of 70

inclined 0.5 Gauss
magnetic eld. The north-south symmetry is broken as expected, cf. Fig. 3.15.
from the electrons and positrons (labelled A
in chapter 2) indeed works cor-

rectly in the MC code and that it was justied to neglect this component in
the earlier theoretical calculations at least for the primarily important centre
region.
In summary, the magnetic eld eects are weak, but non-trivial.
3.7.3 Energy distribution eects
In the theoretical calculations, a change from monoenergetic 60 electrons
to a broken power-law peaking at = 60 introduced only minor changes,
namely a slight redistribution of radiation from medium distances to the centre
region (due to the high-energy particles) and high distances (due to the low-
energy particles). The changes introduced by the energy distribution are bound
to be more complex in the MC simulations, as they are heavily intertwined
with the edge eects discussed before. Nonetheless, a similar redistribution of
radiation from the medium scales to the innermost centre region is observable
in the MC simulations as shown in Fig. 3.23.
The drop visible in the west direction at 200 m arises due to a tran-
0.1
1
10
100
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.23Changes introduced when switching from monoenergetic 60 particles
(thin lines) to a broken power-law peaking at = 60 (thick lines) for emission from the shower
maximum. Solid: to the north, dashed: to the west. See text for explanation of the drop at
200 m.
sition from a dominating east-west polarisation component to a dominating
north-south polarisation component in combination with some resolving out of
pulses due to the lter bandwidth used. (Fig. 3.21 demonstrates that the north-
south and east-west polarisation components become comparable at these dis-
tances.)
It should be pointed out, however, that the strength of the eects intro-
duced by the choice of a specic energy distribution seem misleadingly strong
when one considers the emission from a single shower slice alone. Once the
integration over the shower evolution as a whole is performed, most eects
(including the drop at 200 m) are again washed out almost completely and
the inuence of the specic choice of energy distribution becomes very weak.
For the moment, we therefore continue to use the broken power-law distribu-
tion that we adopted in the theoretical calculations rather than implementing a
more realistic distribution such as the one given by Nerling et al. (2003).
The result that we have reached so far is the emission from the maximum
of a 10
17
eV air shower consisting of 10
8
electrons and positrons at a height
of 4 km, taking into account adequate spatial, energy and trajectory length
distributions and a magnetic eld as present in central Europe. It is illustrated
once more as a surface plot in Fig. 3.24.
-600
-400
-200
0
200
400
600
NS [m]
-600
-400
-200
0
200
400
600
EW [m]
0
20
40
60
80
100
120
F 3.24Pattern of the maximum ltered electric eld amplitude (in V m
1
) emitted
by the maximum of a 10
17
eV air shower consisting of 10
8
particles at a height of 4 km for a
0.5 Gauss 70
inclined magnetic eld, a broken power-law particle energy distribution and a

statistical distribution of track lengths.
3.7.4 Comparison with theoretical calculations
As shown in Fig. 3.7 the individual particle pulses calculated with our MC
code are consistent with the results of the analytic calculations. The analytic
pulses are symmetric because the particle trajectories are implicitly adopted
symmetric to the point of smallest angle to the line of sight the analytic
calculations do not consider edge eects associated with the nite lengths of
the trajectories. When one takes into account the cutting o of the trajectories
correctly in the MC calculations, an observer in the centre region only sees
half of the symmetric pulse for each individual particle (cf. Fig. 3.9).
In section 3.6.1 we discussed that the time integral over the individual par-
ticle pulses is the quantity relevant for the overall pulse as integrated over the
shower as a whole. For simple geometries such as a point source or a line
charge, the change from the symmetric to the half pulses should therefore pro-
duce a drop in the overall pulse amplitude (and the overall pulse spectrum) by
a factor 2. (For more complex shower geometries or at higher distances from
the shower centre the changes introduced by the edge eects are non-trivial but
still important as already discussed in section 3.7.1.)
Fig. 3.25 shows the direct comparison between the analytic and MC sim-
0.01
0.1
1
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.25Spectra emitted by the maximum of a 10
17
eV vertical air shower consisting of
10
8
particles with 60 at a height of 4 km, long (constant 100 g cm
2
), but not symmetric,
particle trajectories and horizontal 0.3 Gauss magnetic eld. Thin lines: analytic calculations
from chapter 2, thick lines: these MC simulations. Solid: 20 m, dashed: 100 m, dotted: 260 m
to north from shower centre.
ulated spectra emitted by a slice of monoenergetic 60 particles for the
case of a horizontal 0.3 Gauss magnetic eld and long (constant 100 g cm
2
),
but not symmetric, particle trajectories. (Because the north-south polarisation
component was neglected in the analytic calculations, we hereafter directly
compare the east-west polarisation component.) This scenario allows a very
direct comparison between the analytic and MC calculations, the only major
dierence being the edge eects introduced due to the non-symmetric trajec-
tories. Consequently, the analytic spectra lie a factor 2 above the MC results.
Scaling down the analytic results by a factor of two shows indeed a very good
agreement between the analytic and MC results as seen in Fig. 3.26. The ra-
dial dependence of the emission pattern, shown in Fig. 3.27, also shows good
agreement between the MC and analytic calculations when one accounts for
the systematic factor of two.
In a next step, we switch on the statistical distribution of trajectory lengths
with a realistic mean free path length of 40 g cm
2
, adopt again a broken
power-law distribution of particle energies and change to the realistic 70
in-
clined 0.5 Gauss magnetic eld present in central Europe. In the analytic cal-
culations, we switch on the broken power-law distribution, but cannot take into
0.01
0.1
1
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.26Same as Fig. 3.25 but scaling down the analytic results by a factor of two.
0.001
0.01
0.1
1
0 100 200 300 400 500
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 3.27Radial dependence of the emission for the same scenario as in Fig. 3.25. Thin
lines: analytic calculations from chapter 2 scaled down by a factor of two, thick lines: these
MC simulations. Solid: = 50 MHz, dashed: = 75 MHz, dotted: = 100 MHz. Distance is
0.01
0.1
1
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.28Spectra emitted by the maximum of a 10
17
eV vertical air shower consisting
of 10
8
particles with broken power-law energy distribution at a height of 4 km, statistically
distributed particle trajectories with 40 g cm
2
mean path length and 70
inclined 0.5 Gauss

magnetic eld. Thin lines: analytic calculations from chapter 2 scaled down by a factor of two,
thick lines: these MC simulations. Solid: 20 m, dashed: 100 m, dotted: 260 m to north from
shower centre.
account the track length eects or the inclined magnetic eld. Obviously, the
results produced by the MC simulations in this scenario therefore cannot be
expected to reproduce the analytic results equally well.
Adirect comparison between the spectral dependences predicted by the an-
alytic and MC calculations is shown in Fig. 3.28, keeping the down-scaling of
the analytic results by a factor of two to compensate for the symmetric trajec-
tories. As discussed earlier, switching on the energy distribution redistributes
ux from medium scales to the centre region. Correspondingly, the spectrum
at 20 m distance shifts to higher amplitudes. The eect is stronger in the MC
calculations than in the analytics. The MC 100 m spectrum ts well with the
analytic results, whereas the MC 260 m spectrum lies at lower amplitudes than
in the analytics: The radial dependence now falls o much steeper in the MC
as compared to the analytic results, as is also visible in Fig. 3.29. Overall, the
radial dependence follows a very much exponential decay. (We will carry out
a detailed analysis of the functional form of dierent properties of the radio
emission in a later paper.) It does not exhibit the prominent plateaus visible in
the centre regions of the analytic calculations. Considering the much higher
3.8. Emission from an integrated shower 93
0.001
0.01
0.1
1
0 100 200 300 400 500
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 3.29Radial dependence of the emission for the same scenario as in Fig. 3.28. Thin
lines: analytic calculations from chapter 2 scaled down by a factor of two, thick lines: these
MC simulations. Solid: = 50 MHz, dashed: = 75 MHz, dotted: = 100 MHz. Distance is
precision of the MC simulations and the neglect of edge eects and statisti-
cal trajectory lengths in the analytic calculations, however, we consider the
agreement quite acceptable.
3.8 Emission from an integrated shower
After having analysed the emission from a shower slice, the next step in our
analysis now is to perform the integration over the air shower as a whole.
3.8.1 Integration over shower evolution
In the theoretical calculations we performed in chapter 2, the integration over
the shower evolution was carried out in a somewhat simplied way: The
shower evolution was discretised into slices of independent generations of par-
ticles, spaced apart by one radiation length each. The overall emission was
then superposed as the sum of the radiation from all these slices. Although
this strategy should allow a good estimate of the emission from the complete
shower, it has at least two problems: First, the total number of particles in the
shower and thus the total eld amplitude is directly inuenced by the
scale introduced through the spacing of the slices. A denser or wider spacing
94 3.8. Emission from an integrated shower
directly leads to higher or lower emission levels, respectively. Although the ra-
diation length is the logical choice for this scale, a scale-free approach would
be a better choice. Second, our theoretical calculations, strictly speaking, are
only valid in the far-eld. Consequently, the emission from slices close to the
ground, especially for high-energy showers, cannot be taken into account with
the desired precision.
Both these pitfalls no longer pose a problem in the MC simulations: No
far-eld approximations at all were applied in the MC calculations, and the
continuous evolution of the shower is correctly taken into account in the cre-
ation of particles according to the corresponding probability distribution func-
tion.
For the shower prole we use the Greisen parametrisation (Greisen 1960)
that we already adopted in chapter 2:
N(s) =
0.31 exp
_
X
m
X
0
23 ln s
3/s1
_
X
m
/X
0
, (3.11)
where the (theoretical) position of the shower maximum X
m
is given by
X
m
= X
0
ln
_
E
p
/E
crit
_
, (3.12)
X
0
= 36.7 g cm
2
denotes the electron radiation length in air, E
crit
= 86 MeV
corresponds to the threshold energy where ionisation losses equal radiation
losses for electrons moving in air, and E
p
species the primary particle energy.
Equation (3.12) predicts the position of the shower maximum for a purely
electromagnetic cascade, in which the shower age as a function of atmospheric
depth then corresponds to
s(X) =
3X
X + 2X
m
. (3.13)
Obviously, this parametrisation of the shower age has to be modied to ade-
quately describe the evolution of the hadronic air showers in our MC simu-
lations. We choose to manually set the depth of the shower maximum to an
empirical value X
m,e
as a function of primary particle energy and shower incli-
nation, e.g. X
m,e
630 g cm
2
for our typical 10
17
eV vertical air shower. The
shower age is then adopted as
s(X) =
3X
X + 2X
m,e
, (3.14)
whereas eq. (3.11) is left unchanged, i.e. retaining the theoretically motivated
value for X
m
.
1
Eq. (3.11) denotes the integrated number of electronic particles that a
detector positioned at atmospheric depth X measures as the shower sweeps
through it. This number is not equal to the number of particles I(X) that are
injected at that atmospheric depth, the quantity we need to describe the prob-
ability distribution function for the creation of particles. The two quantities are
directly related via the path length distribution of the particles. For an expo-
nential path length distribution with mean free path length as given in eq.
(3.10), the injection function is given by
I(X) =
dN
dX
+
N
. (3.15)
As demonstrated in Fig. 3.30, I(X) closely follows the form of N(X) with an
oset of to lower X values, i.e.
I(X) N(X + )/. (3.16)
The shower evolution is thus taken into account in a continuous and con-
sistent way by random creation of particles with probabilities according to eq.
(3.15). Fig. 3.31 illustrates the shower evolution through the particle trajecto-
ries that are followed during the simulation of our 10
17
eV vertical air shower.
3.8.2 Integrated shower results
As in the theoretical calculations, the integration over the shower evolution has
two main eects, visible in Fig. 3.32: First, the emission level is boosted sig-
nicantly. This directly shows that the emission is not described suciently
by just taking into account the shower maximum. The second major eect is a
steepening of the radial emission pattern due to the amplication of coherence
losses. It is mainly the centre region which receives signicant additional radi-
ation. The steepness of the radial dependence is also illustrated by the strong
drop in the maximum amplitude of the ltered pulses when one goes to even
moderate distances of 260 m as shown in Fig. 3.33.
Another important eect is the further fading away of sharp features and
1
A more realistic set of parameters for application of the Greisen function to hadronic show-
ers in the energy range between 10
17
and 10
18
eV was established in Abu-Zayyad et al. (2001).
For the moment, however, we retain the parametrisation as stated above to allow a better com-
parison with our earlier theoretical calculations.
0
0.5
1
1.5
2
0 200 400 600 800 1000
0
1
2
3
4
5
6
7
8
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
X [g cm
2
]
I
(
X
)
[
1
0
6
g
1
c
m
2
]
N
(
X
)
[
1
0
7
]
F 3.30Comparison of I(X) (solid) and N(X) (dashed) as a function of atmospheric
depth X for a vertical 10
17
eV shower with X
m,e
= 631 g cm
2
and = 40 g cm
2
.
-1000
-500
0
500
1000
-1000
-500
0
500
1000
0
2000
4000
6000
8000
10000
12000
14000
N-S [m]
E-W [m]
F 3.31Trace of the trajectories in a complete 10
17
eV air shower.
1
10
100
1000
0 50 100 150 200 250 300 350
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
[
V
m
1
]
F 3.32Eects introduced in the radial dependence of the 40160 MHz rectangle-
ltered pulse amplitude by the integration over the shower evolution. Thick lines: integrated
10
17
eV vertical shower with broken power-law particle energy distribution, statistical track
length distribution with = 40 g cm
2
and 70
inclined 0.5 Gauss magnetic eld, thin lines:

only shower maximum in same scenario. Solid: to the north, dashed: to the west.
the asymmetries associated with the geomagnetic eld in the emission pat-
tern, as illustrated by Fig. 3.34. While some of these features were still quite
prominent in the emission from a single shower slice, they more or less vanish
completely as soon as the integration is performed. Similarly, the eects of
individual changes to the particle distributions, such as the specic choice of
an energy distribution, are therefore much weaker than they are in case of a
single shower slice.
The polarisation characteristics of the integrated shower are very similar to
those of an individual slice. Again, the emission is highly polarised in the east-
west direction, except for some regions at medium distances of 200400 m
as illustrated in Fig. 3.34 and, in a more quantitative way, Fig. 3.35.
Figure 3.36 shows a direct comparison of the radial proles of the particle
density (as given by the NKG-parametrisation) and the ltered radio pulse am-
plitude. The particle density falls o much more steeply than the radio signal
in the central 200250 metres. Further out, the slope becomes comparable
for radio amplitude and particle density.
-1500
-1000
-500
0
500
1000
1500
-20 0 20 40 60 80 100 120
time [ns]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
E
E
W
(
t
)
[
V
m
1
]
F 3.33Pulses in the east-west polarisation component after 40160 MHz rectangle-
ltering for an integrated shower as described in Fig. 3.32. Solid: in the shower centre, long
dashed: 100 m to north of centre, short dashed: 260 m to north of centre.
-300
-200
-100
0
100
200
300
-300 -200 -100 0 100 200 300
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-300
-200
-100
0
100
200
300
-300 -200 -100 0 100 200 300
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-300
-200
-100
0
100
200
300
-300 -200 -100 0 100 200 300
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-300
-200
-100
0
100
200
300
-300 -200 -100 0 100 200 300
d
is
t
a
n
c
e

N
S

[
m
]
distance EW [m]
F 3.34Contour plots of the 40160 MHz rectangle- ltered pulse amplitude for the
full shower described in Fig. 3.32. Contour levels are 20 V m
1
apart. From left to right: total
electric eld strength, north-south polarisation component, east-west polarisation component,
vertical polarisation component.
0.001
0.01
0.1
1
10
0 50 100 150 200 250 300 350
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
E
x
/
y
(
R
,
V
m
1
M
H
z
1
]
F 3.35Radial dependence of E(R, ) for dierent polarisation components at =
55 MHz for the same scenario as in Fig. 3.32. Solid: east-west polarisation to the north from
centre, long-dashed: north-south polarisation to the north from centre, short-dashed: east-west
polarisation to the north-west from centre, dotted: north-south polarisation to the north-west
from centre.
0.001
0.01
0.1
1
10
100
50 100 150 200 250 300 350
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
M
a
x
(
|
E
(
t
)
|
)
v
e
r
s
u
s
N
K
G
F 3.36Comparison of the ltered radio pulse amplitude (solid) and the particle density
as given by the NKG-parametrisation (dashed) as a function of radial distance from the shower
centre. Values were normalised to unity at r = 60 m.
3.8.3 Comparison with theoretical calculations
We now compare the results of our MC simulations of a fully integrated
10
17
eVvertical air shower with the theoretical calculations performed in chap-
ter 2. Fig. 3.37 shows the spectral dependence of the emission in direct com-
parison. The MC results again produce somewhat lower levels of radiation.
Scaling down the theoretical results by the systematic factor of two introduced
in section 3.7.4, the agreement is much better, as shown in Fig. 3.38. Con-
sidering the huge dierences in the two calculations, the agreement is quite
remarkable.
In chapter 2 we compared the theoretical results with the available histori-
cal data. As discussed there in detail, the absolute values of the historical ex-
perimental data are very uncertain and largely discrepant between the dierent
groups. Additional uncertainty arises from ambiguities in the exact denition
of the historical values denoted as
and their conversion to the theoretical

quantity |E(R, )|, which we performed via the relation
=
_
128
|E(R, )| 6.4 |E(R, )| . (3.17)

To deal with the discrepant sets of data we decided to take the well documented
Allan (1971) data as our reference and rescale the spectral data of Prah (1971)
and Spencer (1969) to be consistent with the Allan data at = 55 MHz. In this
work, we use the identically rescaled data for comparison with our new MC
spectra. While the absolute values of the spectral data are therefore somewhat
arbitrary, they at least allow an evaluation of the qualitative spectral depen-
dence. The spectral data are over-plotted in Fig. 3.38. The agreement with the
spectral dependence in the centre region of the shower is very good.
In Fig. 3.39 we compare the radial dependence of the emission as calcu-
lated by our MC code with the earlier theoretical results and the Allan (1971)
data. The scaled-down theoretical results again show good agreement with the
MC results in the centre region. Overall, the MC predicts a somewhat steeper
decrease of the emission levels to higher distances. It was, however, expected
that the theoretical calculations overestimate the coherence and thus emission
levels at high distances. The absolute level of the emission also agrees with
the Allan data within the uncertainties.
0.01
0.1
1
10
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.37Spectra emitted by a complete 10
17
eV vertical air shower with maximum
at 4 km height, broken power-law particle energy distribution, statistically distributed particle
trajectories with 40 g cm
2
mean path length and 70
inclined 0.5 Gauss magnetic eld. Thin

lines: analytic calculations from chapter 2, thick lines: these MC simulations. Solid: shower
centre, dashed: 100 m, dotted: 250 m to north from shower centre.
0.01
0.1
1
10
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 3.38Same as Fig. 3.37 but scaling down the analytic results by a factor of two.
Data from Prah (1971) (gray) and Spencer (1969) (black) were rescaled to be consistent with
the Allan (1971) data at = 55 MHz.
102 3.9. Discussion
0.01
0.1
1
10
0 50 100 150 200 250 300 350
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 3.39Radial dependence of the emission at = 55 MHz for the same scenario as
in Fig. 3.37. Dashed: analytic calculations from chapter 2. Solid: analytic calculations from
chapter 2 scaled down by a factor of two (thin line) in comparison with these MC simulations
(thick line). Data from Allan (1971). Distance is to north from shower centre.
3.9 Discussion
With the design and implementation of our MC simulation, we have taken the
logical next step in the development of our geosynchrotron radiation model
for radio emission from cosmic ray air showers. The MC technique provides
an independent cross-check on our earlier theoretical works and allows us to
model the emission from the air shower with much higher precision based on
a much more realistic air shower model.
To make these simulations feasible on standard o-the-shelf computer
hardware, we conceived and implemented a number of intelligent concepts. In
particular, the smart sampling of the particle trajectories in conjunction with
the automatic inactivation of ground-bins and the adaptive collection of the
time-series data provide the necessary cut-down on computation time com-
pared to a simple brute-force approach. At the same time, high precision is
retained in the results as demonstrated by the detailed consistency checks pre-
sented in section 3.6. The calculation of a typical vertical 10
17
eV air shower
with realistic shower properties and 25, 000, 000 particles combined with au-
tomatic ground-bin inactivation at very high precision takes about 400 seconds
per ground-bin on a standard PC of the 1.6 GHz class. A very detailed calcu-
3.9. Discussion 103
lation with 800 ground-bins (25 radial, 32 azimuthal), parallelised on 4 PCs,
thus takes about one day.
A major new result of our MC simulations is the profound inuence of
edge eects arising from the nite lengths of the particle trajectories. These
could not be taken into account in the theoretical calculations. In combi-
nation with the dependence of the individual particle emission on the mag-
netic eld strength, the edge eects associated with the statistical distribution
of track lengths wash out the asymmetries originally introduced into the to-
tal eld strength emission pattern by the geomagnetic eld to a high degree.
Once the integration over the shower evolution as a whole is carried out, the
asymmetry is gone completely. The absence of asymmetries in the total eld
strength emission pattern from an integrated shower is somewhat unfortunate,
as a prominent asymmetry would have allowed an easy, yet unambiguous test
of the geomagnetic emission mechanism via the statistics of radio pulse total
eld strengths alone.
The decomposition of the electric eld into north-south, east-west and
vertical polarisation components, however, shows that the emission is indeed
highly polarised in the direction perpendicular to the shower axis and the mag-
netic eld direction, as predicted by our theoretical calculations. Polarisation-
dependent radio measurements such as the ones carried out by LOPES could
therefore still unambiguously establish the geomagnetic emission mechanism.
The MC results show good consistency with the theoretical spectra and ra-
dial dependences apart from a systematic factor of two in emission strength
which is plausible considering the implicit assumption of symmetric trajecto-
ries in the analytic calculations. Such good agreement between the theoretical
and MC calculations is remarkable considering the inherent dierences in the
two approaches. In particular, the integration over the shower evolution as a
whole is carried out in a much more sophisticated way in the MC simulations
as compared to the analytical works. As mentioned above, the polarisation
characteristics of the emission are exactly as inferred in the theoretical calcu-
lations.
The spectra and radial dependences predicted by our MC code also agree
well with the historical data of Allan (1971) and the data of Prah (1971) and
Spencer (1969) scaled to the absolute level of the Allan data. The necessary
rescaling, however, demonstrates that the historical data themselves are largely
discrepant with absolute values reaching up to an order of magnitude lower
than the Allan data (see chapter 2 for a detailed discussion). It is therefore
still imperative to gather independent data with good absolute calibration with
new experiments such as LOPES (Horneer et al. 2003). For air showers of
energies around 10
17
eV this should be well feasible as the predicted absolute
levels of emission |E(R, )| around a few V m
1
MHz
1
at 55 MHz are well
above the Galactic noise limit of 0.4/0.15/0.05 V m
1
MHz
1
for a 3
detection with 1/10/100 LOPES antenna(s) (see chapter 2). In areas with high
radio-frequency interference levels such as the site of the KASCADE array,
the noise levels are a factor of a few higher.
3.10 Conclusions
We have successfully advanced our modelling of radio emission from cosmic
ray air showers with elaborate Monte Carlo simulations in the time-domain.
Our MC code takes into account the important air shower characteristics such
as lateral and longitudinal particle distributions, particle energy and track
length distributions, a realistic magnetic eld geometry and the evolution of
the air shower as a whole. The calculation retains the full polarisation infor-
mation and does not employ any far-eld approximations.
We predict emission patterns, radial and spectral dependences for an ex-
emplary 10
17
eV vertical air shower and nd good agreement with our earlier
theoretical works and the historical data available.
A major result that could not be obtained by analytic calculations alone
is that asymmetries introduced into the total eld strength emission pattern by
the magnetic eld direction are washed out completely in the radiation from an
integrated air shower. Statistics of total eld strengths alone can therefore not
establish the geomagnetic emission mechanism. The clear polarisation depen-
dence on the magnetic eld direction, on the other hand, allows a direct test
of the geomagnetic emission mechanism through polarisation-sensitive exper-
iments such as LOPES.
After having documented the implementation details and having demon-
strated the correctness and robustness of our MC simulations, our code is now
in a stage where we can explore the dependence of the radio emission on a
number of parameters such as shower axis direction, primary particle energy,
depth of shower maximum and the like. Consequently, this will be our next
step.
Once these dependences are established, measurements of radio emission
from cosmic ray air showers can be related directly to the underlying charac-
teristics of the observed air showers. Due to the regularity and robustness of
the modelled emission patterns, even a sparse sampling of the radiation pattern
with a limited number of antennas would probably suce for such an analysis.
Furthermore, our code provides a solid basis for the inclusion of additional
eects such as Askaryan-type (Askaryan 1962, 1965)

Cerenkov radiation and
an interfacing to the MC air shower simulation code CORSIKA.
Appendix: Trajectory
Consider rst a simple magnetic eld geometry
B = B
_
_
0
0
1
_
_
. (3.18)
The (unperturbed) trajectory of a charged particle in a homogeneous magnetic
eld is a helix. Aligning the helix along the z-axis, it can be written as
r(t) =
_
_
R
B
cos [
B
(t t
0
)]
+R
B
sin [
B
(t t
0
)]
v cos (t t
0
)
_
_
, (3.19)
where
R
B
=
v sin
B
(3.20)
denotes the radius of the projected circular motion and
B
=
qeB
m
e
c
(3.21)
is the gyration frequency of the particle with charge q times the elementary
charge unit e and velocity
v = c =
_
1
1
2
c. (3.22)
The pitch-angle is given by the (constant) angle between the direction of the
particle velocity vector and the magnetic eld vector.
To derive the general form of this trajectory, we rst rotate the coordinate
system such that the B-eld points in the desired direction. Afterwards, we
adjust the phase t
0
such that the particles initial velocity has the desired direc-
tion as specied by the initial velocity vector V. In the last step we shift the
trajectory so that at t = 0 it coincides with the desired starting position R.
We want to transform the simple geometry eld

B to the desired geometry
B = B
_
_
cos sin
sin sin
sin
_
_
, (3.23)
where [0, 2[ and [0, ] are the azimuth and zenith angles known from
spherical coordinates. This transformation is achieved by applying a rotation
matrix
D =
_
_
cos sin 0
sin cos 0
0 0 1
_
_
cos 0 sin
0 1 0
sin 0 cos
_
_
=
_
_
cos cos sin sin cos
cos sin cos sin sin
sin 0 cos
_
_
, (3.24)
so
B = D

B (3.25)
and, inversely,
B = D
1
B. (3.26)
Applying the same rotation to the trajectory (3.19) yields
r(t) = D r(t). (3.27)
To infer the phase t
0
corresponding to a given initial velocity V, we rotate back
V to the simple geometry,
V = D
1
V. (3.28)
The x- and y-component of

V then directly determine t
0
through the relation
B
t
0
= arctan
_
V
x
V
y
_
, (3.29)
with the upper sign for q > 0 and the lower for q < 0 and where one has to
take into account the correct quadrant for the arctan operation. In terms of the
components of V this yields
t
0
= arctan
_
_
(V
x
cos + V
y
sin ) cos V
z
sin
_
V
y
cos V
x
sin
_
_
/
B
. (3.30)
The last operation that has to be employed is a translation
r
abs
(t) = R
0
+ r(t) (3.31)
of the trajectory such that
r
abs
(t = 0) = R
0
+ r(t = 0) = R, (3.32)
which yields
R
0
= R + v cos t
0
_
_
sin cos
sin sin
cos
_
_
+ R
B
_
_
cos cos cos[
B
t
0
] sin sin[
B
t
0
]
cos sin cos[
B
t
0
] + cos sin[
B
t
0
]
sin cos[
B
t
0
]
_
_
. (3.33)
The resulting trajectory r
abs
(t) is thus fully dened for a given set of parameters
R and V. The time-dependence of particle velocity and acceleration are then
easily derived as the time-derivatives of r
abs
(t).
4
Simulation Results
4.1 Introduction
In chapter 3 we described details of the design and implementation of our
Monte Carlo simulations of radio emission from cosmic ray air showers in
the scheme of coherent geosynchrotron radiation. These simulations are based
on analytic parametrisations of the air shower characteristics and constitute
a precursor to our upcoming full-edged Monte Carlo simulations of radio
emission from extensive air showers based on precise air shower modeling
with CORSIKA (Heck et al. 1998).
In this article, we present the results inferred so far from the simulations
performed with our code. After a short description of the underlying simu-
lation parameters in section 4.2, we describe important characteristics of the
radio emission in general such as the radial dependence, the spectral depen-
dence, the curvature of the radio front and the polarisation characteristics of
the radiation (which play an important role in experimentally verifying the
dominant emission mechanism) in section 4.3. We analyse the dependence
of these characteristics on the associated air shower parameters such as the
shower geometry (zenith and azimuth angle), the primary particle energy, the
depth of the shower maximum and the magnetic eld in a qualitative way in
section 4.4. Afterwards, we parametrise the emissions dependence on the var-
ious observer and shower parameters in a number of individual formulas for
our reference shower (section 4.5) before generalising the parametrisations to
arbitrary shower geometries and piecing together an overall parametrisation
incorporating all dependences in section 4.6. We discuss our results in section
109
110 4.2. Simulation parameters
4.7 and conclude the chapter in section 4.8.
4.2 Simulation parameters
All simulations presented here were done with the Monte Carlo code described
in chapter 3. In this section we specify the simulation strategy as well as the
parameters and conguration options that were used throughout this work.
Obviously, we cannot perform a true n-dimensional analysis of the param-
eter space in question. We therefore choose a vertical 10
17
eV air shower as a
reference and change only one of the shower parameters at a time to analyse
its eect on the radio emission. This implies that the eects introduced by
changes of the dierent parameters are well-separable. We pay special atten-
tion in cases where this is obviously not true (e.g., primary particle energy and
depth of the shower maximum).
Our reference air shower is calculated with a primary particle energy of
10
17
eV, developing to its maximum at an atmospheric depth of 631 g cm
2
as
originally adopted in chapter 2. This corresponds to a distance of 4 km to
the ground in case of a vertical air shower.
The following settings are kept throughout all simulations if not explicitly
stated otherwise (for denitions of the terms in quotation marks see chapter
3): The particle track lengths are distributed following an exponential proba-
bility distribution with mean track length of 36.7 g cm
2
for both electrons and
positrons. The particle energies are set to follow a broken power-law distribu-
tion peaking at = 60 as described in chapter 2. The magnetic eld is chosen
with a strength of 0.5 Gauss and an inclination of 70
, which approximately
corresponds to the conguration present in central Europe. Calculations are
done on a simple grid of 1 ns resolution with smart trajectory sampling
enabled. Automatic ground-bin inactivation is used with a precision goal of
0.25% in 4 consecutive blocks of 10,000 particles each up to a maximum of
25,000,000 particles. A total of 800 bins (32 in azimuth; 25 in radius, up to a
distance of 1000 m) is calculated in each simulation.
4.3 General characteristics
First, we present the general characteristics of the radio emission from a pro-
totypical 10
17
eV vertical air shower, which we take as the reference shower in
this work.
4.3. General characteristics 111
4.3.1 Spectral dependence
In Fig. 4.1 we present the spectral dependence of the emission from a 10
17
eV
vertical air shower at various distances from the shower centre. The spectra
show a steep decline towards higher frequencies due to the coherence dimin-
ishing as the wavelengths become shorter and thus comparable to the scales
present in the shower pancake. The eld strength reaches a rst interference
minimum at a distance-dependent frequency. Afterwards, we see a quickly al-
ternating series of maxima and minima that are insuciently sampled in this
calculation and therefore give rise to the unphysically seeming features at high
frequencies. A realistic modelling of the emission in this incoherent regime
would need a more detailed air shower model taking into account the inhomo-
geneities that are known to be present in the shower cascade. This cannot be
achieved with the currently used analytic parametrisations of air shower prop-
erties, but will be accomplished once our code is interfaced to the air shower
simulation code CORSIKA.
The higher the distance from the shower centre, the steeper the spectral de-
pendence of the emission. In other words, coherence is much better up to high
distances at lower frequencies as compared with higher frequencies: while the
emission is coherent to large distances of >500 m for the 10 MHz frequency
component, it already becomes incoherent at 300 m for the 55 MHz fre-
quency component.
Figure 4.2 shows the spectra of the same 10
17
eVvertical air shower plotted
in a S
diagram, illustrating that most of the power is emitted at frequencies

around 2030 MHz. (The conversion from electric eld frequency compo-
nent E() to the frequency-domain equivalent of the Poynting vector S
is
performed according to eq. (2.14), page 27.)
These two eects strongly point to low frequencies as the most promising
regime for observation of cosmic ray air showers with radio techniques.
To demonstrate the dierences for emission at low and high frequencies,
we compare some of the results presented in the following sections for the two
prototypical frequencies of 10 MHz (good coherence up to high distances as
desirable for experimental measurements) and 55 MHz (frequency band used
in the historical works and LOPES, cf. Horneer et al. (2004), but coherence
only up to medium distances).
112 4.3. General characteristics
0.001
0.01
0.1
1
10
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 4.1Spectra of the emission from a vertical 10
17
eV air shower at various distances
to the north. From top to bottom: 20 m, 140 m, 260 m, 380 m and 500 m.
1e-24
1e-23
1e-22
1e-21
1e-20
1e-19
1e-18
1e-17
1e-16
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
(
E
W
)
[
J
m
2
]
[MHz]
F 4.2Same spectra as in gure 4.1 plotted in a S
diagram.
4.3.2 Radial dependence and emission pattern
Fig. 4.3 shows the 10 MHz component of the electric eld strength in the indi-
vidual linear polarisation directions north-south, east-west and vertical.
The total eld strength pattern is remarkably symmetrical in spite of the intrin-
sic asymmetry of the geomagnetic emission mechanism. A more quantitative
view of the radial dependence of the emission is depicted in Fig. 4.4.
Please note that we can equivalently use the east-west polarisation com-
ponent or the total eld strength in many of the following analyses as there is
no ux in the north-south (let alone vertical) polarisation component along the
north-south direction from the shower centre for air showers coming from the
south.
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
i
s
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
i
s
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
i
s
t
a
n
c
e

N
S

[
m
]
distance EW [m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
i
s
t
a
n
c
e

N
S

[
m
]
distance EW [m]
F 4.3Contour plots of the 10 MHz eld strength for emission from a 10
17
eV vertical
air shower. From top left to bottom right: total eld strength, north-south polarisation com-
ponent, east-west polarisation component, vertical polarisation component. Contour levels are
0.25 V m
1
MHz
1
apart. There is no signi cant ux in the vertical component.
114 4.3. General characteristics
0.01
0.1
1
10
0 100 200 300 400 500 600 700 800
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.4Radial dependence of the 10 MHz emission from a 10
17
eV vertical air shower.
Solid: to the north, dashed: to the north-west, dotted: to the west.
4.3.3 Wavefront curvature
The radio wavefront arriving at the ground is not planar. As demonstrated
in Fig. 4.5, the pulses systematically lag behind at higher distances from the
shower centre. The curvature of the wavefront can be approximated by a spher-
ical surface with a given radius. At distances beyond a few hundred metres,
this approximation, however, breaks down. (Additionally, the curvature ra-
dius derived from the timestamps of the maximum ltered pulse amplitudes
depends on the specic lter used.) The scatter seen in the plot is not of sta-
tistical nature, but rather represents the slight time-shift in the pulses peak
amplitude as a function of azimuth angle.
The curvature of the radio wavefront plays an important role for the beam-
forming performed in digital radio interferometers and has indeed been con-
rmed by LOPES measurements (Horneer, private communication).
4.3.4 Linear polarisation
The radio emission generated by the geosynchrotron mechanism is intrinsi-
cally linearly polarised. Figure 4.6 shows the raw (unltered) pulses arriving
at a distance of 200 m to the north-west from the centre of a 10
17
eV vertical air
-5
0
5
10
15
20
25
0 50 100 150 200 250 300 350 400 450 500
l
a
g

o
f

p
u
l
s
e

p
e
a
k

[
m
]
F 4.5Radio wavefront curvature given by the time-lag of the east-west polarisation
raw pulses emitted by a vertical 10
17
eV air shower. Dotted line: curvature given by a spherical
surface with 6,500 m radius.
shower. The north-south and east-west polarisation components are of similar
strength and arrive almost synchronously. The vertical polarisation-component
is negligible.
Figure 4.7 shows the same data (neglecting the vertical component) vi-
sualised as a scatter plot. For each time-step of the simulated pulse, a point
specifying the north-south versus east-west eld strength component is drawn.
In other words, the series of points directly illustrates the evolution of the (pro-
jected) electric eld vector. The very narrow loop performed by the vector
in the upper-left quadrant of the diagram demonstrates that the emission is in-
deed linearly polarised to a very high degree, even at the already moderate
distances presented here. (In case of perfect linear polarisation, the series of
points would all lie on a straight line, whereas for perfect circular polarisation,
the polarisation vector would follow a full circle around the origin.)
In the centre regions where the emission is strongest, the radiation is almost
perfectly linearly polarised. In these regions, the polarisation vector points
in the direction perpendicular to the air shower and magnetic eld axes, as
predicted in chapter 2, cf. section 4.4.2.
116 4.4. Qualitative dependence on shower parameters
-200
-150
-100
-50
0
50
100
150
200
250
300
0 20 40 60 80 100 120
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
i
(
R
,
t
)
|
[
V
m
1
]
t [ns]
F 4.6Raw(un ltered) pulses in the individual linear polarisation components at 200 m
distance to the north-west from the centre of a 10
17
eV vertical air shower. Solid: east-west
component, dashed: north-south component, dotted: vertical component.
4.4 Qualitative dependence on shower parameters
In the following subsections, we present a number of dependences of the radio
emission on specic air shower parameters in a qualitative way.
4.4.1 Shower zenith angle
An interesting question is that of the radio emissions dependence on the air
shower geometry. Fig. 4.8 shows the radial dependence of the 10 MHz fre-
quency component for air showers coming from the south with dierent zenith
angles. It is well visible that the radial dependence in the north (i.e., shower
axis) direction becomes much atter with increasing zenith angle.
A broadening of the emission pattern in the shower axis direction could
be intuitively expected from projection eects occurring when the air shower
is inclined. One can remove these projection eects by changing the coor-
dinate system from the ground-based distance to the shower centre to the
shower-based (perpendicular) distance to the shower axis. (The electric eld
vector, however, is still denoted with the ground-based north-south, east-west
and vertical components which thus do not change in strength. This method of
4.4. Qualitative dependence on shower parameters 117
-300
-200
-100
0
100
200
300
-300 -200 -100 0 100 200 300
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
E
NS
(R, t) [V m
1
]
E
E
W
(
R
,
t
)
[
V
m
1
]
F 4.7Scatter plot of the north-south and east-west polarisation components shown in
Fig. 4.6. The emission is linearly polarised to a high degree.
back-projection is not the same as an inclination of the full ground-plane.)
Fig. 4.9 shows the back-projected radial dependences for the 10 MHz emis-
sion. It is obvious that the attening is still present and thus cannot simply be
caused by projection, but is an intrinsic feature of the emission. The emission
pattern broadens as a whole (even in the direction perpendicular to the shower
axis) as can be seen when comparing the back-projected patterns for a 45
in-
clined air shower depicted in Fig. 4.13 with that of a vertical shower shown in
Fig. 4.3.
The overall broadening of the emission pattern is due to the fact that the air
shower maximum for inclined showers is much further away from the ground
than for vertical showers. This eect was already predicted from geometri-
cal/qualitative arguments by Gousset et al. (2004). It makes inclined air show-
ers an especially interesting target for observation with radio techniques.
The slight deviation of the 15
zenith angle curve from the trend seen in

Fig. 4.8 is explained by the showers very small angle of only 5
to the 70
in-
clined geomagnetic eld. The weakness of this deviation alone demonstrates
that the dependence of the emission on the strength and orientation of the ge-
omagnetic eld is very slight except regarding the polarisation eects anal-
ysed in Sec. 4.4.2. Consequently, the same diagram for air showers coming
from the north (not shown here) looks very similar.
Fig. 4.10 shows the zenith angle dependence for the 55 MHz frequency
component. The overall trend is the same as in the 10 MHz case, but the co-
herence losses cut o the emission pattern at a zenith angle dependent distance
of a few hundred metres. At zenith angles
>30
, however, the coherence be-

gins to hold up to high distances. This is conrmed by the spectra of a 45
inclined 10
17
eV air shower shown in Fig. 4.11. The spectra are much atter
up to high distances when compared with the vertical case in Fig. 4.1.
Inclined air showers thus not only oer signicantly broader emission re-
gions on the ground, but provide the advantage that the larger nger-print
even extends to signicantly higher frequencies.
0.001
0.01
0.1
1
10
0 100 200 300 400 500 600 700 800 900 1000
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.8Dependence of the 10 MHz east-west electric eld component emitted by a
10
17
eV air shower coming from the south for dierent shower zenith angles as a function of
distance to the north. Red/solid: vertical shower, green/dashed: 15
, blue/dotted: 30
, vio-
let/short dotted: 45
, turquois/dash-dotted: 60
, black/double-dotted: 75
zenith angle.
4.4.2 Shower azimuth angle and polarisation
A very important trait of the radio emission is its predicted polarisation, which
is directly related to the shower azimuth angle. Knowledge of this depen-
dence is imperative for the planning and interpretation of experimental mea-
surements.
0.001
0.01
0.1
1
10
0 100 200 300 400 500 600 700 800 900 1000
distance from shower axis [m]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.9Same as Fig. 4.8 back-projected to the shower-based coordinate system (see
text).
0.001
0.01
0.1
1
10
0 100 200 300 400 500 600 700 800 900 1000
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.10Same as Fig. 4.8 for the 55 MHz frequency component.
0.001
0.01
0.1
1
10
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 4.11Spectra of the emission from a 10
17
eV air shower with 45
zenith angle at
various distances to the north. From top to bottom: 20 m, 140 m, 260 m, 380 m and 500 m.
Fig. 4.12 shows a comparison of the emission at 10 MHz from 10
17
eV air
showers with 45
zenith angle as a function of azimuth angle. The total eld

strength pattern is elongated due to the projection eects arising at high shower
zenith angles (cf. section 4.4.1). Taking out the pure projection eects leads
to the patterns depicted in Fig. 4.13. The patterns are much more circular, but
retain a signicant intrinsic ellipticity and asymmetry.
The total eld strength pattern of the emission (left column) simply rotates
as a function of azimuth angle. (Deviations from a pure rotation are caused by
the symmetry-breaking due to the magnetic eld and shower axes the emis-
sion pattern is no longer supposed to be truly symmetric.) In other words, no
signicant information associated to the geomagnetic eld direction is present
in the signal. As a direct consequence, it is not possible to verify the geo-
magnetic origin of the emission with an experiment measuring only the total
eld strength (or only one circular polarisation component) of the emission.
Furthermore, because the air showers arrive isotropically from all azimuthal
directions, there will not be any azimuthal dependence of the measured pulse
amplitudes in statistical samples of measured total eld strength pulses. It will
therefore not be possible to conrm air showers as the source of measured
radio pulses from statistics of total eld strength data alone. In this case, inde-
pendent information about the simultaneous arrival of cosmic rays, e.g. from
particle detectors, would be necessary.
The situation is dierent when one looks at the individual linear polarisa-
tion components. (As demonstrated in section 4.3.4, the emission is polarised
linearly to a very high degree.) It is visible from the right three columns of
Fig. 4.12 that the eld strength in the dierent polarisation directions has a
direct dependence on the geomagnetic eld direction: the signal is linearly po-
larised mainly in the direction perpendicular to the air shower and magnetic
eld axes, at least in the central regions of high emission. The non-zero con-
tributions in the vertical polarisation direction arise from the 70
inclination of
the geomagnetic eld. Fig. 4.14 illustrates the polarisation characteristics of
the central emission region in a more intuitive way through indicators denoting
the ratio of north-south to east-west polarisation component overplotted over
the total eld strength contours.
Due to these polarisation characteristics, experiments which measure the
polarisation characteristics of the emission can therefore directly verify the
geomagnetic origin of the radio emission from cosmic ray air showers.
4.4.3 Magnetic eld
As discussed in section 4.4.2, the magnetic eld has important inuence on
the polarisation characteristics of the radio emission. The inuence on the
total electric eld strength, however, is very weak.
In Fig. 4.15 we compare the 10 MHz total eld strength and polarisation
characteristics of vertical 10
17
eV air showers in four dierent magnetic eld
congurations: elds of 0.3 Gauss and 0.5 Gauss strength with horizontal and
70
inclined geometry. A 0.3 Gauss horizontal magnetic eld is present in the

equatorial region, whereas a 0.5 Gauss 70
inclined magnetic eld is present

in central Europe.
The change from a horizontal magnetic eld to a 70
inclined magnetic
eld introduces a number of eects. First, a small north-south asymmetry
arises. Second, the overall emission level drops only very slightly although
the projected magnetic eld that the vertical air shower sees drops by a factor
of cos
1
(70
) 3. This demonstrates the very weak eect of the magnetic eld

on the total emission eld strength. The most prominent change is visible in
the polarisation characteristics along the east-west direction from the shower
centre.
Increasing the eld strength from 0.3 Gauss to 0.5 Gauss mainly boosts
the ux in the east-west direction from the shower centre in case of a horizon-
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
F 4.12Contour plots of the 10 MHz emission from a 10
17
eV air shower with 45
zenith angle as a function of shower azimuth. Columns from left to right: total eld strength,
north-south, east-west and vertical polarisation component. Lines from top to bottom: 0
, 30
,
60
and 90
azimuth angle. Contour levels are 0.25 V m

1
MHz
1
apart.
tal magnetic eld. In case of a 70
inclined magnetic eld, the changes are

minimal.
4.4.4 Primary particle energy
Another important characteristic of the radio emission is its dependence on
the primary particle energy. In Fig. 4.16 we present the dependence of the
10 MHz frequency component at various distances from the centre of a vertical
air shower as a function of primary particle energy. Although the depth of the
shower maximum obviously depends on the primary particle energy, it is kept
constant in these calculations to assess the inuence of the primary particle
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
-1000
-500
0
500
1000
-1000 -500 0 500 1000
d
is
ta
n
c
e
N
S
[m
]
distance EW[m]
F 4.13Same as Fig. 4.12 but back-projected to the shower-based coordinate system
(distances are measured perpendicularly from the shower axis).
energy alone. (For a combined dependence see section 4.5.4).
The scaling of the eld strength with primary particle energy is approxi-
mately linear, following a power-law E
0.96
p
. For 55 MHz, the diagram is very
similar until the curves again cut o at distances of a few hundred metres due
to coherence losses (not shown here). The spectra do not change signicantly
in comparison with the 10
17
eV case except for an overall change in amplitude
as demonstrated in Fig. 4.17 in comparison with Fig. 4.1.
An approximate linear scaling of the emission with primary particle en-
ergy is to be expected for coherent emission. In the coherent regime, the eld
strength directly scales with the number of emitting particles. (The emitted
power consequently scales as the number of particles squared.) Since the num-
ber of particles grows approximately linearly with primary particle energy in
F 4.14Contour plot of the total 10 MHz electric eld component emitted by a 10
17
eV
45
inclined air shower with overplotted indicators denoting the ratio of east-west to north-south
polarisation. Left: azimuth of 0
, right: azimuth of 90
. Contour levels are 0.25 V m

1
MHz
1
apart.
the parametrisations at the basis of our simulations (see chapter 3), the linear
scaling directly translates to a linear scaling of the eld strength with primary
particle energy.
4.4.5 Depth of shower maximum
The depth of the air shower maximum is directly related to the nature and
energy of the primary particle (cf., e.g., Pryke 2001). Additionally, it is one of
the parameters that undergoes strong uctuations between individual showers
with otherwise identical parameters. It is therefore interesting to evaluate the
dependence of the emission on this parameter.
As can be seen in Fig. 4.18, there is a signicant dependence of the emis-
sion on the depth of the air shower maximum. The deeper penetrating the air
shower, the steeper becomes the radial emission pattern. This is especially im-
portant for extremely high-energy air showers
> 10
20
eV, where the shower
maximum can develop close to sea-level. The eect is the same as that visi-
ble in the zenith angle dependence (cf. section 4.4.1), where it is much more
pronounced because the (spatial) distance of the shower maximum from the
ground grows very rapidly with increasing zenith angle for a given value of
X
max
.
The eect is very similar at 55 MHz, except for the cutos due to the loss
4.5. Parametrisations for vertical showers 125
F 4.15Contour plots of the total 10 MHz electric eld emitted by a 10
17
eV vertical air
shower with overplotted indicators denoting the ratio of east-west to north-south polarisation.
Top: 0.3 Gauss magnetic eld, bottom: 0.5 Gauss magnetic eld, left: magnetic eld horizontal,
right: magnetic eld 70
inclined. Contour levels are 0.25 V m

1
MHz
1
apart.
of coherence at distances above a few hundred metres.
4.5 Parametrisations for vertical showers
Having analysed the qualitative dependences of the radio emission on various
air shower and observer parameters, constructing a parametrisation of these
dependences would be very useful. In a rst step, we therefore quantify the
dependences in a simple manner for vertical geometry. Afterwards, we gener-
alise these dependences to an arbitrary geometry and piece together an overall
parametrisation taking into account the parameters simultaneously.
126 4.5. Parametrisations for vertical showers
0.001
0.01
0.1
1
10
100
1000
10000
1e+15 1e+16 1e+17 1e+18 1e+19
Primary particle energy [eV]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.16Scaling of the 10 MHz east-west electric eld component emitted by a vertical
air shower as a function of primary particle energy E
p
. From top to bottom: 20 m, 100 m,
180 m, 300 m and 500 m from the shower centre. The data follow a power-law E
0.96
p
.
0.1
1
10
100
1000
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 4.17Spectra of the emission from a vertical 10
19
eV air shower at various distances
to the north. From top to bottom: 20 m, 140 m, 260 m, 380 m and 500 m.
0.1
1
10
0 100 200 300 400 500 600
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
E
W
(
R
,
)
|
[
V
m
1
M
H
z
1
]
F 4.18Radial dependence of the 10 MHz component from a vertical 10
17
eV air
shower with various depths of the shower maximum X
max
. Red/solid: X
max
= 560 g cm
2
,
green/dashed: X
max
= 595 g cm
2
, blue/dotted: X
max
= 631 g cm
2
, violet/short dot-
ted: X
max
= 665 g cm
2
, turquois/dash-dotted: X
max
= 700 g cm
2
, black/double-dotted:
X
max
= 735 g cm
2
.
As there is no direct error estimate for the underlying Monte Carlo re-
sults, we neither specify any
2
values nor make any error estimates for the
derived t parameters in the following sections. We specify our t parameters
with a high number of signicant digits, knowing that the parameters are not
determined with such high precision. Nevertheless, this allows an overall bet-
ter representation of the Monte Carlo results with the t functions. To verify
the quality and estimate the deviation of our parametrisation from the Monte
Carlo data, we then make a direct comparison of our overall parametrisation
and the corresponding Monte Carlo results for a sample of test parameter sets
in section 4.6.5.
4.5.1 Radial dependence
The radial dependence of the emission on distance r from the shower centre
can be t with an exponential decay,
|E(r, 2)| = E
0
exp
_
r
r
0
_
. (4.1)
Fig. 4.19 shows the simulated 10 MHz and 55 MHz total eld strength com-
ponents as a function of distance to the north from the shower centre with the
associated exponential ts. For the 10 MHz component, we t one exponential
in the central 500 m and a second in the outer 500 m region. This increases
the quality of both ts very signicantly. For the 55 MHz component, we use
only the values up to 380 m, as the emission becomes incoherent at higher
distances. We do not take any asymmetry of the emission pattern (cf. Fig. 4.3)
into account in this parametrisation.
[MHz] E
0
[V m
1
MHz
1
] r
0
[m] valid r [m]
10 12.3 135.3 0500
10 84.4 90.44 5001000
55 7.85 51.36 0380
T 4.1Parameters for the radial ts according to eq. (4.1) depicted in Fig. 4.19.
The spectral dependence in the coherent regime can also be parametrised well
with an exponential decay. The dependence in the incoherent regime at high
0.001
0.01
0.1
1
10
100
0 100 200 300 400 500 600 700 800 900 1000
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.19Exponential radial dependence of the 10 MHz emission in the central 500 m
(solid), 10 MHz emission in the outer 500 m (dotted) and 55 MHz emission (dashed) from a
10
17
eV vertical air shower with corresponding Monte Carlo simulated data.
frequencies is not well determined by the Monte Carlo simulations performed
so far. It is, however, probably atter than an exponential decay as demon-
strated from the analytical calculations, where the functional form converges
towards a power-law (cf. Fig. 2.16). Outside the valid -regime, the parametri-
sation therefore is bound to underestimate the real ux. We t the function
|E(r, 2)| = E
0
exp
_
( 10 MHz)
0
_
(4.2)
to spectra at various distances r from the shower centre to the north. The
parameter E
0
in this case directly represents the total eld strength at 10 MHz.
Only data in the coherent regime is used for the tting procedure. The data
range used is indicated in Fig. 4.20 together with the resulting t functions.
The associated t parameters are listed in Table 4.2.
4.5.3 Polarisation characteristics
In the centre region, the emission is almost purely linearly polarised in the
direction perpendicular to the magnetic eld and air shower axes (cf. Fig. 4.14),
r [m] E
0
[V m
1
MHz
1
]
0
[MHz] valid [MHz]
20 10.01 46.501 10180
140 4.693 24.382 10110
260 1.867 14.020 1080
380 0.7942 8.0787 1050
500 0.2858 5.2633 1032
T 4.2Parameters for the spectral ts according to eq. (4.2) depicted in Fig. 4.20.
0.001
0.01
0.1
1
10
10 100
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
)
|
[
V
m
1
M
H
z
1
]
[MHz]
F 4.20Fit of exponential functions to spectra from a 10
17
eV vertical air shower at
various distances to the north. From top to bottom: 20 m, 140 m, 260 m, 380 m and 500 m.
Only the data in the coherent regime is used for the t, as indicated by the colours.
in good agreement with the analytic calculations. In particular, the electric eld
vector points into the direction
E(, ,
B
) =
_
_
sin sin
B
sin
cos cos
B
cos sin sin
B
cos
B
sin sin
_
_
_
(cos cos
B
cos sin sin
B
)
2
+ sin
2
sin
2
, (4.3)
where denotes the shower zenith angle, is the shower azimuth angle with
respect to the magnetic north and
B
species the inclination angle (i.e., com-
plement of the zenith angle) of the magnetic eld.
Multiplication of an
E(r, 2, E
p
)
-value with the unit polarisation vector
E(, ,
B
) then directly yields the estimated north-south, east-west and verti-
cal linear polarisation components (in this order).
The complex dependences at higher distances from the shower centre can-
not be easily parametrised at this stage.
4.5.4 Combined E
p
and X
max
dependence
We have discussed the radio emissions dependence on the primary particle
energy and the depth of the shower maximum separately in earlier sections.
Here, we parametrise the combined dependence on primary particle energy
and appropriately adjusted depth of shower maximum reecting the deeper
atmospheric penetration of higher energy air showers. (We set X
max
to 500,
560, 631, 700 and 770 g cm
2
for E
p
values of 10
15
, 10
16
, 10
17
, 10
18
and
10
19
eV, respectively, see Pryke 2001; Knapp et al. 2003). The steepening of
the radial dependence for increasing X
max
discussed in section 4.4.5 in this case
leads to a radius-dependent steepening or attening of the energy dependence
(originally E
0.96
p
as shown in section 4.4.4) in the central and outer regions,
respectively. However, the combined dependence is still well-described by a
power-law of the type
E(r, 2, E
p
)
= E
0
_
E
p
10
17
eV
_
(r)
. (4.4)
The associated t parameters are listed in Table 4.3.
r [m] E
0
[V m
1
MHz
1
] valid E
p
[eV]
20 10.18 1.057 10
15
10
19
100 5.188 1.004 10
15
10
19
180 2.995 0.965 10
15
10
19
300 1.315 0.907 10
15
10
19
500 0.265 0.808 10
15
10
19
T 4.3Parameters for the combined primary particle energy and depth of shower maxi-
mum dependence according to eq. (4.4) depicted in Fig. 4.21.
132 4.6. Parametrisations for arbitrary geometry
0.001
0.01
0.1
1
10
100
1000
10000
1e+15 1e+16 1e+17 1e+18 1e+19
Primary particle energy [eV]
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.21Scaling of the 10 MHz electric eld emitted by a vertical air shower as a
function of primary particle energy E
p
with appropriately changing depth of shower maximum
X
max
. From top to bottom: 20 m, 100 m, 180 m, 300 m and 500 m to the north from the shower
centre.
4.6 Parametrisations for arbitrary geometry
We now generalise our parametrisations of the radio emission as a function
of air shower and observer parameters to an arbitrary shower geometry. As
our nal result, we piece the individual parametrisations together to an overall
parametrisation incorporating all major parameters.
4.6.1 Radial dependence
The emission pattern becomes increasingly asymmetric with increasing zenith
angle (cf. Fig. 4.12). As discussed in section 4.4.1, most of this asymme-
try is caused by projection eects that can be taken into account by changing
from a ground-based coordinate system (distance r from the shower centre) to
a shower-based coordinate system (perpendicular distance l from the shower
axis), cf. Fig. 4.13. The remaining intrinsic asymmetries in the emission pat-
tern we do not take into account in our parametrisation.
The back-projection from the ground-based to the shower-based coordi-
4.6. Parametrisations for arbitrary geometry 133
nate system is given by
l(r) = r
_
1 cos
2
(
o
) sin
2
(), (4.5)
when and specify the shower azimuth and zenith angle and r and
o
denote
the observer distance from the shower centre and the observer azimuth angle
(azimuth angles being measured with respect to the north). The 10 MHz back-
projected radial dependence can then be well t as
|E(l)| = E
exp
_
l
l
_
. (4.6)
The tting is performed as described in section 4.5.1 for each zenith angle
individually. We restrain ourselves to the inner 500 m (back-projected) radius
to increase the accuracy of the parametrisation. We also base the ts on air
showers coming from the north rather than the south to exclude the deviation
for the 15
zenith angle case arising from the only 5
angle to the geomagnetic

eld in case of an air shower coming from the south. The resulting ts are
depicted in Fig. 4.22, and the corresponding parameters are listed in table 4.4.
0.1
1
10
0 100 200 300 400 500
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
|
E
(
R
,
2
)
|
[
V
m
1
M
H
z
1
]
F 4.22Radial dependence to the north for the 10 MHz emission from a 10
17
eV air
shower with corresponding exponential ts. Red/solid: vertical shower, green/dashed: 15
,
blue/dotted: 30
, violet/short dotted: 45
zenith angle.
E
[V m
1
MHz
1
] l
[m] valid l [m]

0
12.33 135.30 0500

15
11.04 152.80 0500

30
8.33 202.09 0500

45
4.98 339.71 0500

60
2.53 873.54 0500

T 4.4Parameters for the radial ts according to eq. (4.6) depicted in Fig. 4.22.
Similarly as for the radial dependence, we now generalise our parametrisation
for the spectral dependence of the reference shower to an arbitrary shower ge-
ometry using the same exponential ts as adopted in section 4.5.2. Again, for
each shower zenith angle individually, we t a number of spectra at dierent
distances l from the shower axis using the function
|E(l, 2)| = E
(l) exp
_
( 10 MHz)
(l)
_
. (4.7)
We do not showthe individual ts explicitly here. The resulting parameters
E
(l) and
(l) are tabulated in table 4.5. The dependence of
on the distance
to the shower axis l can in turn be parametrised with an exponential function
(l) = a
e
l/b
. (4.8)
In fact, the parameter a
can be xed to the same value for all cases of

analysed here at only minor loss of precision. The ts with a
xed to a value
of 47.96 MHz are shown in Fig. 4.23. The associated parameters for b
are
listed in table 4.6.
4.6.3 Dependence of radial scale factor on X
max
To factor the inuence of the (vertical equivalent) depth of shower maximum
into the parametrisation, we have to parametrise the attening of the emis-
sions radial dependence with increasing X
max
. To achieve this, we t the ra-
dial dependences calculated for vertical 10
17
eV showers with various values
of X
max
(shown in Fig. 4.18) with exponential functions in the central 500 m
(not shown here). As can be seen in Fig. 4.18, the curves overlap at a distance
l [m] E
[V m
1
MHz
1
]
[MHz] valid [MHz]

0
20.0 10.01 46.501 10180

0
140.0 4.693 24.382 10110

0
260.0 1.867 14.020 1080

0
380.0 0.7942 8.0787 1050

0
500.0 0.2858 5.2633 1032

15
19.3 9.594 45.000 10180

15
135.2 4.805 26.705 10110

15
251.1 2.591 14.138 1080

15
367.1 1.183 8.2633 1050

15
483.0 0.433 5.8116 1032

30
17.3 8.185 44.532 10110

30
155.9 4.210 26.770 10110

30
259.8 2.596 17.131 1070

30
363.7 1.563 10.416 1050

30
502.3 0.582 7.0933 1032

45
14.1 6.145 40.824 10110

45
127.3 3.583 36.512 10110

45
212.1 2.653 21.487 10100

45
381.1 1.948 13.550 1060

45
495.0 1.245 8.866 1050

60
10.0 3.251 43.881 10100

60
130.0 2.285 39.652 10100

60
250.0 2.032 34.980 10100

60
370.0 1.964 25.318 1080

60
490.0 1.657 19.828 1070

T 4.5Parameters for the spectral ts of 10
17
eV air showers with arbitrary geometry
according to eq. (4.7). (The arbitrary-seeming values for l are due to calculation of the air
shower in the ground-based coordinate system with subsequent conversion to the shower-based
coordinate system.)
1
10
100
0 100 200 300 400 500 600
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
d
i
s
t
a
n
c
e
l
f
r
o
m
t
h
e
s
h
o
w
e
r
a
x
i
s
[
m
]
[
M
H
z
]
F 4.23Parametrisation of

as a function of distance to the shower axis l according to
eq. (4.8). Red/solid: vertical shower, green/dashed: 15
, blue/dotted: 30
, violet/short dotted:
45
zenith angle.
b
[m]
0
219.41
15
219.16
30
254.23
45
305.17
60
590.03
T 4.6Parameters for the parametrisation of

(l) according to eq. (4.8).
of l = r 200 m. Taking this point as a reference, the eect of changing X
max
can be reduced to a pure change of the slope, i.e., the scale factor l
of the
exponential. We can then quantify the change of this scale factor by the ratio
(X
max
) =
l
(X
max
)
l
(631 g cm
2
)
(4.9)
of the scale factor for a given X
max
and the scale factor of our reference shower.
Figure 4.24 shows as a function of X
max
and a t of this dependence using a
power-law
(X
max
) = 1.00636
_
X
max
631 g cm
2
_
1.50519
. (4.10)
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
1.25
550 600 650 700 750
P
S
f
r
a
g
r
e
p
l
a
c
e
m
e
n
t
s
X
max
[g cm
2
]
F 4.24Ratio of the scale factor for a given X
max
to the scale factor of the reference
shower as a function of X
max
and corresponding power-law t.
4.6.4 Overall parametrisation
We now piece the individual parametrisations together to an overall formula.
This implies that the dierent eects are independent of each other and there-
fore can be separated in an easy way, which need not be true in all cases.
Nevertheless, such an overall parametrisation can be a useful basis for com-
parisons of experimental data with theoretical predictions as long as one keeps
the limitations of the parametrisation in mind. We therefore provide a formula
similar to the parametrisation rst given by Allan (1971) and further enhanced
by Falcke & Gorham (2003). The associated sets of parameters are given for
each zenith angle individually.
At the heart of the parametrisation is the radial dependence for arbitrary
geometry as described in section 4.6.1. This we combine with the spectral
dependence derived in section 4.6.2 and get
|E(r,
o
, )| = E
exp
_
l(r,
o
)
l
_
exp
_
/MHz 10
47.96 exp
_
l(r,
o
)/b
_
_
(4.11)
with the parameters as listed in table 4.7. Additionally, we can factor in the
E
[V m
1
MHz
1
] l
[m] b
[m]
0
12.33 135.30 219.41

15
11.04 152.80 219.16

30
8.33 202.09 254.23

45
4.98 339.71 305.17

60
2.53 873.54 590.03

T 4.7Parameters for the overall parametrisation of the air shower emission according
to eq. (4.12).
primary particle energy dependence (cf. 4.4.4) and the dependence of the radial
scale factor on the (vertical shower equivalent) depth of the shower maximum
derived in section 4.6.3. For the latter, we have to take into account the change
of the eld strength in the shower centre associated with the steepening of the
scale factor l
calculated from the reference point at l = 200 m. The resulting

overall parametrisation is then given by
E(r,
o
, , E
p
, X
max
)
= E
_
E
p
10
17
eV
_
0.96
exp
_
200 m((X
max
) 1) + l(r,
o
)
(X
max
) l
_
exp
_
/MHz 10
47.96 exp
_
l(r,
o
)/b
_
_
(4.12)
with (X
max
) dened in eq. (4.10), l(r,
o
) given by eq. (4.5) and the values
for E
, l
and b
taken from table 4.7. To calculate the eld strength for the
individual linear polarisation components, one can then multiply the result of
eq. (4.12) with the unit polarisation vector given in eq. (4.3).
4.6.5 Quality and validity of the overall parametrisation
To verify the quality of our parametrisation, we take a sample of observer
parameters testing the dierent regimes of the parametrisation and compare
the result of eq. 4.12 with the result from the Monte Carlo simulation. As can
be seen in table 4.8, the deviations are acceptably small: most of the time the
error is below 10%, and only in cases where the parametrisation is expected to
degrade (e.g. in the east-west direction of heavily inclined air showers, where
the intrinsic asymmetries in the emission pattern become relevant) it grows
beyond 20%. This is a very satisfactory result.
E
p
X
max
r
o
|E
param
| |E
MC
| deviation
[
] [eV] [g cm
2
] [m] [
] [MHz]
_
V
m MHz
_ _
V
m MHz
_
[%]
0 10
17
631 0 0 10 12.22 14.07 -13.21
0 10
17
631 0 0 44.43 5.96 6.58 -9.51
0 10
17
631 100 0 10 5.86 5.45 7.47
0 10
17
631 420 45 10 0.56 0.59 -5.93
0 10
17
631 0 0 55 4.78 4.98 -4.1
0 10
17
560 20 0 10 8.49 7.82 8.57
0 10
17
735 60 0 55 2.99 2.55 17.01
0 10
17
735 260 0 10 1.62 1.53 5.73
0 10
18
700 20 0 10 120.28 118.13 1.81
0 10
19
631 220 45 10 172.41 199.45 -13.55
15 10
17
631 60 45 55 2.18 2.24 -2.42
30 10
17
631 100 0 55 1.45 1.59 -9.19
45 10
17
631 20 0 10 4.76 5.67 -16.01
45 10
17
631 180 0 10 3.42 3.42 0.03
60 10
17
631 300 0 10 2.13 1.99 6.89
60 10
17
631 300 45 10 1.93 2.09 -7.67
60 10
17
631 300 0 55 0.64 0.73 -12.59
60 10
17
631 300 45 55 0.47 0.66 -28.56
T 4.8Quality check of the overall parametrisation given by eq. (4.12).
One must of course be careful not to leave the parameter regimes for which
the parametrisation was created. Specically, the back-projected radial dis-
tance l was limited to 500 m in the underlying radial ts. The frequency limits
for the spectral ts can be estimated from table 4.5 as a function of l. As
explained earlier, the parametrisation is bound to severely underestimate the
ux at higher frequencies. The polarisation characteristics given by eq. (4.3)
are only valid for the central region as illustrated by Fig. 4.14. At signicant
zenith angles, the intrinsic asymmetries of the emission pattern which were not
taken into account in the parametrisation lead to a growing deviation from the
Monte Carlo results. Finally, special caution should be used when changing
the depth of shower maximum X
max
for signicantly inclined air showers, as
the underlying parametrisation of (X
max
) was derived for vertical air show-
ers only and the projection eects associated with inclined showers greatly
enhance the depth of shower-maximum eects.
4.6.6 Comparison with Allan-parametrisation
Allan (1971) provided a parametrisation of their experimental data which has
in turn been generalised by Falcke & Gorham (2003) to the form
= 13 V m
1
MHz
1
_
E
p
10
17
eV
_ _
sin cos
sin 45
cos 30
_
exp
_
r
r
0
(, )
_
_

50 MHz
_
1
, (4.13)
where denotes the angle between shower axis and magnetic eld, is the
shower zenith angle and r
0
is a scale factor of about 110 m. To compare this
with our results, we have to convert the experimentally motivated
values to
our theoretically derived E() values. For the conversion we use the relation
=
_
128
|E(R, )| 6.4 |E(R, )| (4.14)

as derived in chapter 2. The conversion results in
|E(R, )| = 2 V m
1
MHz
1
_
E
p
10
17
eV
_ _
sin cos
sin 45
cos 30
_
exp
_
r
r
0
(, )
_
_

50 MHz
_
1
. (4.15)
Our parametrisation yields a value of 3.6 V m
1
MHz
1
for the 50 MHz
emission in the centre of a 30
zenith angle 10
17
eV air shower, which is not far
4.7. Discussion 141
o the Allan-value of 2 Vm
1
MHz
1
. The radial dependence in both cases is
given by an exponential decay, and the resulting scale-factor in our parametri-
sation indeed corresponds to 110 m for the aforementioned set of parameters.
The linear scaling with primary particle energy is identical. Apart from these
similarities, there are of course some dierences: The frequency dependence
in the Allan-formula is specied as
1
. This is obviously very dierent from
the exponential decay in our parametrisation. However, the extrapolated fre-
quency dependence for the Allan-formula rests on rather sparse and uncertain
data, some of which lie in the incoherent regime not included in our parametri-
sation. The dependence on shower zenith angle is much more complex than a
simple cos() trend in our parametrisation and is therefore dicult to compare
with the Allan-formula. Furthermore, we do not predict any signicant depen-
dence of the total eld strength on the angle between shower axis and magnetic
eld (denoted in the Allan-formula). If, however, one looks only at a certain
polarisation component, this could well introduce a sin -dependence, e.g. on
the azimuth angle to the magnetic eld (cf. eq. (4.3)).
Overall, our parametrisation shows many similarities to the historic Allan-
parametrisation. The discrepancies, e.g. regarding the frequency dependence,
are signicant, but considering the sparse experimental data on which the
Allan-formula is founded, these discrepancies should not be over-interpreted.
In this context one should also remember that later experiments measured sig-
nicantly lower values for
, a discrepancy that is yet unsolved, but most

probably due to calibration issues.
On the one hand, we therefore urgently need new, reliable, well-calibrated
experimental data. On the other hand, the modelling eorts have to continue.
In particular, we will improve our simulation by basing it on a more realistic air
shower model as given by the CORSIKA code. This will automatically resolve
the two major shortcomings of our current model: an unrealistic particle pitch-
angle distribution and the assumption of a totally homogeneous air shower
development. Taking into account these eects most likely will redistribute
ux in such a way that the emission levels in the centre region become smaller,
making them more consistent with the historical data.
4.7 Discussion
Our analysis of the Monte Carlo simulation results with regard to the underly-
ing air shower parameters for the rst time establishes a number of experimen-
tally relevant features of the radio emission.
One major result is the predicted polarisation characteristics of the emis-
sion generated in the geomagnetic emission scenario. With this knowledge,
polarisation-sensitive experiments should be able to directly verify that a major
part of the radio emission indeed stems from the geomagnetic mechanism. The
overall weak intrinsic asymmetries in the emission pattern (except for those
associated to projection eects) on the other hand make experimental setups
measuring only the total eld strength or one circular polarisation component
seem less desirable.
Another important insight is provided by the eects arising in air show-
ers with high zenith angles. The intrinsic broadening of the emission pattern
associated to the increasing distance of the air shower maximum in combina-
tion with projection eects and a attening of the spectral dependence makes
highly inclined air showers an especially interesting target for detection with
radio techniques.
Other useful results of our simulations are the predicted wavefront curva-
ture (useful for beam-forming procedures in phased arrays and indeed con-
rmed by LOPES-measurements), the expected quasi-linear scaling of the
eld strengths in the coherent regime, the very weak dependence of the to-
tal eld strength on the specic geomagnetic eld geometry and strength and
the changes to the radial dependence as a function of changing depth of shower
maximum.
The successful incorporation of a signicant number of air shower and
observer parameters into a single parametrisation as achieved in this work
demonstrates that the emission is overall well-behaved and yields a well-
interpretable signal. As a solid estimate, the parametrisation can be a use-
ful tool for the interpretation of experimental data and the planning of exper-
imental setups. Additionally, there are remarkable similarities between the
parametrisation of our Monte Carlo results and the historic Allan-formula.
Our current results represent the most sophisticated simulation of radio
emission from cosmic ray air showers carried out to date. In the future, we will
switch from analytic parametrisations of air shower characteristics as the basis
of our simulations to a full-edged air shower model based on the CORSIKA
simulation code, further improving the modelling accuracy.
4.8 Conclusions
This work presents the cumulative result of our major eort at a realistic mod-
elling of the radio emission from cosmic ray air showers in the scheme of
coherent geosynchrotron radiation. For the rst time, we now have a solid un-
derstanding and a quantitative description of the important emission character-
istics and their dependences on important air shower and observer parameters.
In particular, the emission pattern, spectral dependence, polarisation charac-
teristics, primary particle energy dependence, magnetic eld dependence and
the dependence on air shower geometry and depth of shower maximum for
geosynchrotron emission are now theoretically determined. This information
is imperative for the interpretation and planning of concrete experiments. In
the near future, on the other hand, experiments such as LOPES will provide
well-calibrated, reliable data that will allow a direct comparison with our the-
oretical predictions for the rst time. A direct verication of the geomagnetic
emission mechanism will then be possible.
Having reached this important milestone in the modelling of radio emis-
sion from cosmic ray air showers, the next step will be an interfacing of our
code to CORSIKA (Heck et al. 1998), delivering a full-edged realistic Monte
Carlo simulation of radio emission from cosmic ray air showers. Additionally,
we will incorporate further possible emission mechanisms such as Askaryan-
type

Cerenkov radiation in our model.
5
Conclusions
We have performed a detailed analysis of radio emission from cosmic ray air
showers in the scheme of coherent geosynchrotron radiation. The properties
of the radiation generated by the geomagnetic emission mechanism which
is assumed to be dominant have been established with unprecedented accu-
racy. Our analysis has been performed in three steps.
First, we have carried out an analytic calculation of the radio emission
from an average vertical 10
17
eV air shower (chapter 2). The air shower itself
has been modelled using a set of analytic parametrisations, taking into account
the most important shower properties such as longitudinal and lateral particle
distributions, the particle energy distribution and a simplied integration over
the air shower evolution as a whole. The geosynchrotron emission has been
calculated in the frequency domain, based on well-known synchrotron theory.
While this approach has only limited precision, it allows to gain a solid un-
derstanding of the coherence eects arising from the dierent physical scales
present in the air shower. Our model has predicted the radial and spectral
dependences of the radio emission and identied the major eects shaping the
emission. The spectral dependence is mostly governed by the coherence eects
associated with the longitudinal particle distributions, whereas the radial emis-
sion dependence is dominated by the intrinsic beaming of the geosynchrotron
emission in combination with coherence eects arising from the lateral struc-
ture of the air shower and the integration over the shower as a whole. The
predicted emission strengths are consistent with the (sparsely) available exper-
imental data and should be high enough for LOPES to easily measure 10
17
eV
air showers.
145
146
After having gained a solid understanding of the important eects shap-
ing the emission, we have continued our modelling with the conception and
implementation of a sophisticated Monte Carlo simulation (chapter 3). To re-
tain comparability with the analytic results, our Monte Carlo model has been
based on the same analytically parametrised air shower model as used in the
analytic calculations. The Monte Carlo technique, however, allows a much
more precise integration of the radio emission from the totality of the particles
in the air shower. In particular, no far-eld approximations had to be applied,
the full polarisation characteristics of the emission are retained, the full 2-d
emission pattern on the ground can be calculated and the integration over the
air shower evolution as a whole has been incorporated in a sophisticated way.
Additionally, track length eects that could not be taken into account in the an-
alytical calculations, have been treated with great care. Their inclusion has led
to the important insight that the total eld strength emission pattern is indeed
remarkably symmetrical in spite of the asymmetry introduced by the geomag-
netic eld. It is therefore necessary to measure polarisation data in order to
verify the geomagnetic origin of the emission. Apart from having demon-
strated and thoroughly tested the concepts envisaged to allow the calculation
of the radio emission with high precision on standard PC hardware, we have
compared the Monte Carlo results directly with the earlier analytical calcula-
tions of a vertical 10
17
eV air shower. Overall, the agreement is very good,
with a mild systematic deviation originating from the treatment of trajecto-
ries as always symmetric in the analytical model. In particular, the emission
strengths are still at levels that should alloweasy detection with LOPES. As the
time-domain Monte Carlo simulation pursues a very dierent approach than
the frequency-domain analytical calculations, the good agreement between the
two methods gives us solid condence in our calculations.
Having demonstrated the correctness and robustness of our Monte Carlo
code, we have then performed extensive simulations of a huge set of air show-
ers in order to establish the radio emissions dependence on the air shower
geometry and specic air shower parameters such as the primary particle en-
ergy, the depth of the air shower maximum and the magnetic eld congura-
tion (chapter 4). Major results are the predicted polarisation characteristics,
the intrinsic broadening of the emission pattern as a function of zenith angle
due to the increasing distance of the shower maximum and the conrmation
of a quasi-linear scaling of the emission strength with primary particle en-
ergy. Apart from a qualitative discussion of these and other important eects,
we have successfully incorporated the analysed dependences into an overall
147
parametrisation. This easy-to-use formula will facilitate the comparison of ex-
perimental data and theoretically predicted emission properties signicantly.
Furthermore, it allows for the rst time to relate important air shower parame-
ters directly to specic characteristics of the radio emission. The parametrisa-
tion summarises the cumulative modelling eort performed in this thesis and
thus represents a very useful tool for the study of radio emission from cosmic
ray air showers.
In the course of this thesis, we have built a solid theoretical foundation
for the interpretation of current and future experimental data of radio emission
from cosmic ray air showers. The next interesting step will be to directly com-
pare the theoretical predictions with experimental data, especially regarding
the polarisation characteristics, which allow a direct verication of the geo-
magnetic emission mechanism. Additionally, our model constitutes a solid
basis for further development. In particular, we will substitute the air shower
model based on analytic parametrisations with an interfacing to the air shower
simulation code CORSIKA, yielding a full-edged Monte Carlo simulation of
radio emission from cosmic ray air showers. Furthermore, we will include
additional emission mechanisms such as Askaryan-type

Cerenkov radiation in
our model.
With these advances in the theoretical modelling of the emission and up-
coming results from experimental projects such as LOPES, radio emission
from cosmic ray air showers thus has made great advances on the way to-
wards being established as an additional observing technique for cosmic ray
research.
148
List of Figures
1.1 Schematic view of an extensive air shower . . . . . . . . . . 5
1.2 Spectrum of the cosmic ray ux . . . . . . . . . . . . . . . . 6
1.3 Chemical composition of cosmic rays . . . . . . . . . . . . . 7
1.4 HiRes versus AGASA data . . . . . . . . . . . . . . . . . . . 7
1.5 UHECR anisotropy . . . . . . . . . . . . . . . . . . . . . . . 8
2.1 Geometry of single-particle synchrotron radiation . . . . . . . 21
2.2 Coherence of a particle pair . . . . . . . . . . . . . . . . . . 23
2.3 Comparison of synchrotron radiation components . . . . . . . 25
2.4 Arrival time distribution of shower particles . . . . . . . . . . 32
2.5 Radial dependence of particle arrival time distribution . . . . 33
2.6 Coherence eects in a linecharge . . . . . . . . . . . . . . . 39
2.7 Radial dependence of emission from a point source . . . . . . 40
2.8 Geometry of the air shower maximum . . . . . . . . . . . . . 41
2.9 Coherence eects due to lateral distribution . . . . . . . . . . 44
2.10 Coherence eects in radial dependence due to lateral distribution 44
2.11 Radial dependence for reduced to the north . . . . . . . . 45
2.12 Radial dependence for reduced to the east . . . . . . . . . 46
2.13 Coherence eects in a aring disk . . . . . . . . . . . . . . . 47
2.14 Asymmetry in radial dependence . . . . . . . . . . . . . . . 47
2.15 Radio pulses emitted by a aring disk . . . . . . . . . . . . . 48
2.16 Spectra of an integrated air shower . . . . . . . . . . . . . . . 49
2.17 Radial dependence of emission from an integrated air shower 50
3.1 Ground-trace of a particle trajectory . . . . . . . . . . . . . . 65
3.2 Smart trajectory sampling . . . . . . . . . . . . . . . . . . . 67
3.3 Economic gridding algorithm . . . . . . . . . . . . . . . . . 68
149
150 LIST OF FIGURES
3.4 Raw pulses emitted by the shower maximum . . . . . . . . . 70
3.5 Spectra of the emission from the shower maximum . . . . . . 71
3.6 Filtering of the raw pulses . . . . . . . . . . . . . . . . . . . 72
3.7 Analytics vs. Monte Carlo for a point source . . . . . . . . . 74
3.8 Point source pulses to the north . . . . . . . . . . . . . . . . 75
3.9 Point source pulses to the east . . . . . . . . . . . . . . . . . 75
3.10 Point source pulses for dierent magnetic eld strengths . . . 76
3.11 Robustness of the gridding algorithms . . . . . . . . . . . . . 78
3.12 Robustness of the smart trajectory sampling . . . . . . . . . . 79
3.13 Robustness of the
1
cone cutting . . . . . . . . . . . . . . . 79
3.14 Robustness of automatic bin inactivation . . . . . . . . . . . 80
3.15 Automatic bin inactivation sequence . . . . . . . . . . . . . . 81
3.16 Radial emission dependence for long particle trajectories . . . 83
3.17 Radial emission dependence for short particle trajectories . . 83
3.18 Polarity change in the raw pulses . . . . . . . . . . . . . . . 84
3.19 Radial emission dependence for statistical trajectory lengths . 85
3.20 Emission dependence on magnetic eld conguration . . . . 86
3.21 Contour plots of emission from the shower maximum . . . . . 86
3.22 Automatic bin inactivation sequence for an inclined magnetic
eld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.23 Inuence of the particle energy distribution . . . . . . . . . . 88
3.24 Surface plot of the emission from the shower maximum . . . 89
3.25 Spectra compared between analytics and MC simulations . . . 90
3.26 Spectra comparison incorporating systematic deviations . . . 91
3.27 Radial dependence compared between analytics and MC sim-
ulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.28 Spectra comparison including particle energy distribution . . 92
3.29 Radial comparison including particle energy distribution . . . 93
3.30 Particle injection function . . . . . . . . . . . . . . . . . . . 96
3.31 Trace of the particle trajectories . . . . . . . . . . . . . . . . 96
3.32 Eects of the shower integration . . . . . . . . . . . . . . . . 97
3.33 Pulses from an integrated shower . . . . . . . . . . . . . . . 98
3.34 Contour plots for an integrated shower . . . . . . . . . . . . . 98
3.35 Radial dependence in polarisation components . . . . . . . . 99
3.36 Pulse amplitude vs. particle density . . . . . . . . . . . . . . 99
3.37 Spectra compared between analytics and MC for an integrated
shower . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
3.38 Spectra compared incorporating systematics and data . . . . . 101
LIST OF FIGURES 151
3.39 Radial dependence compared between analytics and MC for
an integrated shower . . . . . . . . . . . . . . . . . . . . . . 102
4.1 Spectra from a vertical air shower . . . . . . . . . . . . . . . 112
4.2 S
spectra from a vertical air shower . . . . . . . . . . . . . 112

4.3 Contour plots of a 10
17
eV vertical air shower . . . . . . . . . 113
4.4 Radial dependence of the emission . . . . . . . . . . . . . . . 114
4.5 Curvature of the radio wavefront . . . . . . . . . . . . . . . . 115
4.6 Polarisation: raw pulses . . . . . . . . . . . . . . . . . . . . 116
4.7 Polarisation: scatter plot . . . . . . . . . . . . . . . . . . . . 117
4.8 Dependence on shower zenith angle at 10 MHz . . . . . . . . 118
4.9 Dependence on shower zenith angle at 10 MHz (backprojected) 119
4.10 Dependence on shower zenith angle at 55 MHz . . . . . . . . 119
4.11 Spectra for a 45
inclined air shower . . . . . . . . . . . . . . 120

4.12 Contour plots for a 45
zenith angle shower . . . . . . . . . . 122

4.13 Contour plots for a 45
zenith angle shower (backprojected) . 123

4.14 Polarisation characteristics . . . . . . . . . . . . . . . . . . . 124
4.15 Magnetic eld dependence . . . . . . . . . . . . . . . . . . . 125
4.16 Scaling with primary particle energy . . . . . . . . . . . . . . 126
4.17 Spectra from a vertical air shower . . . . . . . . . . . . . . . 126
4.18 Inuence of depth of shower maximum . . . . . . . . . . . . 127
4.19 Exponential ts to the radial dependence . . . . . . . . . . . 129
4.20 Fit of spectra from a vertical air shower . . . . . . . . . . . . 130
4.21 Scaling with primary particle energy . . . . . . . . . . . . . . 132
4.22 Exponential ts to the radial dependences . . . . . . . . . . . 133
4.23 Parametrisation of spectral dependence parameters . . . . . . 136
4.24 Scaling of with X
max
. . . . . . . . . . . . . . . . . . . . . 137
152 LIST OF FIGURES
List of Tables
3.1 Parameters for the atmospheric layers . . . . . . . . . . . . . 63
4.1 Radial dependence t parameters (vertical shower) . . . . . . 128
4.2 Spectral dependence t parameters (vertical shower) . . . . . 130
4.3 Energy dependence t parameters . . . . . . . . . . . . . . . 131
4.4 Radial dependence t parameters . . . . . . . . . . . . . . . 134
4.5 Spectral dependence t parameters . . . . . . . . . . . . . . 135
4.6 Spectral scale factor t parameters . . . . . . . . . . . . . . . 136
4.7 Overall parametrisation t parameters . . . . . . . . . . . . . 138
4.8 Quality check of the overall parametrisation . . . . . . . . . . 139
153
154 LIST OF TABLES
Bibliography
Abu-Zayyad, T., Belov, K., Bird, D. J., et al. 2001, Astropart. Physics, 16, 1
Agnetta, G., Ambrosio, M., Aramo, C., et al. 1997, Astropart. Physics, 6, 301
Aharonian, F. A., Akhperjanian, A. G., Aye, K.-M., et al. 2004, Nature, 432,
75
Alkhofer, O. C. 1975, Introduction to cosmic radiation (Verlag Carl Thiemig
M unchen)
Allan, H. R. 1971, Prog. in Element. part. and Cos. Ray Phys., Vol. 10, 171
Allan, H. R., Clay, R. W., & Jones, J. K. 1970, Nature, 227, 1116
Allan, H. R., Clay, R. W., Jones, J. K., Abrosimov, A. T., & Neat, K. P. 1969,
Nature, 222, 635
Allan, H. R., Jones, J. K., Mandolesi, N., Prah, J. H., & Shutie, P. F. 1971, in
Proc. of the Twelfth Intern. Cosmic Ray Conf., Hobart, Vol. 3, 1102
Allan, H. R., Neat, K. P., & Jones, J. K. 1967, Nature, 215, 267
Allan, H. R., Shutie, P. F., Sun, M. P., & Jones, J. K. 1973, in Proc. Thirteenth
Intern. Cosmic Ray Conf., Denver, Vol. 4, 2407
Allan, H. R., Sun, M. P., & Jones, J. K. 1975, in Proc. Fourteenth Intern.
Cosmic Ray Conf., M unchen, Vol. 8, 3082
Aloisio, R. & Blasi, P. 2002, Astropart. Physics, 18, 183, astro-ph/0201310
Alvarez-Mu niz, J., V azquez, R. A., & Zas, E. 2000, Phys. Rev. D, 62, 63001
155
156 Bibliography
ANTARES collaboration. 1999, A Deep Sea Telescope for High Energy Neu-
trinos, Tech. rep., astro-ph/9907432
Antoni, T., Apel, W. D., Badea, F., et al. 2003, Nucl. Instr. Meth., 513, 490
Antoni, T., Apel, W. D., Badea, F., et al. 2001, Astropart. Physics, 14, 245
Askaryan, G. A. 1962, Soviet Phys. JETP, 14, 441
Askaryan, G. A. 1965, Soviet Phys. JETP, 21, 658
Atrashkevich, V. B., Vedeneev, O. V., Allan, H. R., et al. 1978, Sov. J. Nucl.
Phys., 28, 366
Atrashkevich, V. B., Vedeneev, O. V., & Khristiansen, G. B. 1975, in Proc.
Fourteenth Intern. Cosmic Ray Conf., M unchen, Vol. 8, 3086
Auger, P., Ehrenfest, P., Maze, R., Daudin, J., & Fr eon, R. A. 1939, Reviews
of Modern Physics, 11, 288
Balkanov, V. A., Belolaptikov, I. A., Bezrukov, L. B., et al. 1999, in Dark
matter in Astrophysics and Particle Physics, 805+
Belletoile, A., Ardouin, D., Charrier, D., et al. 2004, in Proc. of the SF2A
2004, astro-ph/0409039
Bergman, D.R. for the High Resolution Flys Eye Collaboration. 2003, in
Proceedings of the 28th ICRC, Tsukuba, Japan, ed. T. Kajita, Y. Asaoka,
A. Kawachi, Y. Matsubara, & M. Sasaki, Vol. 1 (Universal Academy Press,
Inc. - Tokyo, Japan), 397
Biermann, P. L. & Strittmatter, P. A. 1987, ApJ, 322, 643
Bird, D. J., Corbato, S. C., Dai, H. Y., et al. 1995, ApJ, 441, 144
Buniy, R. V. & Ralston, J. P. 2002, Phys. Rev. D, 65, 16003
Bury, K. V. 1975, Statistical Models in Applied Science (John Wiley & Sons,
New York), 299
Castagnoli, C., Silvestro, G., Picchi, P., & Verri, G. 1969, Nuovo Cimento,
63B, 373
Bibliography 157
Catalano, O., Maccarone, M. C., Santangelo, A., Scarsi, L., & The Euso Col-
laboration. 2003, in Astronomy, Cosmology and Fundamental Physics, 427
+
Cheng, K. S., Ho, C., & Ruderman, M. 1986, ApJ, 300, 500
Colgate, S. A. 1967, J. Geophys. Res., 72, 4869
Dova, M. T., Epele, L. N., & Mariazzi, A. G. 2003, Astropart. Physics, 18, 351
Dova, M. T., Fanchiotti, H., Garcia-Canal, C., et al. 1999, in Proceedings of
the 26th ICRC, Utah, USA, ed. D. Kieda, M. Salamon, & B. Dingus, Vol. 1,
514
Elbert, J. W. & Sommers, P. 1995, ApJ, 441, 151
Erlykin, A. & Wolfendale, A. 1987, J. Phys. G: Nuclear and Particle Physics,
23, 979
Falcke, H. & Gorham, P. W. 2003, Astropart. Physics, 19, 477
Fermi, E. 1949, Physical Review, 75, 1169
Fog, A. 2003, http://www.agner.org/random/randomc.htm
Fuji, M. & Nishimura, J. 1969, in Proc. 11th Internat. Conf. on Cosmic Rays,
Budapest, Vol. 3, 709
Gaisser, T. K. 1990, Cosmic Rays and Particle Physics (University Press, Cam-
bridge)
Gousset, T., Ravel, O., & Roy, C. 2004, in , astro-ph/0402426
Green, K., Rosner, J. L., Suprun, D. A., & Wilkerson, J. F. 2003, Nucl. Instr.
Meth., 498, 256, astro-ph/0205046
Greisen, K. 1956, Prog. Cosmic Ray Phys., Vol. 3, 1
Greisen, K. 1960, Ann. Rev. Nucl. Sci., Vol. 10, 63
Greisen, K. 1966, Phys. Rev. Lett., 16, 748
Halzen, F. 1999, New Astron. Rev., 42, 289
Harding, A. K. & Gaisser, T. K. 1990, ApJ, 358, 561
158 Bibliography
Heck, D., Knapp, J., Capdevielle, J. N., Schatz, G., & Thouw, T. 1998,
Forschungszentrum Karlsruhe Report FZKA 6019
Hess, V. F. 1912, Z. Phys., 13, 1084
Horneer, A., Antoni, T., Apel, W. D., et al. 2004, in Astronomical Telescopes
and Instrumentation 2004: Gravitational Wave and Particle Astrophysics
Detectors, Proceedings of the SPIE, Vol. 5500, 129
Horneer, A., Falcke, H., Haungs, A., et al. 2003, in Proceedings of the 28th
ICRC, Tsukuba, Japan, ed. T. Kajita, Y. Asaoka, A. Kawachi, Y. Matsubara,
& M. Sasaki, Vol. 2 (Universal Academy Press, Inc. - Tokyo, Japan), 969
Huege, T. & Falcke, H. 2002, in Proceedings of the 6th European VLBI Net-
work Symposium, ed. P. R. Z. J. Ros, E., MPIfR Bonn, Germany, 25
Huege, T. & Falcke, H. 2003, A&A, 412, 19
Huege, T. & Falcke, H. 2004a, A&A, in press
Huege, T. & Falcke, H. 2004b, Astropart. Phys., in preparation
Jackson, J. D. 1975, Classical Electrodynamics, 2nd edn. (John Wiley & Sons,
New York)
Jelley, J. V., Fruin, J. H., Porter, N. A., et al. 1965, Nature, 205, 327
Kahn, F. D. & Lerche, I. 1966, in Proc. Roy. Soc., Vol. A-289, 206
Kamata, K. & Nishimura, J. 1958, Prog. Theoret. Phys. Suppl., Vol. 6, 93
Kampert, K.-H., Antoni, T., Apel, W. D., et al. 2004, in Proceedings of the
Cosmic Ray International Seminar 2004: GZKand Surroundings, to be pub-
lished, astro-ph/0410559
Knapp, J., Heck, D., Sciutto, S. J., Dova, M. T., & Risse, M. 2003, Astropart.
Physics, 19, 77, astro-ph/0206414
Kravchenko, I., Frichter, G. M., Seckel, D., et al. 2003, Astroparticle Physics,
19, 15
Kulikov, G. V. & Khristiansen, G. B. 1959, Soviet Phys. JETP, 35, 441
Lagage, P. O. & Cesarsky, C. J. 1983, A&A, 118, 223
Bibliography 159
Linsley, J. 1986, J. Phys. G: Nucl. Phys., 12, 51
Maier, G. 2003, PhD thesis, Universit at Karlsruhe
Matsumoto, M. & Nishimura, T. 1998, ACM Trans. on Modeling and Com-
puter Simulation, Vol. 8, No. 1, 3
Matthews, J. & Pierre Auger Observatory Collaboration. 2004, AAS/High En-
ergy Astrophysics Division, 8
Nerling, F., Engel, R., Guerard, C., Perrone, L., & Risse, M. 2003, in Pro-
ceedings of the 28th ICRC, Tsukuba, Japan, ed. T. Kajita, Y. Asaoka,
A. Kawachi, Y. Matsubara, & M. Sasaki, Vol. 2 (Universal Academy Press,
Inc. - Tokyo, Japan), 611
Nikolsky, S. I. 1995, Nucl. Phys. B. (Proc. Suppl.), 39, 228
Prah, J. H. 1971, Masters thesis, University of London
Pryke, C. L. 2001, Astropart. Physics, 14, 319
Ptuskin, V. S., Rogovaya, S. I., Zirakashvili, V. N., et al. 1993, A&A, 268, 726
Rachen, J. P. & Biermann, P. L. 1993, A&A, 272, 161
Rohlfs, K. & Wilson, T. L. 1996, Tools of Radio Astronomy, 2nd edn.
(Springer Verlag, Berlin)
R ottgering, H., de Bruyn, A. G., Fender, R. P., et al. 2003, in Texas in Tuscany.
XXI Symposium on Relativistic Astrophysics, 6976
Rybicki, G. B. & Lightman, A. P. 1979, Radiative Processes in Astrophysics
(John Wiley & Sons, New York), 167194
Saltzberg, D., Gorham, P., Walz, D., et al. 2001, Physical Review Letters, 86,
2802
Sciutto, S. J. 1999, astro-ph/9911331, see also
http://www.sica.unlp.edu.ar/auger/aires
Shaviv, N. J. 2004, in Proceedings of the 19th European Cosmic Ray Sympo-
sium, in preparation
Sigl, G. 2003, Annals Phys., 303, 117
160 Bibliography
Sigl, G., Schramm, D. N., & Bhattacharjee, P. 1994, Astroparticle Physics, 2,
401
Sokolsky, P. 2003, in Particle Astrophysics Instrumentation. Edited by Peter
W. Gorham. Proceedings of the SPIE, Volume 4858, pp. 103-114 (2003).,
103114
Spencer, R. E. 1969, Nature, 222, 460
Sun, M. P. 1975, PhD thesis, University of London
Suprun, D. A., Gorham, P. W., & Rosner, J. L. 2003, Astropart. Physics, 20,
157
Takeda, M., Hayashida, N., Honda, K., et al. 1999, ApJ, 522, 225
The AGASA collaboration. 2004, http://www-akeno.icrr.u-
tokyo.ac.jp/AGASA
The Pierre Auger Collaboration. 1996, The Pierre Auger Project Design Re-
port, http://www.auger.org
Ulrich, H. 1997, Diploma thesis, Universit at Karlsruhe
Ulrich, H. 2003, PhD thesis, Universit at Karlsruhe
Watson, G. N. 1944, Theory of Bessel Functions (University Press, Cam-
bridge)
Wefel, J. P. 1991, in NATO ASIC Proc. 337: Cosmic Rays, Supernovae and
the Interstellar Medium, 29+
Zas, E., Halzen, F., & Stanev, T. 1992, Phys. Rev. D, 45, 362
Zatsepin, G. & Kuzmin, V. 1966, Zh. Eksp. Teor. Fiz., 4, 114
6
Danksagung
Eine Doktorarbeit ohne die Hilfe und Unterst utzung Anderer zu schreiben,
d urfte geradezu unm oglich sein. Auch an meiner Dissertation haben viele
Menschen Anteil gehabt.
Besonderer Dank geb uhrt meinem Betreuer Heino Falcke, der mein Inter-
esse f ur die Forschung am MPIfR bereits vor Beginn meines Master-Studiums
in Cambridge weckte. Er gab mir mit diesem spannenden Promotionsthema
eine einmalige Chance. Ohne die vielen Diskussionen, f ur die er sich immer
Zeit nahm, ohne seine Ratschl age und die Motivation, die seine Begeisterung
immer wieder in mir weckte, h atte ich diese Dissertation so nicht fertigstellen
k onnen. Meinem Koreferenten Peter Schneider m ochte ich f ur seine Hilfe und
seine vielen Ideen, die mir oft neue L osungsans atze aufzeigten, ebenfalls sehr
herzlich danken.
Trotz seiner vielf altigen Verpichtungen als gesch aftsf uhrender Direktor
des MPIfR verfolgte und begleitete Anton Zensus meine Arbeit und meine
Zukunftsplanung immer mit groem Interesse. Dies habe ich immer sehr zu
sch atzen gewusst. Auch Arno Witzel hatte stets ein oenes Ohr f ur meine
Belange und half mir beim Umschien diverser Klippen. Die Unterst utzung
durch die International Max-Planck Research School (IMPRS) gab mir die
M oglichkeit des Besuchs von Konferenzen und Sommerschulen, die mir einen
Blick uber den Tellerrand erlaubten und mir erm oglichten, wichtige Kon-
takte zu kn upfen.
Die freundliche Atmosph are in der VLBI-Gruppe trug entscheidend dazu
bei, da mir meine Arbeit trotz der obligatorischen Tiefs immer Freude
machte. Von den Mitarbeitern des MPIfR m ochte ich besonders Thomas
161
162
Beckert hervorheben, der mir mageblich bei der Einarbeitung in die Theo-
rie der Synchrotronstrahlung half und auch bei sp ateren Problemen immer ein
wertvoller Ansprechpartner war. Auch Peter Biermann verfolgte meine Arbeit
mit besonderem Interesse und machte dar uber hinaus unerm udlich Werbung
f ur meine Ergebnisse. Daneben gab es nat urlich viele Menschen ausserhalb
des MPIfR, die mir in Diskussionen, mit Ratschl agen oder sonstiger fachlicher
Unterst utzung geholfen haben. Besonders hervorheben m ochte ich hier John
Ralston, Ralph Engel und Andreas Haungs.
Ohne die vielen Studenten der VLBI-Gruppe w are die t agliche Arbeit
sicher langweilig gewesen. Ich hatte mit Andreas Brunthaler, Elmar K ording
und Andreas Horneer uber die Jahre viel Spa in unserem B uro. Dar uber
hinaus haben mir alle drei oft beim L osen kleinerer und gr osserer Probleme
sehr geholfen. Ganz besonders m ochte ich Elmar daf ur danken, da er im-
mer dazu bereit war, mit mir uber kniige Probleme zu diskutieren das hat
mich wirklich oft entscheidend weitergebracht. Das t agliche Mitagessen mit
Uwe Bach und Enno Middelberg, bei dem wir zusammen die kulinarischen
K ostlichkeiten unserer damaligen Kantine genossen, war ebenfalls immer ein
Highlight.
Martin Stumpe werde ich zum Dank dafuer, da er meine Arbeit Korrektur
gelesen hat und auch sonst immer ein wirklich guter Freund war, bei einem
Computerspiel seiner Wahl einmalig gewinnen lassen. Meinen Eltern und
meinen Br udern danke ich f ur ihre bedingungslose Unterst utzung und daf ur,
da sie immer f ur mich da waren. Ganz besonders bedanken m ochte ich mich
bei Kathrin, die mein Leben und mein Studium uber so viele Jahre bereichert
und begleitet hat.
Zu guter Letzt danke ich meiner Freundin Tatjana, die f ur mich ein ganz
besonderer Mensch ist und mit ihrer unb andigen Lebensfreude auch mein
Leben mit so viel Freude und Zuversicht erf ullt.

164
Publications
The contents of this thesis have been published in the following refereed jour-
nal articles:
Chapter 2: Radio emission from cosmic ray air showers: Coherent
geosynchrotron radiation
Huege, T., Falcke, H., 2003, A&A 412, 1934
Chapter 3: Radio emission from cosmic ray air showers: Monte Carlo
simulations
Huege, T., Falcke, H., 2004, A&A in press, astro-ph/0904223
Chapter 4: Radio emission from cosmic ray air showers: Simulation
results
Huege, T., Falcke, H., Astropart. Physics, in preparation
165

Radio Emission From Cosmic Ray Air Showers

Uploaded by

Copyright:

Available Formats

Radio Emission From Cosmic Ray Air Showers

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Radio Emission From Cosmic Ray Air Showers

Uploaded by

Copyright:

Available Formats

Radio Emission from Cosmic Ray Air Showers

. Shaded circles indicate event clustering within 2.5

< 100 Mpc (Sigl et al.

. The particle trajectory lies in the x-y plane.

dened in Figure 2.1. Retaining the phase

contributions from the two particles add up to 2A

contributions are only partial. Furthermore, as the

for an eective electron-positron

< with the observers line of sight.) In a typical

| (short-dashed) for = 100 MHz, = 60 and

|, especially for small where most of the radiation

is not dominating the

, we analytically reconstruct the time-dependence of the electromagnetic

|E(R, )| 6.4 |E(R, )| . (2.11)

. They, however, still trace the distribution of e

by far exceeds the number of

= (5.307 0.032) ns (2.29)

< 100 m that we

(R, ) by the original

only gives signicant

> 50 MHz) and therefore are an important limiting

(0, 0), which is justied due to the minute changes in

(, ) over the shower surface. The general polarisation di-

> 80 % of the particles are included in this

, and we manually scale them

inclined magnetic eld present in central Europe.

inclined 0.5 Gauss magnetic eld (thick lines).

in chapter 2) indeed works cor-

inclined magnetic eld, a broken power-law particle energy distribution and a

inclined 0.5 Gauss

inclined 0.5 Gauss magnetic eld, thin lines:

and their conversion to the theoretical

|E(R, )| 6.4 |E(R, )| . (3.17)

inclined 0.5 Gauss magnetic eld. Thin

diagram, illustrating that most of the power is emitted at frequencies

zenith angle curve from the trend seen in

, however, the coherence be-

zenith angle as a function of azimuth angle. The total eld

inclined geometry. A 0.3 Gauss horizontal magnetic eld is present in the

inclined magnetic eld is present

) 3. This demonstrates the very weak eect of the magnetic eld

azimuth angle. Contour levels are 0.25 V m

inclined magnetic eld, the changes are

. Contour levels are 0.25 V m

inclined. Contour levels are 0.25 V m

-value with the unit polarisation vector

zenith angle case arising from the only 5

angle to the geomagnetic

[m] valid l [m]

12.33 135.30 0500

11.04 152.80 0500

8.33 202.09 0500

4.98 339.71 0500

2.53 873.54 0500

(l) are tabulated in table 4.5. The dependence of

can be xed to the same value for all cases of

[MHz] valid [MHz]

20.0 10.01 46.501 10180