Course Notes for Ocean Colour Remote Sensing Course

Erdemli, Turkey
September 11 - 22, 2000
Module 1: Satellite Orbits
prepared
by
Assoc Professor Mervyn J Lynch
Remote Sensing and Satellite Research Group
School of Applied Science
Curtin University of Technology
PO Box U1987
Perth Western Australia 6845
AUSTRALIA
tel +618-9266-7540
fax +618-9266-2377
email <[email protected]>
Module 1: Satellite Orbits
1.0 Artificial Earth Orbiting Satellites
The early research on orbital mechanics arose through the efforts of people such as Tyco
Brahe, Copernicus, Kepler and Galileo who were clearly concerned with some of the
fundamental questions about the motions of celestial objects. Their efforts led to the
establishment by Keppler of the three laws of planetary motion and these, in turn, prepared
the foundation for the work of Isaac Newton who formulated the Universal Law of
Gravitation in 1666: namely, that
F = GmM/r
2
, (1)
Where F = attractive force (N),
r = distance separating the two masses (m),
M = a mass (kg),
m = a second mass (kg),
G = gravitational constant.
It was in the very next year, namely 1667, that Newton raised the possibility of artificial
Earth orbiting satellites. A further 300 years lapsed until 1957 when the USSR achieved the
first launch into earth orbit of an artificial satellite - Sputnik - occurred.
Returning to Newton's equation (1), it would predict correctly (relativity aside) the motion of
an artificial Earth satellite if the Earth was a perfect sphere of uniform density, there was no
atmosphere or ocean or other external perturbing forces. However, in practice the situation is
more complicated and prediction is a less precise science because not all the effects of
relevance are accurately known or predictable. Some of these complicating perturbing forces
arise from the non-spherical (oblate) shape of the earth, ocean tides, atmospheric drag, impact
of the Sun (solar wind) on the atmosphere and the gravitational attraction of other planetary
objects. This leads to the use of the term osculating orbit to describe the motion of the
satellite. This expression and the associated osculating orbital elements used to define the
motion will be discussed in further detail subsequently.
We could decide to describe the position of a satellite at any given time by its instantaneous
position vector (3 components) and instantaneous velocity vector (3 components). This
description would be quite adequate, if generated as a function of time. However, it is
normal practice to compute the orbital descriptors asa function of time by executing a
computer-based orbital model using inputs that are more appropriate to the precise
formulation of the numerical model.
Because of the perturbing forces mentioned above, a simple formulation such as equation 1 is
not adequate and we resort to a perturbation model to achieve the higher degree of prediction
accuracy required. One such formulation, that conveniently has an analytical solution, is the
so-called Brouwer-Lyddane model developed from the work of both Brouwer (1959) and
Lyddane (1963). While we will discuss this model further, it should be noted that there are a
number of other models that are available for this purpose.
Why do we require accurate orbital prediction models? The prime requirement is that in
order to receive direct broadcast data from a satellite as it passes overhead, we need
information on the time at which the satellite will rise over the horizon as well as the azimuth
and zenith angles so we can direct a receiving antenna (a satellite dish) at that location. The
path the satellite takes through the sky (the ephemeris), if known from the prediction model,
enables the ground receiving antenna to track the satellite and therefore receive data
continuously until the satellite sets at the completion of the overpass. The particular type of
orbit into which a satellite is deployed is predetermined by what we require the satellite to
observe. For example, if we wish a satellite to observe a particular region of the earth every
day at approximately the same local time each day throughout the annual cycle, this would
require a special orbit with a particular orbital plane precession. If we wished to observe
events that evolve relatively rapidly in time (eg typhoons, tornadic storm cells, tropical
cyclones), we would require a satellite that has the capacity to stare at a particular point on
the Earth and record imagery of that region perhaps every hour or every 15 minutes. If we
require observations of the whole Earth, at least once per day and every day (eg for observing
synoptic weather systems, or changes in sea surface temperature), again we would require a
specific orbit. However, if the requirement was for high spatial resolution (a so-called small
ground footprint or small field of view (FOV)) we would need to compromise on daily
coverage and adopt an orbit that images the whole of the Earths surface perhaps only once
every 15 days.
Other application require that we know satellite orbits to extremely high precision. For
example, in geodesy, global position systems (GPS) have revolutionised our ability to obtain
our position on the Earth's surface. This is achieved however only if we know the position of
constellation of associated satellites to very high accuracy (cm or mm accuracy). Similarly,
satellite altimetry, which measures the height of the ocean surface relative to a hypothetical
surface, the geoid, requires ideally millimeter accuracy in the knowledge of the distance from
the satellite's orbit to the ocean surface. Such applications maintain the pressure to
understand better the nature of the perturbing forces on orbits, to improve the accuracy of
orbital models and to enhance our knowledge of how to improve ephemeris prediction with
these models and supporting measurements.
2.0 Terminology and Coordinate Systems
2.1 Coordinate Systems
There are two coordinate systems that are usually encountered in remote sensing. These are
geocentric and geoedetic systems. The prime difference in these systems is where the origin
of the coordinate system is located. We will be using the geocentric system that has the
centre of mass of the Earth C as the origin of coordinates. Next we must construct the
coordinate system about this origin. The first axis is from the origin through the North Pole.
The second axis is from the origin through the Earths equatorial plane in the direction of
Aries (see below). The final axis is that which is perpendicular to the two previously defined
axes.
Aries is a point in space the direction of which is determined by the intersection of the
equatorial plane of the Earth (ie the plane that contains the equator) and the orbital plane
defined by the Earths annual motion around the Sun ( see Figure 1.1) such that a line from
the Suns centre to the Earth along this direction points toward Aries at the time of the Vernal
Equinox. This latter event occurs at a date near March 20 each year when the Suns apparent
position, as it moves from South to North in the sky, crosses the equatorial plane.
Figure 1.1 Shown are two orientations of the Earth with respect to the Sun for the months of
January and July. In July, at location A, the Sun appears directly overhead of point in the
northern hemisphere. In January, at location B, the Sun appears directly overhead of this
point in the southern hemisphere. Accordingly, as the Earth progresses in its orbit about the
Sun in the solar ecliptic plane, it will reach a point at about March 20, when the Sun will be
directly overhead of a point on the Earths equator as the Sun makes this transition from
southern to northern latitudes. This particular time, the Vernal Equinor, defines a geometry
such that a line from the Earths centre C through the line of intersection of the solar ecliptic
plane and the plane of the Earths equator defines the direction of Aries.
2.2 A Simple Elliptical Orbit
While orbital motion may be rather complex, we commence with the description of an
elliptical orbit and define some of the important variables. With reference to figure 1.2, the
orbital position r(t), at any time t, of a smaller mass m (eg a satellite) in orbit about the larger
spherically symmetric and uniform density mass M, such as the Earth (note: because of
spherical symmetry the larger mass may be considered a point mass with its centre of mass at
C), is described by the equation of an ellipse, namely,
r = a(1 - e
2
)/(1 + cos ), and (2)
e = orbit eccentricity,
a = ellipse semi-major axis,
= the true anomaly, measured from perigee P (the lowest point in the orbit where
m is closest to M).
E
W
S
N
E
W
S
N
Sun
Solar Ecliptic
Plane
Earths orientation
in July
Earths orientation
in January
A
B
Summer
Wint
er
Note that apogee A (point in the orbit where m is furthest from M) is radian (or 180)
removed from perigee.
Figure 1.2 The orbital plane of an elliptical orbit, of eccentricity e, with the satellite shown
at S. the Ascending Node N, Perigee P and Apogee A are shown. The Earths centre of mass
is located at C a focus of the ellipse. The vector r identifies the instantaneous location of
the satellite in the orbital plane. Q is the mid-point of the major axis AP of the ellipse and is
of length 2a as shown by the sum of the three segments AQ, QC and CP. The angles w and
are explicitly defined in figure 3.
The period T of this elliptical orbit is given by,
T = 2(a
3/2
/ G
1/2
M
1/2
), (3)
where, for the case of the Earth and an orbiting satellite,
G = Earth's gravitational constant, and
M = mass of the earth.
We have neglected all perturbing effects (no atmosphere so atmospheric drag is zero, no solar
wind, no tidal effects, no oblateness of the Earth).
P
Q
A
N
S
r
a a e a (1-e)

Figure 1.3: The sub-satellite track defines an imaginary line on the Earths surface such as
would be traced out by a straight line drawn from the Earths centre C to the instantaneous
position of the satellite. When the sub-satellite track intersects the Earths equatorial plane,
as the satellite moves from south to north, it defines the location of the Ascending Node N.
the orbits inclination, i, is the angle between the plane of the satellite orbit and the Earths
equatorial plane measured anticlockwise from the latter. The Right Ascention of the
Ascending Node is the angle measured eastward from the direction of Aries to the Ascending
Node N. The angle W, the Argument of Perigee, measures the angle in the satellites orbital
plane between the Ascending Node N and the Point of Perigee P. The satellites
instantaneous position in the orbital plane, the True Anomaly, is measured by the Angle
from the Point of Perigee in the sense of the direction of satellite motion to the vector r . The
term Mean Anomaly is substituted frequently for the True Anomaly.
For the satellite orbit illustrated in Figure 1.3, the centre (of mass) of the Earth C is the origin
of two of the coordinate axes. That is, an axis from C through the North Pole and a second
axis directed to the Point of Aries have been defined as two coordinates. For clarity the third
Earths Equatorial
Plane
Sub-Satellite
Track
S
E W
C
P

Aries
N

r
r
i
axis of the coordinate system has not been drawn on the diagram but it is at right angles to the
two defined axes.
The satellites ephemeris, namely ts position in space as a function of time, as described by
the coordinate system, is shown in figure 1.3 together with the sub-satellite track which is
that imaginary line traced out over the Earths surface by the line joining the instantaneous
position of the satellite to the Earths centre.
The point at which the sub-satellite track crosses the equator as it crosses from the southern
hemisphere to the northern hemisphere defines the location of the Ascending Node N. The
angle measured eastward from Aries to the location of the ascending node is the Right
Acsention . The angle measured anti-clockwise between the equatorial plane to the sub-
satellite track is the inclination i of the orbit. As we will see the inclination is an important
satellite injection variable that controls the north-to-south latitude extent of the satellite orbit
and the total number of orbits per 24 hour period.
SUN
Aries
Aries
N
N

Earths
Orbit around
the Sun
A
B
C
Figure 1.4 View, from above the North Pole, two sequential positions of the Earth in its orbit
around the Sun. The direction of Aries (fixed direction relative to the celestial sphere) is
shown and we view the system from above the Earths orbital plane with the location of the
North Pole as shown. The satellite orbital plane A will remain fixed with respect to Aries as
the Earth moves around the Sun resulting in the orbital plane B. If we wish to achieve orbit
C we will need to introduce a precession of the orbital plane by changing the right Ascention
as a function of time. Thus the sequence A to B has
&
= O whereas A to C requires a
specific value of
&
that matches the motion of the Earth around the Sun (2 radian/year).
This results in a sun-synchronous orbit.
The angle , the Argument of Perigee, is determined by the angle in the satellites orbital
plane between the Ascending Node N and the Point of Perigee P. The satellites
instantaneous position in its orbital plane, the True Anomaly, is described by the angle
from the direction of the Point of Perigee (in the sense of the direction of satellite motion) to
the direction of the satellites instantaneous position vector r. Note that figure 1.2, which
represents the satellites orbital plane, shows these angles more clearly than figure 1.3. The
term Mean Anomaly is sometimes used instead of True Anomaly.
2.3 Polar or Sun Synchronous Orbits
The value of the inclination angle i will determine key attributes of an orbit. If i =0, the orbit
will be in the equatorial plane. If i > 90, the orbit is classified as retrograde whereas if i <
90, the orbit is classified prograde. Most Earth observing satellites are launched so as to
have retrograde orbits. The reason for this is shown in figure 1.4 where the Earth is viewed
from above the North Pole N. As the Earth orbits the Sun we examine the motion of the
satellites orbital plane. For the orbital configuration A, the satellite could observe a position
on the Earths surface at local noon. At configuration B, we see the satellites orbital plane
has maintained its orientation with respect to the Earths orbital plane and the Point of Aries.
The consequence of this, however, is that the satellite in orbit B now cannot make
observations of the Earth at local noon. If it is important (eg for making global change; or to
minimise effects due to Sun-pixel-satellite geometry bidirectional effects) for Earth
observations to be made at the same time each day over the whole annual cycle, it is clear
that the Earths orbital plane must precess; that is, move to orientation C, as shown in figure
1.4. Over an annual cycle the orbit will need to precess at the rate of 2 radian (or 360) per
year. Note that the sense of the precession is to the East and corresponds to the Right
Ascention changing as a function of time. This type of orbit is term sun-synchronous
meaning its observations are locked into a fixed (solar) time of the day. So, orbits with i >
90 are sun-syncronous and have retrograde precession of their orbits with the value of d/dt
positive. It is quite common for such satellite orbits to be described as polar orbiting
satellites. Figure 1.5 illustrates the geometry for retrograde and prograde orbits.
In contrast, i < 90 classifies an orbit as prograde and the precession is to the West with
decreasing and d/dt is negative. While it is true these orbits cross the equator and travel
poleward they in general do not cross the poles. That is, an orbit of inclination angle i = 70
will only pass between the latitudes +70 and 70 and will never pass over the poles. Note,
however, if i = 90, the orbit will pass over the poles and the orbital plane will not precess.
A more detailed treatment of the orbital equations for Earth satellites is provided in Maul
(1985), Stewart (1984). A particularly detailed treatment of orbital determinations for
celestial and plantary objects is provided by Escobal (1976). The intention here is not to go
into the details of orbital dynamics but to appreciate those aspects that are important for Earth
sensing systems. Accordingly, when required the appropriate expression will be stated and
typical values for earth satellite substituted to illustrate parameter values of interest.
For example, as noted above, to achieve a sun-synchronous orbit the orbital precession must
be
. 10 99 1 1
24 . 365
2
1 7 1 1

s rad x day year rad

u
dt
d
o
(4)
If a satellite in a sun-synchronous circular orbit was deployed at an altitude above the Earths
surface of 1450 km, the inclination may be determined as follows.
We have,
), (
3
2
cos
2
2 / 7
o e ty eccentrici for
a GM J
a
dt
d
i
e

(5)
where a = radius of satellite orbit = 7828 km,
J
2
=1082.63 x 10
-6
(see below)
GM = the Earths gravitational constant = 398, 600 km
3
S
-2
(product of G and M the gravitational constant and the mass of the
Earth),
a
e
= radius of the Earth = 6378 km at the equator.
This yields, . 7 101 , 203 0 cos
o
i and i
Hence, the orbit is retrograde and constrained to be sun-synchronous. The figures provided
here are somewhat typical of the NOAA/AVHRR series of satellites of which there are
normally two on orbit at any time. Normally, one is deployed with a morning ascending node
and the other an afternoon ascending node.
The quantity J
2
in equation (5) is indirectly a measure on the no-sphericity of the Earth itself
which is normally described as an oblate spheroid. This equatorial bulge produces a
gravitational potential field (the analogue of the electrical potential with which we are more
familiar). When this gravitational potential is expanded mathematically into a series of
spherical harmonics, the term J
2
represents the bulk of the non-sphericity.
If we consider a circular orbit, we see that the rate of change of the True Anomaly (see
Figures 1.2 and 1.3) will determine the period of the orbital motion . An approximation to d
/dt for a circular orbit is provided by the expression,
.
3
2
r
GM
dt
d

,
_

(6)
So the period of the satellite orbit T is,
T = ( )
2 / 1
3
2 / / 2

1
]
1

r
GM
u dt d u
Hence for a 1450 km altitude orbit,
Orbital Period = T . min 9 114 7 6892
600 , 398
) 7828 (
2
2
1
3

1
]
1

s u
During this interval, the Earth will have rotated 29 in an easterly direction. Hence, once this
satellite has completed one orbit and returned, the point it originally observed (say, the
original sub-satellite point) will now be displaced approximately 3200 km to the East (29/360
of 40,074 km, where the latter figure is the circumference of the Earth at the equator). In
simple terms (neglecting orbital precession), we may think of this as the Earth rotating under
a satellite orbit that maintains a fixed orientation is space. The result of this is that satellites
in sun-synchronous orbits may achieve global coverage typically in one day. The projection
of ephemeris onto the Earths surface, the sub-satellite track, illustrates this very clearly.
2.4 Geostationary or Geosynchronous Orbits
An important class of satellite orbits is the geostationary or geosynchronous orbit. For such
orbits the inclination angle i = 0 and the orbit lies in the equatorial plane. The normal
requirement for such orbits is to place them directly over the equator at a particular longitude.
For example, the GOES East and GOES West satellites are placed in orbits over the
Americas (normally at 75 W and 135 W, the Japanese Geostationary Meteorological
Satellite (GMS) is located over New Guinea and Meteosat over Europe. For this to occur it is
clear that the orbital period of these satellites must equal the period of rotation of the Earth.
That is, the satellite period is one sidereal day or 86,164 s.
If geostationary orbits are such that i 0, but is small, then the sub-satellite track is what is
frequently termed a figure 8. This type of orbit is shown in figure 1.6 for the case where i
= 10, and we see the satellite N-S extent matches the inclination angle. A non-zero i may
not always be intentional. If a satellite is injected into orbit it is sometimes difficult to set i
precisely to zero and, of course, valuable on-board fuel is consumed in correcting the orbit.
So, a small offset is not particularly problematic as long as the wandering of the satellite in
its figure 8 does not stray outside of the reception angle of the fixed-direction ground
station antenna.
Equation (6),with appropriate substitutions, becomes,
T (min) = 1.66 [6378 + h (km)]
3/2
.10
-4
,
where again, h is the altitude of the satellite above the Earths surface.
So for a geostationary satellite orbit with a period of 2 day
-1
, we may determine h. For the
values given, we see that for a geostationary orbit
h = [24*60 min ]/[1.66*10
-4
] 6378 = 35,843 km.
Accordingly, geostationary satellites are positioned some 35 times the altitude of the typical
polar orbiting satellites. A major consequence of this contrast in orbital deployment is that the
geostationary satellites have to be equipped with much larger aperture (diameter) optical
systems if they are to resolve spatially comparable detail on the surface of the Earth.
4.0 Computer-based Modelling of Satellite Orbits
Models of satellite orbits are routinely used to support satellite data collection from receiving
stations. Typically, data downlink station or direct readout station must have predictions of
when the satelliote of interst will rise over the horizon and the appropriate azimuth angle at
which the receiving antenna or dish must be pointed. In order to track the satellite as it
moves overhead the antenna may require a sequence of pairs of azimuth and zenith angles as
a funtion of time to maintain accurate tracking. Because many of the perturbing forces
(atmospheric drag, solar wind, approximations to the Earth's gravitational field as expressed
by the higher order terms in the gravitational potential) are not captured in the prediction
models the model predictions do not maintain their accuracy for an extended period.
Typically, one runs the prediction model approximately every few days (maybe weekly -
depending on the satellite) to update the orbital prediction. To achieve this with acceptable
accuracy, models resort to use of Orbital Elements to reset the initial conditions in the
computer-based model. These orbital elements are derived from direct measurement of the
satellites position made using satellite laser ranging (SLR) instruments. These instruments
locate the satellite position in the sky by reflecting a laser pulse off reflectors (corner cube
reflectors) on the satellite. The time for the round trip of the laser pulse provides an accurate
range (order of a cm accuracy is typical) and the azimuth and elevation at which the laser
fired record the direction. By tracking the satellite across the sky the SLR defines the
satellite ephemeris (ie its orbit in space and time). From these measurements a set of orbital
elements are derived that are appropriate to input to the orbital prediction model. There are
two common sets of orbital elements used for satellite prediction, namely T-bus elements and
(NORAD) two-line elements. Several of these orbital elements will be familiar. here we will
discuss just the two-line elements (2LE). The elements as illustrated below are provided as
two-lines of consecutive characters. In order to aid location of particular elements, the
special characters are used above the first line of elements and immediately below the second
line of elements. The identity of those elements used in orbital prediction models will be
described below.
* * ## ++$ $
1 21263U 91 32 B 93231.76315203 .00000177 00000-0 88271- 4 0 6491
2 21263 98.6545 260.6933 0013797 33.2603 326.9449 14.22300920117899
! ! ? ? " " : : % % / / \ \
Satellite Number = 21263 (see line 1 characters between and below the asterisks (*)
Each satellite is assigned a distinct identification number.
Launch Year = 91 (line 1 below the symbols ##)
The year in which the spacecraft was launched (eg 1991)
Epoch Year = 93 (line 1, between and below (++) characters)
The year the elements were obtained (eg 1993)
Epoch Day = 231.763115203 (line 1 between and below the $ characters)
The epoch day of year in GMT or UTC (1 epoch day 367)
Inclination = 98.6545 (line 2 above and between the ! characters)
The angle at which the orbit plabecrosses the equatorial plane as described earlier.
Right Ascention of the Ascending Node = 260.6933 degree ( line 2 between and above the ?
characters). This parameter was defined earlier.
Orbit Eccentricity = 0.0013797 (line 2 between and above the " characters)
Note the decimal point is not provided in the elements. A zero to the left of the decimal point
has been added post extraction to emphasise the eccentricity e is less than unity) As defined
earlier the eccentricity defines the shape of the elliptical orbit. Ab eccentricity of defines a
circular orbit. The degree of ellipticity increases as the eccentricity increases toward unity (0
e 1).
Argument of Perigree = 33.2603 degree (line 2 between and above the : characters)
Defined previously as the angle in the orbital plane from the ascending node to the point of
perigee in the direction of the satellite's motion.
Mean Anomaly = 326.9449 (line 2 between and above the % characters)
The angle in the orbital plane from the the point of perigee to the line of the vector from the
Earth to the satellite's position in the direction of the satellite's motion.
Mean Motion = 14.22300920 (line 2 between and above the / characters)
The mean number of orbits per day.
Orbit number = 11789 (line 2 between and above the \ characters)
The current number of Earth revolutions completed by the satellite since launch. The value is
incremented when the satellite crosses the equator on an ascending pass (northward bound).
If a model requires any other elements (eg semi-major axis) they may be computed from the
existing elements
A software package SeaTrack (Lambert et al, 1993) describes a particular orbital prediction
model. The above description of the two-line elements has been drawn from that publication.
In the Workshop, a particular orbital prediction model will be studied in the Laboratory
program. Experince will be gained in model execution and its application to the orbital
prediction of the SeaStar platform on which SeaWiFS is deployed. The model selected will
also permit the determination of the area over which a particular satellite overpass will
deliver observational data when the satellite is tracked by a specific ground receiving station
(at some predetermined lat and long). Such information is particularly useful when planning
research cruises, enabling a cruise vessel to position itself favourable so that is within the
geographical scene (reception area) captured by the satellite orbit for that particular day.
4.2 Model Outputs
Another orbital prediction model is available at the URL <http://eospso.gsfc.nasa.gov> and
may be executed online. The type of information provided by this site is appropriate for
tracking particular satellites since it provides the coordinates for initialising the antenna when
the satellite rises over the horizon and permits programming the antenna tracking system's
pointing direction and rate of progress as it follows the satellite's motion across the sky.
6.0 Concluding Remarks
The intention of this segment of the Workshop was to make the point that satellites are
instruments, just like laboratory instruments, and we need to understand how to use them and
why we use them in different ways. Further, it is apparent that we have considerable control
over the precise way we deploy a satellite on-orbit so that it can make the observations
appropriate to our needs. The variables used to describe a satellite's orbit are not the simple
variables that we might have hoped for - for example position (x, y and z ) velocity (v
x
, v
y
and v
z
) at time t. The variables,in most cases, have their origins in astronomy. Direct read-
out (or downlink) satellite stations need information so they are ready to receive satellite data
transmissions as soon as the satellite appears over the horizon. This requirement has
stimulated the development of models of satellite orbits. The initial conditions are provided
to the models via a set of orbital elements that are in turn derived for direct observations of
the satellite ephemeris using laser tracking and ranging instruments
7.0 References
Brouwer, D. 1959. Solution of the Problem of Artificial Satellite Theory Without Drag.
Astron. J., 64, 378-397.
Escobal P. R. (1976). Methods of Orbit Determination. (Robert E Krieger Publ., Florida)
Second Edition.
Lambert, K., Gregg, W., Hoisington, C. and Platt. F. S. 1993. SeaTrack - Ground Station
Orbit Prediction and Planning Software for Sea-Viewing Satellites. NASA Refereence
Publication #1331, Dec. 1993, NASA Scientific and Technical Information Branch.
Lyddane, R. H. 1963. Small Eccentricities or Inclinations in the Brouwer Theory of the
Artificial Satellite. Astron. J., 68, 555-558.
Maul, G. A. 1985. Introduction to Satellite Oceanography. (Martinus Nijhoff, Dordrecht).
Stewart, R. H. 1985. Methods of Satellite Oceanography. (Scripps Institute of
Oceanography, Univ of California, San Diego)
Course Notes for Ocean Colour Remote Sensing Course
Erdemli, Turkey
September 11 - 22, 2000
Module 2: Basic Radiometric Quantities and
Definitions
prepared
by
Assoc Professor Mervyn J Lynch
Remote Sensing and Satellite Research Group
School of Applied Science
Curtin University of Technology
PO Box U1987
Perth Western Australia 6845
AUSTRALIA
tel +618-9266-7540
fax +618-9266-2377
email <[email protected]>
Module 2: Basic Radiometric Quantities and Definitions
1.0 Geometry, Angles and Solid Angles
An angle is defined in two dimensional space as a ratio of two lengths - an arc of a circle
divided by the radius of that circle - see figure 2.1. It is by definition dimensionless and the
unit of angular measure is termed the radian (abbreviated to rad). When the length of the
arc is equal to the radius the angle is 1 rad. If we increase the angle progressively unit we
occupy a complete circle the angle now is measured by the circumference (2r) divided by
the radius r of the circle. Thus a complete circle corresponds to an angle of 2 rad or
approximately 6.28 rad. Frequently angles are measured in degrees with
360 = 2 rad.
(insert figure 2.1 and caption here)
Figure 2.1 Definition of an angle in two dimensions is that angle subtended by an arc of
length a at a radius r.
Note angle is a ratio of two lengths and therefore is dimensionless. If a = r, the angle is 1
radian.
In the real world we more frequently encounter what are termed solid angles. For example, if
we view a "circular" object (the moon) then the eye and the moon's area (its area appears to
be a flat disk of r
2
m
2
where r is now

the radius of the moon ) define a conical volume in
space. This "angle" is termed a solid angle and is measured by the area of the object divided
by the radius squared (in this case the eye - moon distance. The unit of solid angle measure is
the steradian (abbreviated sterad). If now we imagine travelling toward the moon we will
reduce the distance (eye - moon) but the area presented by the moon is unchanged. This
causes the solid angle to increase. If now we imagine ourselves at the centre of a sphere of
radius r then the surface of the sphere 4r
2
is now the appropriate "area" and it is at radius r.
This is the largest solid angle we can create and it corresponds to 4r
2
/ r
2
or 4 steradian.
Figure 2.2 illustrates the measurement of a solid angle and introduces the azimuth and zenith
angles and respectively.
(insert figure 2.2 and caption here)
Figure 2.2. The element of solid angle d is defined in terms of the elemental area dA at
distance r from the origin O. dA is produced by small increments d and d in the zenith
angle and azimuthal angle respectively, so that
2
/ r dA d .
Note that, in figure 2.2, the elemental surface area dA has sides of length rsind and rd
giving the area dA as,
dA = r
2
sindd. (1)
2.0 The Basic Radiometric Quantities
2.1.1 Irradiance
An optical beam may be thought of as a collection or ensemble of photons each carrying
energy. If this beam is incident on a surface then the rate that energy (joule or J) arrives at
that surface is termed power may be measured in joule s
-1
or J s
-1
This latter unit is the
definition of the watt or W. Frequently in radiometry the power in an optical beam per unit
area (see figure 2.3a) is termed the radiant flux d but more commonly the irradiance E
(the unit is W m
-2
). Hence, in figure 2.3a, the irradiance E incident on the surface due to the
radiant flux d falling on the area dA is d/dA. This quantity will be encountered frequently
in remote sensing.
In figure 2.3a the beam was normal to the surface. Figure 2.3b shows the direction of the
beam now impacting the surface at an angle of incidence . Since the beam, of the same
power, is now distributed over a larger surface area, when compared with the case in figure
2.3a, the irradiance will be reduced by the factor cos since the area impacted by the beam
has increased from dA to dA/cos.
(insert figure 2.3 and caption here)
Figure 2.3. Consider a beam of radiation of radiant flux or power d (W) incident on a
horizontal surface. In case (a) the radiant flux per unit area, the irradiance
dA d E
a
/ (Wm
-2)
.
For case (b) where the beam is now inclined to the surface at a zenith angle we see the area
over which the beam irradiates the surface is now dA/cos. The reduced irradiance in case
(b) becomes dA d dA d E
b
/ cos ) cos / /( .
2.1.2 Radiance
Consider the geometry illustrated in figure 2.4 where the the normal to a small element of
surface dA is inclined at an angle to the direction from which we observe the radiant flux
from the surface. The determine the radiant flux or radiant power d leaving area dA (eg
radiation from the ocean to the atmosphere) that fills or passes through the solid angle d.
This quantity is defined as the radiance L, where, from its definition,
L = d / (dA.cos.d), (2)
and the units of radiance are W m
-2
.ster
-1
. Sometimes radiance exiting a surface is termed
exitance.
(insert figure 2.4 and caption here)
Figure 2.4(a). The radiant power d leaving an elemental projected area dA on an extended
source (such as the ocean surface) per unit solid angle d is defined as the radiance L and
expressed as,
). (
cos
1 2

sr Wm
d dA
d
L

Figure 2.4(b) The projected area concerned, namely, dAcos is shown.

The quantity radiance corresponds to the type of observation we might conceive of making in
practice. For example, if we viewed the small element of surface area dA in figure 2.4 from,
say, an aircraft, using a telescope that had an aperture such that the solid angle involved was
d, then, according to equation 2, we would determine the radiance. The same circumstance
applied when we make observations from a satellite platform.
It is worth noting that solid angle is important generally in characterising laboratory sources.
For example, we could compare a beam from, say, a 60 W laser to that from a 60 W light
bulb. The important issue here is the relative solid angles over which the radiation from these
two sources is distributed. For example, these might be 10
-5
sterad for the laser versus 4
sterad for the light bulb - a ratio of some 10
+6
in solid angle. So it is the appropriate solid
angle that is the key variable in determining the radiance from these two sources that have
identical output powers. Useful sources to consult are Maul (1985), Stewart (1985) and Kirk
(1986).
3.0 More About Radiance
Making measurements in radiance units is an important concept in radiometry. We define an
elementary beam using figure 2.5 as an aid. The notion of an elementary beam is that all rays
that pass through the aperture dA
1
also pass through aperture dA
2
. If the latter holds, then the
power passing through dA
1
equals the power passing through dA
2
. These two aperatures,
with surface normals as shown, are centered at X and Y respectively and subtend solid angles
to each other as shown in the figure. Note that
1
and
2
respectively are the angles between
the surface normals and the direction of propagation of the beam of radiation, namely from X
to Y.
(insert figure 2.5 and caption here)
Figure 2.5. The concept of an elementary beam is illustrated: all rays that pass through
d, and d
2
dA, also pass through dA
2
. The respective solid angles are shown along
with the surface normals .
. 2 1 n n
U and U
We now can write down the appropriate radiances at the two apertures,
L
1
= d
1
/ (dA
1
. cos
1
. d
1
), and (3)
L
2
= d
2
/ (dA
2
. cos
2
. d
2
). (4)
However, since,
d
1 =
d
2,
we have

from (3) that,
L
1
.dA
1
.cos
1
.d
1
=

L
2
.dA
2
.cos
2
. d
2
. (5)
However, using the definition of solid angle we also have that,
d
1
=

(dA
2
.cos
2
)/r
2
and d
2
=

(dA
1
.cos
1
)/r
2
. (6)
If we now substitute equations (6) into (5) we obtain the particularly important relationship
that, for an elementary beam,
L
1
= L
2
(7)
4.0 Making Earth Observations from Space
Figure 2.6 illustrates schematically how we might use a simplified sensor on a satellite to
observe the surface of the Earth. Note that while we have not included lenses or mirrors in
the sensor, these do not make the arguments any more complex if we observe the elementary
beam concept. For example, if a ray enters the sensor foreoptics (front aperture) it is normal
to ensure that that ray finally falls on the detector surface. In other words, to achieve
maximum performance out of a satellite instrument, we normally design it to conform to an
elementary beam system since, to do otherwise would degrade the sensor.
We have that the radiant power leaving the elemental surface area dA propagates to the
satellite, and (neglecting any atmospheric absorption or scattering) enters the sensor
aperturearea A. Hence the area dA at the Earth's surface and the sensor aperture A conform
to an elementary beam system.
(insert figure 2.6 and caption here)
Figure 2.6. Schematic of a space-based radiometer of aperture A and detector of area D
viewing the ocean surface with the solid angles
d s
and referencing the sea and detector
respectively.
Thus the radiant power appears to originate at the aperture. Since,

D
= D / d
2
,
the radiance passing through the aperture toward the detector may be written,
L = d / A
D
= d / (A.D/d2) .
However, A / d
2
is the solid angle
S
at the detector, and
L = d / (D.
S
).
This is a key result in that it shows that if we make a measurement of the radiance based on
thegeometry (design) of the sensor, the elementary beam concept indicates that this radiance
is identical to that of the target we are observing on the Earth's surface!
4.0 Vector and Scalar Quantities
Later in the Workshop terms like scalar or vector irradiance will be introduced as will
conversion from radiance to irradiance and vice versa. The important quantity normalised
water leaving radiance will be encountered frequently in other modules and the definitions of
these new quantities from the basic variables will be developed. On these more detailed
issues, the text by Kirk (1986) is a particularly useful resource.
6.0 Concluding Remarks
For thisModule, the aim has been to introduce the basic concepts and to account for how and
why these quantities are defined as they are. . While some of the concepts, such as the solid
angle, may be new, particularly with respect to its use in measurment, it is clear that solid
angles are more commonly encountered in our three dimensional world of observations than
two-dimensional angles. The importance of the elementary beam concept to radiometry, and
in particular to space-based radiometry, is hopefully clear. It ought also be clear as to why
the basic quantity - radiance - is almost the natural variable for use in making Earth
observations with instruments. When we encounter radiometry applied to remote sensing of
the oceans, issues such as optical propagation across the air-water interface and the processes
of scattering and absorption in the ocean water itself as well as particulate material and
pigments complicates the science. Nevertheless, at all stages, the processes involved and
their impact on a propagating beam need to be represented in terms of signal amplitudes,
propagation directions and just a few basic quantities and their associated units.
7.0 References
Kirk, J. T. O. 1986. Light and Photosynthesis in Aquatic Ecosystems.(Cambridge University
Press).
Maul, G. A. 1985. Introduction to Satellite Oceanography. (Martinus Nijhoff, Dordrecht).
Stewart, R. H. 1985. Methods of Satellite Oceanography. (Scripps Institute of
Oceanography, Univ of California, San Diego)