VIRTIS: The Visible and Infrared Thermal Imaging Spectrometer

G. Piccioni1, P. Drossart2, E. Suetta3, M. Cosi3, E. Ammannito4, A. Barbis3, R. Berlin6,

A. Boccaccini1, G. Bonello1, M. Bouyé2, F. Capaccioni1, G. Cherubini3, M. Dami3, O. Dupuis2,
A. Fave2, G. Filacchione1, Y. Hello2, F. Henry2, S. Hofer5, G. Huntzinger2, R. Melchiorri2, J. Parisot2,
C. Pasqui3, G. Peter6, C. Pompei3, J.M. Rèess2, A. Semery2, A. Soufflot7 and the VIRTIS Co-I team:
A. Adriani4, F. Angrilli8, G. Arnold6, K. Baines9, G. Bellucci4, J. Benkhoff6, B. Bezard2, J.-P. Bibring7,
A. Blanco10, M.I. Blecka11, R. Carlson9, A. Coradini4, A. Di Lellis1, T. Encrenaz2, S. Erard7, S. Fonti10,
V. Formisano4, T. Fouchet2, R. Garcia12, R. Haus6, J. Helbert6, N.I. Ignatiev13, P. Irwin14, Y. Langevin7,
S. Lebonnois15, M.A. Lopez Valverde16, D. Luz12, L. Marinangeli17, V. Orofino10, A.V. Rodin13,
M.C. Roos-Serote18, B. Saggin19, A. Sanchez-Lavega20, D.M. Stam21, F. Taylor14, D. Titov22,
G. Visconti23 & M. Zambelli1
1
Istituto di Astrofisica Spaziale e Fisica Cosmica (INAF-IASF), via Del Fosso Del Cavaliere 100,
I-00133 Rome, Italy
Email: [email protected]
2
Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique (LESIA), Observatoire de
Paris/Meudon 5, Place Jules Janssen F-92195 Meudon cedex, France
3
Galileo Avionica, via A. Einstein 35, I-50013 Campi Bisenzio (FI), Italy
4
Istituto di Fisica dello Spazio Interplanetario (INAF-IFSI), via Del Fosso Del Cavaliere 100,
I-00133 Rome, Italy
5
Kayser-Threde GmbH, Wolfratshauser Strasse 48, D-81379 Munich, Germany
6
German Aerospace Center (DLR) Institute of Planetary Exploration Planetary Physics, Berlin-
Adlershof Rudower Chaussee 5, Geb. 16.16, D-12489 Berlin, Germany
7
Institut d’Astrophysique Spatiale, Bâtiment 120 Université Paris-Sud, F-91405 Orsay cedex, France
8
CISAS Università di Padova, via Venezia 1, I-35131 Padova, Italy
9
Jet Propulsion Laboratory, MS 183-601 4800 Oak Grove Drive, Pasadena CA 91011, USA
10
Università degli Studi di Lecce Dipartimento di Fisica, Via Arnesano, I-73100 Lecce, Italy
11
Space Research Centre of Polish Academy of Science, Bartycka 18A, 00-716 Warszawa, Poland
12
Département des Études Spatiales Institut de Physique du Globe de Paris 4, Avenue de Neptune
F-94107 Saint Maur des Fossés cedex, France
13
Space Research Institute (IKI), Profsojuznaja 84/32, 117997 Moscow, Russia
14
University of Oxford, Dept. Physics Atmospheric, Oceanic and Planetary Physics,
Clarendon Laboratory, Parks Road, Oxford OX1 3PU, UK
15
Laboratoire de Meteorologie Dynamique, Jussieu, Box 99 F-75252 Paris cedex 05, France
16
Instituto de Astrofísica de Andalucía (CSIC), Camino Bajo de Huétor, E-18080 Granada, Spain
17
International Research School of Planetary Sciences, Dipartimento di Scienze Universita’ d’Annunzio
Viale Pindaro 42, 65127 Pescara, Italy
18
Observatorio Astronomico de Lisboa Centro de Astronomia e Astrofisica da Universidade de Lisboa,
Tapada da Ajuda P-1349-018, Lisboa, Portugal
19
Politecnico di Milano, Polo di Lecco, via Marco D’Oggiono 18/A, I-23900 Lecco, Italy
20
Dpto. Física Aplicada I Escuela Superior de Ingenieros Universidad del Pais Vasco, Alda. Urquijo s/n
E-48013, Bilbao, Spain
21
Astronomical Institute ‘Anton Pannekoek’ University of Amsterdam, Kruislaan 403 1098 SJ,
Amsterdam, The Netherlands
22
Max-Planck-Institute for Aeronomy, Max Planck Str. 2, D-37191 Katlenburg-Lindau, Germany
23
Dept. of Physics, University of L’Aquila, via Vetoio Loc. Coppito, I-67010 Coppito, L’Aquila, Italy
SP-1295

The VIRTIS imaging spectrometer built for ESA’s Rosetta cometary mission is a
versatile instrument that is also well-suited to Venus observations. The discovery of
the near-IR windows in the atmosphere of Venus from ground-based observations in
the 1980s showed that the surface of the planet can be studied via IR observations
over the nightside. Imaging spectroscopy in the visible and near-IR can study the
atmosphere from the uppermost layers down to the deepest levels. With its unique
combination of mapping capabilities at low spectral resolution (VIRTIS-M) and
high spectral resolution slit spectroscopy (VIRTIS-H), the instrument is ideal for
making extensive IR and visible spectral images of the planet.

1. Scientific Objectives Observations of the nightside of Venus in the near-IR from Earth, first made by
Allen & Crawford (1984), gave the first evidence of deep atmospheric sounding.
These were followed by accurate spectroscopic measurements of the atmosphere
with high spectral resolution spectroscopy (Bézard et al., 1990), including the
water vapour and carbon monoxide that is important for thermochemical
equilibrium studies (Kamp et al., 1988).
The first attempts at imaging spectrometry of the Venus nightside from space
in the near-IR were made by Galileo’s Near-IR Mapping Spectrometer (NIMS)
in 1990 (Fig. 1; Carlson et al., 1991) and Cassini’s Visual & IR Mapping
Spectrometer (VIMS) in 1999 (Baines et al., 2000). These fast flybys indicated
how powerful this method of investigation could be at Venus. Unfortunately,
their brief durations strongly limited the investigations possible, particularly the
meteorological evolution of the clouds. Observations of Venus with the VIRTIS
new-generation imaging spectrometer is a unique opportunity to extend these
investigations.
VIRTIS is able to produce day and nightside IR and visible spectra. The
observation strategy for these scientific objectives is following the observing cases
detailed in Section 5. The whole atmosphere is being observed from the
mesospheric levels down to the surface (Taylor et al., 1997), and the surface itself
is accessible to VIRTIS IR observations on the nightside. The main topics for
VIRTIS science observations are:

— study of the lower atmosphere composition below the clouds and its
Fig. 1. The Venus nightside at 2.3 μm, taken by variations (CO, OCS, SO2, H2O) from nightside observations (Collard et al.,
Galileo (Carlson et al., 1991). The thermal 1993). The 2.3 μm window (at high resolution with VIRTIS-H) gives access
emission from the deep atmosphere is modulated to accurate measurement of minor species. At shorter wavelengths, H2O is
by the cloud structure of the deeper atmosphere.
Cloudy regions appear in blue (lower emission), measurable with the lower spectral resolution of VIRTIS-M for mapping
while bright regions (red) correspond to less H2O variations, as done by NIMS/Galileo (Drossart et al., 1993). From these
cloud. observations, correlation with volcanic or meteorological activity is being
sought (some signatures of volcanic outgassing are related to minor-species
variations, in particular for sulphur compounds).
— study of the cloud structure, composition and scattering properties (dayside
and nightside). The different geometries of cloud reflections along with the
spectral range of VIRTIS will constrain the cloud structure. The average
structure is known from previous missions, but the temporal and spatial
variabilities are less well documented. VIRTIS can sound the different layers
and measure their optical thickness in the IR, and measure the upper layer
on the dayside.
— cloud tracking in the UV (~70 km altitude, dayside) and IR (~50 km altitude,
nightside). The correlation of UV and IR observations at different times,
along with the 4-day cloud rotation period, gives access to the vertical
variation of the wind field up to 70 km altitude.
— measurements of the temperature field with subsequent determination of the
zonal wind in the altitude range 60–100 km (nightside). The 4–5 μm range is

2
 scientific instruments

sensitive to thermal structure, which can be retrieved (on the nightside) and
compared with models (Roos et al., 1993). Such retrieval gives access to the
vertical wind variations through cyclostrophic approximation.
— lightning search (nightside). Although tentative (there is no reliable
information on the frequency of lightning on Venus), observations of
transient lightning illumination are of high scientific interest, and is an ‘open
search’ option for VIRTIS observations.
— mesospheric sounding. Understanding the transition region between tropo-
sphere and thermosphere:
– non-local thermodynamic equilibrium (non-LTE) O2 emission
(night/dayside) at 1.27 μm (95–110 km; Drossart et al., 1993). As observed
from the ground and NIMS/Galileo, these emissions have spatial and
temporal variabilities, which make them of high interest for accurate
spatial mapping by VIRTIS.
– CO2 fluorescence (dayside). Non-LTE emissions at 4.3 μm (>80 km;
Roldan et al., 2000) on limb scans by Galileo provide important
information on the physics of the upper atmosphere through the
collisional/radiative equilibrium sounded through the CO2 band.
– limb observations (CO, CO2): atmospheric vertical structure (>60 km)
(day/nightside).
— search for variations related to surface/atmosphere interaction, dynamics,
meteorology and volcanism. Global observation or partial observations at a
regional scale on a temporal scale of one Venusian day is allowing the search
for correlations between different physical processes.
— temperature mapping of the surface and searching for hot spots related to
volcanic activity. NIMS/Galileo observations of Io showed that lava lakes
are easily detected on a planet by imaging spectroscopy. Even if the
atmosphere of Venus precludes the clear detection of free lava, a temperature
anomaly could be a signature of some volcanic activity. This has never been
attempted with the spatial/temporal performance of an instrument like
VIRTIS.
— search for seismic waves from the propagation of acoustic waves amplified in
the mesosphere by looking for high-altitude variations of pressure/
temperature in the CO2 4.3 μm band (Artru et al., 2001). Although clearly a
challenge, it is one of the tentative science objectives that would be of
primary scientific importance if detected. Gravity waves also have signatures
in the IR mesospheric emissions accessible to VIRTIS. Very little is known
about wave activity in the upper atmosphere of Venus.

These objectives during the nominal Venus Express mission mean that VIRTIS
is providing a very broad scientific return.

2.1 Overview 2. Instrument Description

Although with some modifications, VIRTIS for Venus Express is essentially
identical to the instrument carried by Rosetta (Coradini et al., 1998). It is an
imaging spectrometer with three focal planes and two channels. The mapping
channel has two 2-D focal planes: visible (0.25–1 μm) and IR (1–5 μm). The
spectroscopic channel has a single aperture but provides high-resolution spectra
at 2–5 μm. The IR focal plane array (FPA) uses HgCdTe arrays of 270 x 438-
pixel detectors designed to provide high sensitivity and low dark current (about
1 fA at 80K), with a read noise lower than 300 e–.
As on Rosetta, the electronic module and data handling support unit were
built in Germany (Kayser-Threde for hardware, DLR for software and
management), VIRTIS-M by Galileo Avionica Florence-Italy and VIRTIS-H by
Paris Observatory, France. Integration and programme management were in

3
SP-1295

Table 1. VIRTIS principal characteristics.

VIRTIS-M Visible VIRTIS-M Infrared VIRTIS-H

Spectral range (nm) 280 – 1100 1050 – 5130 1840 – 4990

Spectral resolution λ/Δλ 150 – 500 100 – 500 1300 – 3000
Spectral sampling (nm) [note 1] 1.89 9.47 0.6
Field of view (mrad x mrad) 64 (slit) x 64 (scan) 0.44 x 1.34
Max spatial resolution (μrad) 250 (slit) x 250 (scan) –
Image size, full FOV high resolution (pixels) 256 x 256 –
noise equivalent spectral radiance 1.4 x 10–2 1.2 x 10–4 1.2 x 10–4
(central band, Wm–2 sr–1 μm–1)

Telescope Shafer Telescope Shafer Telescope off-axis parabolic mirror

Pupil diameter (mm) 47.5 32
Imaging F# 5.6 3.2 2.04
Etendue (m2 sr) 4.6 x 10–11 7.5 x 10–11 0.8 x 10–9
Slit dimension (mm) 0.038 x 9.53 0.029 x 0.089
Spectrometer Offner Relay Offner Relay Echelle spectrometer
Detectors Thomson TH7896 CCD HgCdTe [note 2] HgCdTe [note 2]
Sensitivity area format 508 x 1024 270 x 436 270 x 436
Pixel pitch (mm) 19 38 38
Operating temperature (K) 150 – 190 65 – 90 65 – 90
Spectral range (μm) 0.25 – 1.05 0.95 – 5.0 0.95 – 5.0
Mean dark current < 1 e/s < 2 fA @ 90K < 2 fA @ 90K

Notes. 1: depends on selected mode of operation; the maximum value is shown. 2: VIRTIS-M and VIRTIS-H use identical IR detectors.

Florence. There are differences between the Rosetta and Venus Express
instruments owing to the very different targets: the cold nucleus of a comet and
the hot planetary atmosphere of Venus. The Venus Express design improves
thermal dissipation, adapts the integration timescale and reduces the repetition
time between successive observations. Table 1 provides the instrument’s main
characteristics.
The output from VIRTIS-M (Fig. 2) can be considered to be a large set of
stacked monochromatic 2-D images (spatial direction in the x-y plane) in the
range 0.25–5 μm (l axis), at moderate spectral resolution. The VIRTIS-H field-of-
view is centred in the middle of the -M image and provides spectra at high
spectral resolution in this small portion of the image.
As shown in Fig. 3, VIRTIS is consists of four Modules: the Optics Module
(OM) containing the -M and -H Optical Heads; the two Proximity Electronics
Modules (PEM-M and PEM-H) and the Main Electronics Module (ME).
The Optics Module is externally mounted on the –X panel of the spacecraft
with the Optical Heads co-aligned and boresighted in the +Z direction. Both
optical systems have their slits parallel to the Y axis; VIRTIS-M can point and
scan by rotating the primary mirror around the Y-axis. Two views of the Optics
Module are shown in Figs. 4 and 5.
VIRTIS-M uses a Si CCD (Thomson TH7896) to image from 0.25 μm to 1 μm
and an HgCdTe IR focal plane array (IRFPA, Raytheon) to image from 1 μm to
5 μm. VIRTIS-H uses the same HgCdTe IRFPA to perform spectroscopy in
selected bands from 2 μm to 5 μm (see Table 1).

4
 scientific instruments

Fig. 2. The output from VIRTIS-M can be considered to be a large set of Fig. 3. The VIRTIS functional block diagram. DHSU: Data Handling and
stacked monochromatic 2-D images (spatial direction is in the x-y plane) in Support Unit.
the range 0.25–5 μm (l axis), at moderate spectral resolution. The
VIRTIS-H field-of-view centred in the -M image provides spectra at high
spectral resolution in this small portion of the frame.

Fig. 4. The Optics Module mounted on its support structure shortly before Fig. 5. The Optics Module structure. From a thermal-mechanical point of
installation of the multi-layer insulation and final integration on the view, the structure is divided into two (top and bottom), at very different
spacecraft at Alenia Spazio in Turin. The -M channel is at right top; the -H temperatures. The cold side (top) is directly coupled to the radiator (and
channel is at left top. cold space) and the warm side (bottom) is the baseplate in contact with the
spacecraft.

The electronics boards required to drive the CCD and the two IRFPAs and to
acquire housekeeping information from the optical heads are housed in the
PEMs. VIRTIS-ME controls the interface with the PEMs and the spacecraft,
controls the cryocoolers and the power conditioning and distribution, and
performs all the digital data-handling.
The IR detectors require active cooling to minimise the detector dark current
owing to the thermally generated Johnson noise. Their operating temperature is
typically 80K.
In order to minimise the thermal background radiation seen by these two
IRFPAs, the Optics Module is divided into two parts:

5
SP-1295

Fig. 6. The baseplate Pallet with the two

cryocoolers and the connector bracket, the eight
titanium rods fixing the Cold Box mounting
ledge on the baseplate, and the empty ledge to
which the two Optical Heads are rigidly
connected.

— the Pallet directly interfaces with the warm spacecraft (i.e. it is thermally
controlled by the spacecraft) and houses the two cryocoolers.
— the Cold Box structure is rigidly mounted to the much warmer Pallet but
thermally insulated from it. It contains the cold detectors and the two Optical
Heads and is thermally disconnected from the spacecraft and passively
cooled below 150K by radiating one of its surfaces towards cold space. The
optical heads are mounted on the Cold Box Ledge.

Two cold fingers connect the two active coolers inside the Pallet to their
corresponding IRFPAs inside the Cold Box; they must maximise the thermal
pathway from the coolers to the IRFPAs while remaining mechanically pliant. In
contrast, the stand-off insulators (eight titanium rods) connecting the Cold Box
to the baseplate of the Pallet minimise the thermal pathway between the warm
spacecraft and the cold optical subsystems while remaining mechanically rigid. In
this way, the structure and the delicate subsystems that it supports were not only
guaranteed to survive launch vibrations, but the structure also helps in
minimising the usual thermal gradients that adversely affect the alignment of
low-temperature optical systems.
Since the CCD is a frame-transfer device and the IRFPAs are direct injection
devices, exposure times can be controlled electronically. However, measurement
of the detector dark current plus background radiation requires calibration
shutters to be placed at the entrance light-limiting slit of each spectrometer
(telescope background contribution is negligible because of its low temperature
and the narrow width of the slit).
Each VIRTIS-M and -H channel has a cover to protect the instrument from
direct Sun illumination and to preserve the cold environment inside the Cold
Boxes when the Sun is within the field of view when VIRTIS is not operational.
Each cover has an emergency actuator to force it open in case of malfunction.
The covers are machined as retro-reflectors for in-flight calibration. This
feature also came in useful during ground-testing of the instrument, especially
when it was not on an optical test bench.

6
 scientific instruments

2.2 The team

VIRTIS is the result of the joint efforts of several scientific institutes and
companies in different European countries. The institutes involved in the
hardware development, calibration and testing were:

— Istituto di Astrofisica Spaziale e Fisica Cosmica (IASF), Rome, Italy;

— Laboratoire d’Etudes Spatiales et d’Instrumentation en Astrophysique
(LESIA), Observatoire de Paris, Meudon, France;
— Galileo Avionica (GA), Florence, Italy (VIRTIS prime contractor);
— Institut für Planeteterkundung DLR, Berlin, Germany;
— Kayser-Threde GmbH, Munich, Germany;
— Istituto di Fisica dello Spazio Interplanetario (IFSI), Rome, Italy;
— Institut d’astrophysique Spatial (IAS), Orsay, France;
— Università di Lecce, Italy;
— CISAS-DIM, Università di Padova, Italy.

The tasks were defined on the basis of the VIRTIS-Rosetta structure. The
general sharing of responsibility among countries was:

VIRTIS-M Italy
VIRTIS-H France
Optics Module structure Italy
Active cooling subsystem Italy
Main Electronics Module Germany
Cover subsystem Italy
Ground Support Equipment Italy
Shutter Unit France
Sunshields Italy

The VIRTIS prime contractor is Galileo Avionica, responsible, apart from the
overall integration, for the production and testing of VIRTIS-M, the Optical
Module Structure and the Sunshields for both -M and -H.
The two Co-Principal Investigators are G. Piccioni (INAF-IASF Rome) and
P. Drossart (LESIA Paris). Forty Co-Investigators from ten countries are
participating in the data analysis.

2.3 The Optical Module structure

The OM’s main function is to provide mechanical support for the VIRTIS optical
subsystems while maintaining them at an operating temperature below 150K and
the two IR focal planes at 80K. To achieve this, the OM is divided into two
principal functional parts (Fig. 6): a warm enclosure and a cold enclosure (the
‘Cold Box’).
The warm enclosure includes the OM baseplate, the Inter-Unit Harness
brackets, the Active Cooling Subsystem (a pair of cryocoolers, one each for -M
and -H) and the structural truss that thermally insulates the upper Cold Box from
the baseplate. The Cold Box includes the two optical subsystems, mounted on a
cold ledge that also supports the cover mechanisms, the two baffles and the
passive cooling hardware.
The mechanical interface with the spacecraft is the OM baseplate connected to
a special supporting structure provided by the Venus Express spacecraft (not
present on Rosetta). A set of four thermal pipes in the structure is dedicated to
the thermal power dissipation of the VIRTIS cryocoolers. In this way, the power
generated on the baseplate, mostly from the cryocoolers, is conductively
dissipated to deep space through a dedicated radiator.
The cryocoolers’ cold-tips are conductively connected to the VIRTIS IR
detectors via flexible thermal straps.

7
SP-1295

The passive cooling to keep the -H and -M temperatures below 150K consists
of a primary radiator, conductively connected to the ledge, and two secondary
radiators, one each for -M and -H. The three radiators are placed on the
spacecraft’s –X-axis, looking at cold space; their overall envelope is about
590 x 650 mm.
The primary radiator refrigerates the ledge to slightly higher than the
instrument’s temperature, by dissipating the heat from the truss and from part of
the internal cabling. The secondary radiators are each mounted directly on top of
the instruments.
The Cold Box and the warm enclosure are externally covered with multi-layer
Fig. 7. VIRTIS-M during assembly. The grating insulation (MLI). An MLI blanket separates, from a radiative point of view, the
is in the centre, the FPAs are on the right-hand Cold Box from the warm enclosure.
side and the slit is on the left. The low-conducting heat load between the cold and warm enclosures is
achieved by mounting the Cold Box on eight titanium hollow struts. The number
and attachment positions were selected to avoid distortion upon cooling from
room temperature. All of the struts are blanketed with MLI to reduce the thermal
load and avoid differential heating of the cold ledge.
The heat flow through the Cold Box is small enough that the VIRTIS optical
subsystems are mounted inside an essentially isothermal cavity. To help maintain
this isothermal cavity, the covers and actuators for the -M and -H entrance ports
are inside the Cold Box, between the optical subsystems and the external MLI
blanket.

2.4 VIRTIS-M
VIRTIS-M is the mapping spectrometer of VIRTIS. It consists of the -M optical
head inside the OM and the PEM-M proximity electronics close to it but residing
in the spacecraft in a warm thermally controlled environment. The optical head
consists of the telescope and spectrometer that are shared by the visible and IR
channels and then naturally co-aligned into one unique optical axis (Fig. 7).

2.4.1 Telescope
The Shafer-type telescope is the combination of an inverted Burch telescope and
an Offner relay. The Offner relay takes the curved, anastigmatic virtual image of
the inverted telescope and makes it flat and real without losing the anastigmatic
quality. Coma optical aberration is eliminated by putting the aperture stop near
the centre of curvature of the primary mirror, thus making the telescope
monocentric. The result is a telescope system that relies only on spherical mirrors
yet remains diffraction-limited over an appreciable spectrum and all the vertical
field (slit direction).
The horizontal field is realised by rotating the mirror of the Shafer telescope
around an axis parallel to the Y-axis of the spacecraft. The optical system is not
diffraction-limited over the overall field but the optical quality is still sufficient.

2.4.2 Spectrometer
The imaging spectrometers of the Cassini (VIMS) and Mars Express (OMEGA)
Bibring et al., 2004) missions required separately housed co-aligned instruments
to cover the visible and IR spectra. The VIRTIS-M Offner spectrometer not only
does away with redundant optical systems, but also eliminates the need for
collimators, camera objectives and beam splitters, thereby simplifying fabrication
and minimising volume and mass. However, a grating spectrometer that does not
rely on a collimator and camera objective requires perfect matching with its
collecting telescope. Not only must they have matching F-numbers, but the
telescope must be telecentric or have its exit pupil positioned on the grating. The
Shafer telescope is matched to the Offner spectrometer because both are
telecentric. This means that the entrance pupil is positioned in the front focal
length of the optical system 750 mm in front of the primary mirror.

8
 scientific instruments

The VIRTIS-M spectrometer grating does away with the beam splitter by
realising two different groove densities (Piccioni et al., 2000) on a single substrate
(Fig. 8). Since the pupil optics conjugate is on the grating, the same spectral beam
splitting is performed for each FOV angle. The grating profiles are
holographically recorded into a photoresist coating and then etched with an ion
beam. Using various masks, the grating surface can be separated into different
zones with different groove densities and different groove depths. The ‘V’ regions
(inner parts), which make up the central 30% of the conjugate pupil area,
correspond to the higher groove density needed to generate the higher spectral
resolution required in the ‘visible’ channel extending from the UV to the near-IR. Fig. 8. The VIRTIS-M grating used for the
The smaller pupil area allows the visible channel to operate partially coherently development model. The two inner regions are
for the visible range, the external one is for IR.
and achieve a smaller point spread function.
The dispersion axis is roughly along the ruler.
The IR channel (outer part) has a pupil area equal to 70% of the total. Since The surface geometry is convex.
the IR channel does not require such high resolution as the visible channel, it was
decided to accept for this channel a lower Modulation Transfer Function (MTF),
caused by the visible zone’s obscuration of the IR pupil. In the visible channel,
the diffraction spot diameter is smaller than the pixel size, so the main factor for
the optical quality is the optical system aberrations. Conversely, in the IR
channel, the optical quality is affected mainly by the diffraction. Furthermore,
the central obscuration in the IR reduces the diffraction MTF by a further 10%.
A laminar grating with a rectangular groove profile is used for the visible
channel pupil zone to enable two different groove depths to alter the grating
efficiency spectrum and compensate for low solar energy and low CCD quantum
efficiency in the UV and near-IR regions. The resulting efficiency, improves the
instrument’s dynamic range by increasing the signal-to-noise ratio (SNR) at the
extreme wavelengths and preventing saturation in the central wavelengths.
To minimise the effects of background radiation emitted in the form of
photons with wavelengths longer than 4.2 μm, both IRFPAs have to be protected
by spectral blocking filters with cut-off wavelengths at 5.1 μm. In addition,
VIRTIS-M also requires prevention of the superimposition of higher diffraction
orders coming from the grating (VIS and IR channels), by inserting a dedicated
long-pass filter in the optical path.
Since the 4.2–5.1 μm background radiation cannot be filtered within the same
spectral band in which the data are collected, the spectrometer must be cooled to
below 150K.

2.4.3 PEM-M
The PEM-M, housed in the spacecraft, contains all the electronics needed to
interface the Main Electronics, to drive the FPAs, the scan mirror and the cover
mechanism and to perform the acquisition and conversion of the science and
housekeeping data.
The electronics package consists of a mother board and four daughter boards,
of which two are committed to the visible channel, one to the IR channel and one
to the scan and cover control units. Fig. 9 shows the PEM-M block diagram
together with the Optical Head devices.
The interface circuits with the Main Electronics are located on the mother
board and are implemented with monodirectional balanced interfaces.
The visible channel comprises four main blocks: data acquisition; analogue-to-
digital converter (ADC); the CCD bias and drivers; the FPGA.
The data-acquisition block is based on a low-noise amplifier chain containing
a correlated double-sampling circuit to eliminate every uncertainty caused by the
reset noise. The ADC (7805ALPRP) ensures a 16-bit resolution with a 10 μs
conversion time.
The CCD phase drivers translate the patterns generated by a dedicated ASIC
to MOS levels in order to drive correctly the large capacitive loads of the CCD
rows and columns with low power consumption.

9
SP-1295

Fig. 9. VIRTIS-M block diagram. The Optical

Head mechanisms and devices are on the left
side, while the electronics (PEM-M) are
represented on the right with their block
structure.

The main function of the FPGA (ACTEL RT14100) is to initialise the ASIC
timing generator, and to control the ME command interpreter, the data
transmission, the calibration source and annealing heaters.
From a functional point of view, the IR channel is very similar to the visible
channel. The IRFPA control unit and acquisition chain are simpler, with all the
electronics on one board. The IR FPGA (ACTEL RT14100) is dedicated to
controlling the shutter mechanism.
The acquisition, conversion and transmission of the VIRTIS-M housekeeping
data (temperature, voltage and current measurements) are divided between the
visible and IR channel.
The scan unit board performs accurate positioning of the primary mirror to
the commanded angle. The scanning movement is done step by step. Each step
moves a new portion of the field of view of the interferometer slit. The control is
accomplished using a two-phase DC brushless motor to provide the torque; the
current flowing in the stator windings produces a torque at the permanent
magnet rotor. The Inductosyn position transducer provides 10 arcsec accuracy. It
is essentially a variable coupling transformer, the magnitude of which varies
according to the position of the rotating element.
The control logic interface is based on an ACTEL FPGA (RT14100) receiving
scan unit commands via a synchronous serial link and calculating the position
error. This FPGA also contains the cover mechanism control logic, which
actuates the cover by a stepper motor. The end positions (open and closed) are
computed step-by-step starting from a mechanical reference (end-stop). In
addition, two Hall-effect sensors (HESs) monitor the position reached by the
cover after the opening or closing command. The Cover Mechanism Electronics
include two drivers for the windings of the stepper motor and the front end for
the two HESs. Two different supply lines close or open the cover, then power-off
the motor and continue to monitor the cover position by the HES outputs.

2.4.4 Focal Plane Array: visible detector

The visible FPA is based on the Thomson-CSF TH7896 CCD detector. It uses a

10
 scientific instruments

buried channel design and two poly-silicon N-MOS technology to achieve good
electro-optical performance. It includes a multi-pinned phase boron implant to
operate fully inverted and to reduce substantially the surface dark current,
residual images after strong exposure and other effects due to ionising radiation.
The TH7896M is a full-frame image sensor with 1024 x 1024 sensitive elements,
two registers and four outputs, but it is used as a frame-transfer device, shielding
half the sensitive area that works as a memory section. Only one horizontal
register and one output are actually used.
To obtain sensitivity to 250 nm, there is a UV-coating on columns 1–112. This
coating emits light at approximately 540–580 nm when excited with wavelengths
shorter than 450 nm. It is transparent in the visible and near-IR, and does not
significantly influence quantum efficiency in this portion of the spectrum.
In order to generate the required information (to meet the instantaneous FOV
of 250 μrad and to have the same pixel size in the IR and visible detectors), 2 x 2
binning is implemented at the detector level. This approach achieves the required
pixel size with a less expensive solution; a 38 μm pixel size is feasible but would
require a custom design.
Effects on charge transfer efficiency owing to radiation-induced dark current
spikes and residual bulk image are minimised when operating at very low
temperatures. Proton interactions with the silicon lattice produce deep-level traps
that begin to freeze out at 200K and become inactive at 160K. In fact, once filled
with electrons, these traps remain filled owing to the long emission time constant.
The challenge is then to keep the CCD temperature as close as possible to that of
the Cold Box while preventing degradation of the charge transfer efficiency
owing to buried channel freeze-out and bulk trap characteristics. The thermal-
mechanical design of the FPAs fulfils this goal.
The detector is also equipped with two ordering filters with a boundary at
band 215 (680 nm) in order to attenuate the higher orders of the visible grating
region.

2.4.5 Focal Plane Arrays: IR detectors

The IR detectors used in VIRTIS-H and VIRTIS-M are based on a 2-D array of
IR-sensitive photovoltaic mercury-cadmium-tellurium (HgCdTe). The growth
technique for the crystal is critical for high-performance FPAs because structural
defects in the material result in defective photodiodes. The preferred growth
technique for producing the best material, in terms of crystalline quality,
electrical uniformity and reproducibility, is the liquid or molecular epitaxy
process. These devices have the potential to operate at a higher temperature than
the more established indium antimonide (InSb) detectors owing to dark current
reduction by a factor of 10 or more.
The detector is manufactured by Raytheon Infrared Operations (Santa
Barbara, USA). The charge trans-impedance amplifier technology used on this
detector is the best solution when external low background flux (6 x 107 to
2 x 1011 photon cm–2 s–1) has to be detected, as is the case for Rosetta’s comet and
the Venus nightside.
The array is formed through hybridisation of mercury-cadmium-telluride
(MCT) material with dedicated Si CMOS. The device is a MCT-Si hybrid array
of 270(w) x 436(h) 38μm MCT photodiodes with line and column spacing of
38 μm between diode centres, a spectral wavelength range from 0.95 μm up to
5 μm and an operating temperature of 80K. The device has a flatpack
encapsulation, adapted to take a pressure-bonded window and with an internal
filter to separate the grating’s higher orders and to suppress background
radiation.
The detector is packaged into a housing which includes an optical window and
provides mechanical, thermal and electrical interfaces for its integration on the
VIRTIS-H and -M FPAs.

11
SP-1295

The VIRTIS-M filter window is used to stop the superimposition of higher

diffraction orders from the grating and also to eliminate/reduce the background
thermal radiation due to the temperature of the instrument housing. The window
is covered with anti-reflection coating optimised in the 1–5.1 μm region. Five
segment filters are coated on to the window with the following bandpasses:

— 1st region: 0.9–1.6 μm;

— 2nd region: 1.2–2.5 μm;
— 3rd region: 2.4–3.75 μm;
— 4th region: 3.6–4.4 μm;
— 5th region: 4.3–5.0 μm.

2.4.6 VIRTIS-M modes

The possible modes for VIRTIS-M are summarised in Table 2. The overall data
rate and volume depend essentially on the selected modes for -M because -H does
not much affect the rate (unless special testing modes are used, involving the
transfer of the full frame).

2.5 VIRTIS-H
The VIRTIS-H optical design is based on a cross-dispersion concept (Fig. 10). Its
beam etendue is 0.8 x 109 m2 sr for an instantaneous FOV of 0.58 x 1.74 mrad. A
mean resolution of λ/Δλ = 2000 in the spectral region 2–5 μm gives about 3500
spectral elements. The 32 mm entrance pupil is placed near the F/1.6 off-axis
parabolic telescope. The magnification of the spectrometer (between the slit and
the detector) is 1.3, so the slit image covers 3 pixels, although 5 pixels are summed
per spectral element to avoid any signal loss.
To reach the SNR specifications, the spectrometer is cooled below 150K and
the detector is cooled to 80K.
VIRTIS-H consists of four main parts: an afocal module, the dispersion
module, an objective and a focal plane assembly. The afocal module is composed
of two aluminium off-axis parabolas (telescope and collimator) with the same
focal plane in which the entrance slit is placed. A shutter near the entrance slit
allows background measurements. This assembly is mounted on a bench made
with the same material as the mirrors; this makes a compact low-straylight
achromatic system that can be adjusted in the visible at room temperature.
The dispersion module consists of a prism and a reflection grating. Eight
orders of the grating (6–13), perpendicularly separated by a magnesium oxide
prism, are used (Table 3). The grating is split into two differently blazed parts to
allow a smoother efficiency profile along each order.

Table 2. VIRTIS-M modes.

Number of pixels Binning

spectral spatial spectral spatial Detectors

NOMINAL_3X4_FULL_WIN 144 64 3 4 VIS + IR

VIS_ONLY_1X4 288 64 1 4 VIS
IR_ONLY1X4 288 64 1 4 IR
HIGH_SPECTRAL1X4_FULL_WIN 432 64 1 4 VIS+IR
HIGH_SPATIAL_3X1_FULL_WIN 144 256 3 1 VIS+IR
ALL_PIX_FULL_WIN 432 256 1 1 VIS+IR
REDUCED_SLIT_3X1 144 64 3 1 VIS+IR
ALTER_IR_ONLY_1X4 288 64 1 4 IR

12
 scientific instruments

Fig. 10. A simplified view of the VIRTIS-H

optical scheme.

The F/2 objective is composed of two silicon and one germanium lenses. The
entrance face of the first Si lens is elliptical (e = –0.1171). As it is close to the
grating (the instrument white pupil), the objective spherical aberration is
drastically reduced. The position of the white pupil makes the objective and the
system strongly telecentric.
The FPA includes a two-zone filter that isolates the lower order (4.0–5.0 μm)
from the others. The main goal of this filter is to remove the thermal background
in the shorter wavelengths, thereby increasing the SNR in this spectral region.
The overall FPA can be adjusted in tilt and focus with respect to the objective.
Since the 4.0–5.0 μm background radiation cannot be filtered within the same
spectral band in which the data are collected, the spectrometer must be cooled to
below 150K to cope with the low signal levels.
The straylight is reduced via several baffles and black-anodised mechanical
mounts in the spectrometer. A baffle is placed between the spectrometer entrance
and the telescope, and another one near the entrance slit. The objective and the
FPA are linked using baffles.
The instrument is optically adjusted using hard wedges, which makes it very
stable. First, the afocal whole (telescope-slit-collimator) was adjusted at room
temperature using wedges on the telescope and the collimator; the material
uniformity means there is no misalignment at the working temperature of this
system. The IR adjustment at the working temperature of the overall instrument Table 3. VIRTIS-H grating orders.
was made using wedges on the FPA and the grating.
The optical quality of the instrument is very good because it is strongly Order (μm) λ min λ max
diffraction-limited in the overall spectral band. Owing to the presence of a
Order 0 4.00342 4.98660
grating, the spectral resolution (λ/Δλ) varies in each order between 1500 and
Order 1 3.43527 4.28712
2900.
VIRTIS-H uses the same IR detector as VIRTIS-M but, owing to the different Order 2 3.00538 3.75714
designs of the two channels, the detector is used rather differently. VIRTIS-H is Order 3 2.67117 3.34079
a high-resolution spectrometer and does not perform imaging. Its IR detector Order 4 2.40335 3.00673
acquires spectra spread over its surface, so only a portion of the pixels contains Order 5 2.18426 2.73313
useful scientific data. The eight spectral orders are spread over the entire surface Order 6 2.00126 2.50554
matrix. In each spectral order the spectrum covers 432 x 5 pixels (where 5 pixels
Order 7 1.84486 2.31273
represent the image of the slit size when imaged on the detector).

13
SP-1295

Fig. 11. The PEM-H block diagram.

Thus overall only 15% of the 438 x 270-pixel matrix surface is used. To reduce
the overall data rate and volume, VIRTIS-H uses pixel mapping to give the exact
location of the spectra over the IR detector. The ME calculates the location of
the pixels to be downloaded and passes it to PEM-H, which then downloads
them accordingly. The downloaded data are the -H spectra, a set of 432 x 8 x 5.
As VIRTIS-H has no spatial resolution, the 5 pixels are averaged together, so the
end-product in the nominal acquisition mode is a 3456- (or 432 x 8)-pixel
spectrum representing the full spectral range of the instrument from 2 μm to
5 μm.

2.5.1 PEM-H
PEM-H, housed in the warm spacecraft environment, contains all the electronics
needed to interface the Main Electronics, to drive the FPAs, the shutter and the
cover mechanisms and to perform the acquisition and conversion of the science
and housekeeping data. The electronics package consists of a mother board and
four daughter boards: CME board for the cover control; FPGA board; ADC
board; video board. Fig. 11 shows the PEM-H block diagram.
The VIRTIS-H detector has a CMOS-based readout multiplexer. It needs
only two Transistor-Transistor Logic (TTL) clocks (main clock and integrate
pulse), two power supplies (5 V) and two biases. The Video Amplification and
Filtering module amplifies and filters the video signal and sends it to the
analogue-digital conversion stage. The Analogue Housekeeping Handling

14
 scientific instruments

processes the analogue housekeeping signals (temperatures, voltages and

currents), and multiplexes and sends them to the further analogue-digital
conversion. This conversion multiplexes the analogue signals (video &
housekeeping) and converts them into 16-bit serial data.
The PEM Controller, implemented in an FPGA (ACTEL RT14100), handles
the Data Handling and Support Unit (DHSU) commands, the PEM and
detector control signals, the PEM and detector clocks and all the input/outputs.
The Command Interpreter in the FPGA translates the DHSU commands into
PEM commands for further processing.
The PEM Parameters Memory stores the current PEM configuration
parameters in the FPGA. Some of these (command bits) are sent directly to the Fig. 12. The VIRTIS electronics boxes shortly
other functions by the PEM Parameters Memory itself. before integration at IAS Orsay, Paris, during
the calibration campaign at the beginning of
The Detector and ADC Controller inside the FPGA generates the detector
2004. The main electronics are at top, PEM-H
clocks and synchronises accordingly the detector acquisitions by the ADC. The is at left and PEM-M is at right.
acquisitions are controlled by the Pixel Map, which contains the image of the
detector pixels and is used to control the pixel acquisition (only the illuminated
pixels are converted). It is written into by the Command Interpreter, controls the
Detector & ADC Controller and can be read by the Output Data Formatting.
The Output Data Formatting handles the science and analogue housekeeping
data (generated by the ADC) and the digital housekeeping data (assigned values
and status), formats them and generates the Data Blocks.
The Detector Biases includes four detector bias sources controlled by four
DACs. The Calibration Lamps Controller powers the three calibration lamps
via three programmable current sources (three DAC channels). A Shutter
Controller controls and adjusts shutter closing by means of a programmable
current source (one DAC channel). A Cover Controller (CME) controls the
cover device and handles the cover status data. Finally, the DHSU Interface
handles the commands and data exchanges with the DHSU: unidirectional
control signals and serial synchronous data (RS422), PEM synchronisation
clock at 8 MHz and Reset control signal. The electronics boxes are shown in
Fig. 12.

2.5.2 VIRTIS-H IR detector

The VIRTIS-H IR detector is identical to the -M detector but has different
filters. The filter window eliminates/reduces the background thermal radiation
owing to the temperature of the instrument housing.
Its transmittance characteristics are optimised for each corresponding
detector position. The window has an anti-reflection coating optimised in the
2–5 μm region. Two segment filters are coated on to the window with the
following passbands:

— 1st region: 2.0–4.4 μm;

— 2nd region: 3.9–5.0 μm.

2.5.3 VIRTIS-H modes

The possible modes for VIRTIS-H are:

H_Nominal_Observation. The Pixel Map is used to select the region of the IR

detector to be downloaded to -ME.
H_Calibration. Start of a calibration sequence.
H_Nominal_Simulation. Data structure is as for H_Nominal_Observation but
the data content is simulated by the FPGA.
H_Science_Backup. Full window reading of the IR detector (270 x 438 pixels)
H_Spectral_Calibration_Simulation. An H_Image_Slice (same size as
H_Science_Backup but only simulated data) is delivered to the ME.
H_Test. Test mode to command directly the PEM-H without any ME filter.

15
SP-1295

2.5.4 Data reduction tool for VIRTIS-H analysis

A software web tool was developed to provide a quick-look facility far all
Co-Investigators. The tool can:

— change the displayed frame (if the Planetary Data System file contains several);
— change the contrast and the colouring scheme;
— draw horizontal and vertical profiles by clocking on the picture;
— compute the spectra for selected orders; if necessary, it can change the pixel
map used.

If a more sophisticated treatment is needed, the data must be downloaded to

a dedicated machine. A ‘VIRTIS Virtual Instrument’ tool was built to simulate
observations for a source whose flux is described by the user. The goal was to help
the Co-Investigators to prepare the observations to be performed with the real
VIRTIS instruments.

2.6 Main Electronics

The Main Electronics are physically separated from the Optics Module and the
Proximity Electronic Modules (Fig. 12).
The ME consists of two Data Processing Units (DPUs), including the
spacecraft interface electronics and the related power supply units, the power
supply for VIRTIS-M/H, and the VIRTIS-M/H interface electronics (M/H IFE),
which include the respective Cooler Controller Electronics (CCE) and Cover and
cooler Switching Electronics (CSE) and the M/H reference clock generator
(MCLK/RCLK). The DPUs, spacecraft interfaces and VIRTIS-M/H interfaces
are also known as the Data Handling and Support Unit.
There is a cold-redundant DPU, with the respective redundant power supply
unit and redundant common clock generators.
To power the VIRTIS-M/H subsystems, there is only one supply unit. It is not
redundant owing to the logical redundancy of the two VIRTIS-M and -H
channels (heritage from Rosetta). If the main DPU or main power supply unit
(main channel) fails, the redundant channel (redundant DPU and power supply
unit) can be switched on by the spacecraft redundant power interface.
The cross-strapping between the two VIRTIS-M and -H subsystems and the
main or redundant channel (power supply and DPU) is implemented in the ME
M-IFE and H-IFE. An additional cross-strapping is implemented to provide the
+28 V input power for the VIRTIS-M/H power converter. The VIRTIS-M/H
interface electronics (M/H-IFE) are supplied by the main or redundant DPU
power. The resulting power is distributed by respective switches in order to turn
on or off the VIRTIS-M/H IFE separately.
The VIRTIS subsystems (VIRTIS-M, -H, coolers and covers) are switched
on/off via telecommands and controlled by the DHSU.
For controlling the two coolers (inside the OM), two electrically independent
CCEs are located in the ME box and can be switched on or off by the respective
+28 V CSE and controlled by the DPU via the M/H-IFE.
The DHSU provides the platform for the data processing/handling and the
instrument-control functions, which are performed by software. The main tasks
of the DHSU are:

— acquisition, pre-processing, compression and formatting of the science and

calibration data of VIRTIS-M and -H subsystems;
— control and power switching of VIRTIS-M, -H (including the
electromechanical devices, the scan mirror for -M and shutters), the coolers
and covers;
— health check of the instrument and the operational status of VIRTIS to the
spacecraft;

16
 scientific instruments

— execution of up/downlink activities to and from VIRTIS;

— interpretation, execution and acknowledgement of telecommands;
— management and synchronisation of activities between VIRTIS and the
spacecraft.

The data-processing and the data-handling activities by the DHSU are

performed on-line. The DHSU communicates with VIRTIS-M, VIRTIS-H and
the spacecraft via serial interfaces and an ME common bus, which is separated
from the local bus of the DPU.
Each DPU has a high-speed data interface to the Solid-State Mass Memory
(SSMM), a low-speed telemetry interface (TM I/F) and low-speed telecommand
interface (TC I/F) to the Remote Terminal Unit (RTU).
All interfaces are handled by interrupt, in order to decouple the asynchronous
processes between the DPU and the interfaces and to put the DPU in a power-
down mode while not operating.
Data processing and handling are sub-slice-oriented (a sub-slice is a 64 pixel x
144 or 64 x 72 spectels). The 16-bit organised data from VIRTIS-M and -H are
transferred via serial interfaces, FIFO (First-In, First-Out) buffer and common
bus into the local data memory of the DHSU. This data acquisition process is
performed in the background, while the processor completes the processing of
data from previous exposures. The DPU performs the data processing, including
data compression and telemetry packing. The steps and the kind of data
processing and instrument control are defined by special operation modes
selected by telecommands.
After processing and compression, the scientific and housekeeping data are
formatted and transferred to the FIFO buffer of the spacecraft data interface.

Fig. 13. VIRTIS mode transitions. Once the

primary boot is executed, VIRTIS enters safe
mode, where only a few commands can be
executed. One of these is the ‘enter idle’ mode,
from where it can be operated in full.
TC: telecommand. VTC: VIRTIS telecommand.

17
SP-1295

The interface hardware controls the transfer of the serial block data to the
spacecraft (SSMM or RTU).
The DHSU also controls and monitors the status of the different subsystems,
including the DHSU itself. The operational housekeeping data are periodically
acquired and transferred via the spacecraft low-speed telemetry interface.
A low-power and radiation-hardened Digital Signal Processor (DSP, TSC
21020) is used for the DPU with a computing power of about 20 MIPS.
The DPU memory is divided into three sections: program code; program data;
and image data. The program memory consists of the PROMs, EEPROMs and
SRAMs. The program code for execution of the Safe Mode is stored in the
radiation-hardened PROM and is downloaded by primary boot after power-on
to the fast and low-power program SRAM, also radiation-hardened (Fig. 13).
The EEPROM stores the main part of the application software and parameters.
The EEPROM content and the Program Memory RAM content can be uploaded
by telecommand and downloaded by sending telemetry packets. The validity of the
EEPROM content is verifiable by checksum, which is read before secondary boot.
The checksum and other status data are stored three times in physically
independent configuration areas of the EEPROM. An image data memory stores
the raw science data and the intermediate products. The architecture of the DPU
has a watch-dog to detect program lockups and endless loops.

2.6.1 DPU
The core of the DPU is the DSP TSC21020E. It is a high-reliability version of the
ADSP21020 and is functionally compatible with this DSP. The DPU
communicates with the VIRTIS-M and VIRTIS-H subsystems by serial
command/data interfaces. It provides the IEEE-1355 high-speed science data
interface (HS link) to the spacecraft SSMM, the low-speed data (SDT) interface
for the transfer of housekeeping and the Memory Load Command (MLC)
interface to the RTU for VIRTIS telecommanding by the spacecraft or to the
EGSE/Spacecraft Interface Simulator (SIS) for on-ground tests. Internally, the DPU
controls all other ME units, using separate low-speed serial or parallel interfaces.
The main functionality of the DPU is concentrated around the DSP, which
provides computational resources and control capabilities. The DSP operates at
20 MHz, yielding 40 MFLOPs (typical) to 60 MFLOPs (peak) processing power.
A further key function is the high-speed serial link communication, which is
realised by an IEEE-1355 ASIC (SMCS332). The SMCS332 has three IEEE-1355
full-duplex links (two are used for VIRTIS) that execute data transfers with low
Central Processing Unit intervention. Each of the links supports high-level
protocol-handling running full-duplex at up to 200 Mbit/s. For VIRTIS only two
links with a maximum speed of 10 Mbit/s are used, one for the transfer of
scientific data to the SSMM, the other for interacting with a controlling Host PC
for development and verification purposes.
The DPU local memory concept is based on a fast static RAM. It provides
128 kword (48-bit) program memory, for 512 kword (40-bit) data memory and
8 kword (16-bit) dual-ported communication memory. The communication
memory provides bidirectional buffering between the local processor and the
SMCS332. It decouples the different speed characteristics of the DSP and the
communication link.
The DPU Extension Board is an extension unit for the DPU that contains only
VIRTIS-specific functionalities. These are the On-Board Data Handling
(OBDH) interfaces, a data memory extension, an EEPROM bank, internal
interfaces to other ME units and additional control logics located in an ACTEL
FPGA. Reading the status of the DPU Extension Board, as well as writing the
configuration of the local resources, is managed by simple memory accesses of
the TSC21020E.
A 2 M x 16-bit image memory extension is used for intermediate buffering of

18
 scientific instruments

the spectral data before compression. A 1024 K x 8-bit EEPROM memory

provides non-volatile storage of program code and mode parameters. The
EEPROMs are only switched-on/off during the secondary boot load of
EEPROM content into the Program Memory in order to minimise radiation
influences during VIRTIS mission operation. The necessary separation of the
EEPROM lines from the DPU busses is realised by additional bus drivers in order
to avoid power loading conflicts.
Serial 16-bit unidirectional interfaces to the spacecraft (MLC and SDT) are
used for telecommand and telemetry transmission from and to the spacecraft.
TM and TC data are buffered by two separate 8 K x 9-bit FIFOs.

2.6.2 Software
The PROM software is hard-coded in the Main Electronics/DPU and the
EEPROM software is changeable by memory upload (memory management
service). The PROM software is written in Assembler, it has a size of 5139 48-bit
instructions and consists of 66 modules.
After Primary Boot the PROM software runs using only the Program Memory
for safety reasons. Fig. 13 shows a state block diagram.
Primary Boot is performed by the DPU Board and Boot Controller (BBC)
after +28 V power-on. The PM has a very good Single Event Upset (SEU)
performance (almost SEU-free) while the Data Memory (DM) is SEU-sensitive.
Safe behaviour of all PROM software functions can therefore be assumed.
The EEPROM software is stored in EEPROM as PM and DM segments with
segment checksum for verification during upload and start (i.e. Secondary Boot
from EEPROM in RAM). It runs in PM and DM RAM. The EEPROM software
is mainly written in C with low-level functions in Assembler for speed and code
optimisation. The VIRTUOSO Real Time Operating System (RTOS) is used and
51 processes can be active simultaneously.
The Secondary Boot software consists of about 200 functions, excluding
VIRTUOSO functions/library (VIRTUOSO V4.1 R2.05 is used). The size of the
secondary boot software is about 83 Kwords with 68 K Instructions and 15 K
Program data/parameter. The compressed code stored in EEPROM is 354 Kbyte
(about 60 Kwords). This means that two executables can be stored in EEPROM
from the size point of view. Generally, about 170 K of Instructions can be stored
in EEPROM at maximum and up to eight different executables.

2.6.3 Data compression

Data compression is done for reducing the telemetry data volume either by
lossless or lossy compression. A compression unit is always a 144 x 64-element
sub-slice. The compression is applicable for VIRTIS-M IR, -M VIS and -H using
the same algorithms, depending on the operational parameters.
The reversible algorithm (lossless) is derived from that developed for the
OMEGA/Mars Express and VIMS/Cassini imaging spectrometers. It takes
advantage of spatial/spectral correlations for pre-processing, then a Rice coding
is applied on the residuals. It was mainly used for validation purposes early in the
mission and it provides typical compression ratios of 2–4 depending on the
entropy content of the data.
The lossy algorithm is based on wavelet transforms. The following compression
modes with related factors are applicable:

— lossless compression is data compression without any loss. A differential

algorithm is performed with Rice encoding. The compression factor is low
(about 1.5) for very noisy M-PEM IR data up to high (about 14) for synthetic
IR ramp data. The nominal average compression factor is about 2–4.
— wavelet F1 compression is ‘low’ lossy. The data compression factor is 8
(2-bit/16-bit).

19
SP-1295

— wavelet F2 compression is ‘medium’ lossy. The data compression factor is

10.67 (1.5-bit/16-bit).
— wavelet F3 compression is ‘high’ lossy. The data compression factor is 16
(1-bit/16-bit).

2.7 Instrument budgets

The total mass of the instrument including internal harness is 33 kg. Total power
during science operations is about 50 W. During the cool-down of the IR
detectors a long peak of 70 W can be reached even though it is strongly
dependent on the thermal environment conditions of the Cold Boxes and
baseplate.
The data-rate range is obviously very wide, depending on which mode is being
used for observations. As an example, a typical rate of 600 kbit/s is generated for
the high resolution when a lossless compression is used, considering an average
compression factor of 3 and a repetition time of 2.5 s. The data rate engages
mostly the SSMM, while the RTU is used only for the science housekeeping.
In case of failures on both fast interface channels, science data can be
transmitted in a degraded mode via the slower OBDH interface.

3. Calibration An intense calibration campaign was performed early on at the subsystem level
for VIRTIS-M and -H (Drossart et al., 2004) and then at the system level at the
large facility in IAS Orsay (F). VIRTIS carries several internal calibration devices
in order to check the spectroscopic and radiometric parameters from time to time
during the mission.

3.1 Ground calibration

All the FM detectors were tested and calibrated separately in an ad hoc facility in
advance of installation (Fig. 14). The main parameters were measured at 60K,
70K, 80K, and 100K, and different wavelengths when relevant:

— dark current without spectrometer background;

— charge response (linearity);
— photon flux response (linearity);

Fig. 14. The VIRTIS-H detector on the

calibration bench before integration.

20
 scientific instruments

Fig. 16. VIRTIS-H spatial profile along the spatial direction (parallel to the
-M slit). The FWHM of IFOV along the slit is 1.34 mrad (about 5–6 high-
Fig. 15. Gas-cell transmission measured with VIRTIS-H (black curve) resolution pixels of -M). In the cross-slit direction, the FWHM is about
compared with the same cell measured by a laboratory FTS (red curve). 0.44 mrad (about 2 high-resolution pixels of -M).

— flat fields;
— readout noise;
— blooming effects.

A dedicated irradiation test was performed on a test-detector to verify its

behaviour over the required life. The conditions were:

— proton flux at 180K at: 20 MeV, 30 MeV, 40 MeV and 50 MeV;

— bombardments up to 2 x 1010 particles/cm2 on a single chip detector.

The dark current showed a variation of less than 1% after the irradiation tests,
illustrating the detector’s robustness.
VIRTIS underwent several calibration sessions, performed with different
Cold-Box temperatures, devoted to spectral and spatial calibration, radiometric
calibration, geometric calibration, internal calibration and co-alignment
measurements, among others.
A monochromator was used to measure the spectral registration and
calibration. For VIRTIS-H, an average of 10 evenly-distributed wavelengths was
recorded per grating in order to interpolate the spectral resolution. Owing to the
use of the grating’s high orders, the spectral resolution varies significantly along
the orders. The specification of an average resolution of 2000 around 3.3 μm was
achieved. CO2 and CH4 gas cells and mineral samples were used for spectral
registration of real targets for both VIRTIS-M and -H. Fig. 15 shows the
spectrum of the gas cell measured with VIRTIS-H compared with that measured
on a laboratory Fourier Transform Spectrometer.
For the VIRTIS-M spectral profile measurement, a scan about the centre of
the slit was performed by a monochromator using a wavelength step of the order
of 1/5 of a spectel. The resulting Full Width at Half Maximum (FWHM) is 2 nm
for visible and 11 nm for IR.
To measure the spatial IFOV, the slit was scanned in the spatial direction with
a pinhole. Fig. 16 shows the FWHM and expected width for VIRTIS-H.
Considering a pupil diameter of 32 mm, the etendue is ~0.6 x 10–9 m2 sr at
FWHM of the FOV and 0.8 x 10–9 m2 sr at 18% of the full maximum of the FOV.
At FWHM, the VIRTIS-H IFOV along the slit is 1.34 mrad.

21
SP-1295

Fig. 17. The NESR of VIRTIS-M visible (blue) compared with a typical Fig. 18. The NESR of VIRTIS-M IR (blue, lower curve) measured at 135K
modelled radiance of the Venus dayside (red) with nadir-viewing. The for the Cold Boxes, compared with the modelled weak radiance from the
NESR is obviously well below the bright radiance from the dayside, so there Venus nightside (red, upper curve) with nadir-viewing. The NESR is well
is a good SNR for short exposures of a fraction of a second. The ratio is below the expected radiance, thus VIRTIS is able to detect most of the
essentially limited by the dynamic noise of the analogue-digital converter. radiance escaping from the planet. The thermal background is not
considered here; it depends strongly on the temperature of the Cold Boxes.

The VIRTIS-M spatial profile is bi-dimensional, one direction being along the
slit (the ‘pixel’ function) and the other across the slit (the ‘slit’ function). The
visible and IR FWHMs are similar and close to the spatial sampling of 250 μrad.
The -M visible slit function is also similar to the nominal spatial sampling along
the scan direction (250 μrad), while the IR equivalent is slightly worse by a factor
of two.
To measure the VIRTIS-H defocusing and image quality, a pinhole was used
at several wavelengths with a Hartmann mask near the entrance pupil. For each
position of the mask, an image was recorded and the position of the impact
estimated by gaussian fit. The set of positions allowed an estimate of the
wavefront to be reconstructed. At 3.4 μm the measured wavefront error is about
λ/5RMS, showing that the spectral resolution is not reduced by the image
quality.
From a radiometric point of view, a blackbody at different temperatures was
used to estimate the sensitivity of VIRTIS in terms of flux. The setup was placed
in a neutral nitrogen environment to remove atmospheric absorptions. Images
were recorded and the spectrum reconstructed for each blackbody temperature
(600, 460, 455, 451, 450ºC). The linearity of the response at 4.4 μm allowed the
temperature variations to be measured to better than half a degree. The
maximum relative deviation to linearity is less than 0.06%.
For the VIRTIS-M radiometric performance and Noise Equivalent Spectral
Radiance (NESR), see Figs. 17 and 18 and Section 4.

3.2 In-flight calibration

VIRTIS-M uses two custom lamps optimised for the respective visible and IR
channel (Melchiorri et al., 2003). Nevertheless, the lamps can be seen by both
channels, providing redundancy. The lamp illuminates the internal part of the -M

22
 scientific instruments

cover, designed to reflect the radiance diffusely back towards the instrument. The
cover has two areas of different materials chosen to have higher spectral
reflectance for the lamp and the related channel. The areas correspond roughly to
the visible and IR part of the instrument pupil. This choice comes from a trade-
off between the mechanical constraints and the need for the best calibration. The
lamp’s spot size is equal to the pupil dimension and also spatially homogeneous
at the distance of the cover.
The lamps are highly stable and an intense qualification campaign was carried
out on the qualification lamps while checking the limited degradation with time.
Two different filters in front of the lamp window checked the spectroscopic
performances: an Holmium filter for the visible lamp and a polystyrene filter for
the IR.
For VIRTIS-H, two spectral and one radiometric internal calibration modules
are used in the spectrometer. The spectral calibration module uses a tungsten
lamp and a temperature-calibrated Fabry-Pérot that gives 20–10 lines depending
on the grating order. One of the modules is used through the entrance cover
mirror that returns the flux to the telescope and slit. The other is used through
the back of the slit that returns the flux to the collimator. The radiometric
calibration module uses a tungsten lamp through the entrance cover mirror.
This Fabry-Pérot system allows the spectral registration of the instrument to
be confirmed after a shift is suspected, for example, following the vibrations
induced by launch. The absolute position of each known spectral line on the
detector is gauss-fitted for each order and used to reconstruct the spectral
registration of the instrument to a 2-degree polynomial function. Those functions
are then used to give the correspondence between the pixel number and the
wavelength for each pixel.

The predicted radiometric performance at Venus is good enough to work with 4. Expected Performance
short exposure times (and thus high repetition time). This achieves the scientific
objectives not only for the dayside but also for the nightside, where the radiance
is very weak. From the expected DN rate for the VIRTIS-M visible channel when
looking at the dayside using nadir geometry, the maximum exposure time to
avoid saturation is 0.36 s.
For the nightside, the predominant factor is the lack of photons. From the
expected DN rate for VIRTIS-M IR over the nightside, the maximum usable
exposure time is 4.5 s.
The actual maximum integration time will be limited by the background
(depending on the temperature of the Cold Box) rather than the flux from Venus.
The typical value for the exposure is significantly less than 4.5 s. A typical
exposure time of 1 s for nadir-viewing is used for -M in order to avoid saturation
at wavelengths longer than 4.2 μm. The temperature of the Cold Box may in this
sense dramatically affect the thermal background, especially at the longest
wavelengths. The saturation of the thermal region and thus a larger integration
time is acceptable when the scientific objectives from time to time do not require
the wavelength region above about 4.2 μm.
The NESR is a very useful figure for predicting the SNR during the
observations. The NESR is given by the ratio of the noise sigma (referred to 1 s
exposure time) and the Instrument Transfer Function (responsivity). An
equivalent source having a radiance equal to the NESR gives a signal equal to the
noise; in other words, an SNR of 1. Fig. 17 shows the NESR of the VIRTIS-M
visible channel compared with a typical modelled dayside radiance with nadir
geometry. Obviously, there is no SNR problem for the dayside, as
the NESR is well below the radiance. This is also true for VIRTIS-H and -M IR.
Fig. 18 shows the NESR and the expected radiance for VIRTIS-M IR (and
similarly VIRTIS-H) using nadir geometry over the nightside. The NESR is well

23
SP-1295

Fig. 19. Example of a true-colour

reconstruction. At left is the (rather poor) target,
acquired by a computer scanner. At right is the
resulting image from the VIRTIS-M visible raw
data without any correction. Only three channels
(RGB) were used from the 432 bands.

below the expected radiance, so VIRTIS is able to detect most of the weak
radiance escaping from the planet.
The ground-spot size is small enough for all the scientific objectives, even at
apocentre (66 000 km altitude), where it is about 15 km in high-resolution for
VIRTIS-M and about 80 km for -H. The minimum and typical repetition time for
all the observations is 2.5 s.
Another important task for VIRTIS-M visible is providing true-colour images
for public outreach. During the subsystem calibration campaign, VIRTIS-M
visible (and IR, although not shown here) imaged a target placed in the focal
plane of the set-up collimator to simulate an image at infinite distance. Fig. 19
shows the result, comparing the original (rather poor) picture and the raw image
from VIRTIS. The image uses only raw data; there is no correction for the
illuminating lamp, no flat-field correction and only three channels (RGB) were
used out of the 432 bands.

5. Observations The following observation cases are being used by VIRTIS at Venus (not in order
of priority):

Case 1 (pericentre observations)

These observations are usually made for statistical coverage of different regions
of Venus, night and day. The use of the scanning mirror (in fixed or scanning
mode) improves the statistical coverage, but image reconstruction is difficult
owing to the high speed of the spacecraft (Fig. 20). The pixel size is defined as the
linear dimension of the VIRTIS-M high-resolution spatial pixel (250 μrad in
angle) on the surface of Venus at the centre of the slit. The dwell time is typically
less than the minimum repetition time, which means that image reconstruction is
not possible, or the pixels are elongated in the direction of motion or there are
‘holes’ when the -M scanning mirror is used.
Observing the same region at a variety of phase angles (‘feature tracking’) is
essential for characterising the phase function of the scattering particles (H2SO4).
The knowledge of the phase function is important for accurate modelling of the
scattering in radiative transfer retrievals. It is also provides information on the
mean particle size.

Case 2 (off-pericentre observations)

These observations allow VIRTIS to perform spectral mapping (for -M) at the
best spatial resolution, with image cube reconstruction obtained by internal
mirror motion synchronised with orbital motion (an altitude higher than

24
 scientific instruments

Fig. 20. The main observation parameters for

VIRTIS in science Case 1. The colour of each
curve and the axis to which it refers are the
same. The abscissa is the time along the orbit in
hours; time 0 is the pericentre pass of the
spacecraft. The plotted parameters are the
absolute speed of the spacecraft and its
components: along the direction of the
observation (component parallel to the optical
axis of VIRTIS) and orthogonal to the direction
of the observation (perpendicular to the optical
axis of VIRTIS). Groundspeed is the
spacecraft’s relative speed over the surface.

12 000 km is needed). This mode is being intensively used, as most of the

scientific objectives related to combined spectroscopic/mapping are obtained in
this phase of the orbit.

Case 3 (apocentre observation)

This mode allows VIRTIS to observe the full disc of the southern hemisphere,
through off-nadir pointing in a mosaic of 3 x 3 images, as shown in Fig. 21. This
mode is essential for studying the dynamics for global meteorology; repetition
during several orbits provides a full atmospheric rotation. It is also used to map
the temperature field, cloud properties, winds and atmospheric chemistry, and it
is important for public outreach images.

Case 5 (stellar occultation/3-axis stabilisation) Fig. 21. Spacecraft repointing strategy with
This mode is used in parallel with the SPICAV instrument, which is the driver for numbered sequence for VIRTIS in science case 3
(global mosaic from apocentre). The dark side of
this case. VIRTIS is able to retrieve upper atmospheric structure and composition
Venus is at top and the radiator (–X) is pointed
by studying spectral absorption of stellar flux. In particular, minor constituents to shadow. The Y axis is parallel to the solar
can be detected, and the upper atmosphere structure is being deciphered from panels. The ellipse at the centre arises from the
CO2 absorption. This case is naturally combined from a geometrical point of fact that the ground track at apocentre is at
view with case 7. about 80ºS and not 90ºS according to the actual
orbital parameters. VIRTIS can observe the
dayside in the visible, and both sides in the IR
Case 7 (limb sounding) using two exposure times.
The instrument’s unique capability of imaging in the 4.3 μm CO2 band provides
powerful sounding of the mesospheric emissions. Mars Express/OMEGA
observations of the Martian limb have provided interesting observations, which
VIRTIS is complementing at Venus in a similar observing mode.
Limb observations provide vertical resolution and allow higher altitudes to be
probed owing to longer atmospheric paths (at and above cloud tops) up to
115 km, thus reaching the thermosphere (‘cryosphere’ in this case) where LTE no
longer applies.

High priority is given to selected surface targets; the planet’s slow rotation
offers limited opportunities for each target. Even though nadir or near-nadir
viewing (with some off-track or across-track angles) is being used most, feature-

25
SP-1295

tracking pointing is helping greatly to study the radiance for different phase
angles, especially around pericentre, and thus to infer the radiative transfer
properties more extensively.

6. Conclusions Venus Express, ESA’s second planetary mission, will not solve all the mysteries
behind the dense atmosphere. Nonetheless, it is a major step in the long-term
systematic exploration of Venus. VIRTIS is playing a significant role in ensuring
a high science return from the mission. It is highly desirable that the results from
Venus Express triggers a new wave of exploration of our sister planet.

References Allen, D.A. & Crawford, J.W. (1984). Cloud Structure on the Dark Side of Venus.
Nature 307, 222–224.
Artru, J., Lognonné, P. & Blanc, E. (2001). Normal Modes Modeling of Post-
Seismic Ionospheric Oscillations. Geophys. Res. Letters, 28, 697.
Baines, K.H., Bellucci, G., Bibring, J.-P., Brown, R.H., Buratti, B.J.,
Bussoletti, E., Capaccioni, F., Cerroni, P., Clark, R.N., Coradini, A.,
Cruikshank, D.P., Drossart, P., Formisano, V., Jaumann, R., Langevin, Y.,
Matson, D.L., McCord, T.B., Mennella, V., Nelson, R.M., Nicholson, P.D.,
Sicardy, B., Sotin, C., Hansen, G.B., Ajello, J.J. & Amici, S. (2000). Detection
of Sub-micron Radiation from the Surface of Venus by Cassini/VIMS. Icarus
148, 307–311.
Bézard, B., De Bergh, C., Crisp, D. & Maillard, J.P. (1990). The Deep
Atmosphere of Venus Revealed by High-Resolution Night Side Spectra.
Nature 345, 508–511.
Bibring, J.-P., Langevin, Y., Poulet, F., Gendrin, A., Gondet, B., Berthé, M.,
Soufflot, A., Drossart, P., Combes, M., Belluci, G., Moroz, V., Mangold, N.,
Schmitt, B. & the OMEGA team (2004). Perennial Water Ice Identified
Around the Mars South Pole. Nature 428, 627–630.
Carlson, R.W., Baines, K.H., Encrenaz, Th., Taylor, F.W., Drossart, P.,
Kamp, L.W., Pollack, J.B., Lellouch, E., Collard, A.D., Calcutt, S.B.,
Grinspoon, D., Weissman, P.R., Smythe, W.D., Ocampo, A.C., Daniel-
son, G.E., Fanale, F.P., Johnson, T.V., Kieffer, H.H., Matson, D.L.,
McCord, T.B. & Soderblom, L.A. (1991). Galileo Infrared Imaging
Spectroscopy Measurements at Venus. Science 253, 1541–1548.
Collard, A.D., Taylor, F.W., Calcutt, S.B., Carlson, R.W., Kamp, L.W.,
Baines, K.H., Encrenaz, Th., Drossart, P., Lellouch, E. & Bezard, B. (1993).
Latitudinal Distribution of Carbon Monoxide in the Deep Atmosphere of
Venus. Planet. Space Sci. 41, 487–494.
Coradini, A., Capaccioni, F. Drossart, P., Semery, A., Arnold, G., Schade, U.,
Angrilli, F., Barucci, M.A., Bellucci, G., Bianchini, G., Bibring, J.P.,
Blanco, A., Blecka, M., Bockelee-Morvan, D., Bonsignori, R., Bouye, M.,
Bussoletti, E., Capria, M.T., Carlson, R., Carsenty, U., Cerroni, P., Colan-
geli, L., Combes, M., Combi, M., Crovisier, J., Dami, M., DeSanctis, M.C.,
DiLellis, A.M., Dotto, E., Encrenaz, T., Epifani, E., Erard, S., Espinasse, S.,
Fave, A., Federico, C., Fink, U., Fonti, S., Formisano, V., Hello, Y., Hirsch, H.,
Huntzinger, G., Knoll, R., Kouach, D., Ip, W.H., Irwin, P., Kachlicki, J.,
Langevin, Y., Magni, G., McCord, T., Mennella, V., Michaelis, H., Mondello,
G., Mottola, S., Neukum, G., Orofino, V., Orosei, R., Palumbo, P., Peter, G.,
Pforte, B., Piccioni, G., Reess, J.M., Ress, E., Saggin, B., Schmitt, B.,
Stefanovitch, Stern, A., Taylor, F., Tiphene, D. & Tozzi, G. (1998). VIRTIS:
An Imaging Spectrometer for the Rosetta Mission. Planet. Space Sci. 46(9/10),
1291–1304.
Drossart, P., Bezard, B., Encrenaz, Th., Lellouch, E., Roos, M., Taylor, F.W.,
Collard, A.D., Calcutt, S.B., Pollack, J. & Grinspoon, D.H. (1993). Search for

26
 scientific instruments

Spatial Variations of the H2O Abundance in the Lower Atmosphere of Venus

from NIMS Galileo. Planet. Space Sci. 41, 495–504.
Drossart, P., Piccioni, G., Coradini, A., Réess, J., Sémery, A., Suetta, E.,
Cosi, M., Dami, M. & Arnold, G. (2004). VIRTIS Imaging Spectrometer for
the ESA/Venus Express Mission. SPIE Infrared Spaceborne Remote
Sensing XII, Denver, Colorado, USA, 2–6 August 2004.
Melchiorri, R., Piccioni, G. & Mazzoni, A. (2003). VIRTIS-M Flight Lamps.
Rev. Sci. Instrum. 74(8), 3796–3801.
Kamp, L.W., Taylor, F.W. & Calcutt, S.B. (1988). Structure of Venus’s
Atmosphere from Modelling of Night Side Infrared Spectra. Nature 336,
360–362.
Piccioni, G., Amici, S., Fonti, S., Coradini, A. & Dami, M. (2000). Efficiency
Measurements of the VIRTIS-M Grating. Planet. Space Sci. 48(5), 411–417.
Roldan, C., López-Valverde, M.A., López-Puertas, M. & Edwards, D.P. (2000).
Non-LTE Infrared Emissions of CO2 in the Atmosphere of Venus. Icarus 147,
11–25.
Roos, M., Drossart, P., Encrenaz, Th., Lellouch, E., Bezard, B., Carlson, R.W.,
Baines, K.H., Kamp, L.W., Taylor, F.W. & Collard, A.D. (1993). The Upper
Clouds of Venus: Determination of the Scale Height from NIMS-Galileo
Infrared Data. Planet. Space Sci. 41, 505–514.
Taylor, F.W., Crisp, D. & Bézard, B. (1997). Near-Infrared Sounding of the
Lower Atmosphere of Venus. In Venus II, Univ. of Arizona Press, Arizona,
USA, pp325.

27