Instrumental concept and performances of the POLDER instrument.
Y.André, J-M.Laherrère, T.Bret-Dibat, M.Jouret, J.M.Martinuzzi, J.Perbos.
Centre National d'Etudes Spatiales
18 avenue Edouard BELIN, 3 1055 TOULOUSE Cedex
. ABSTRACT
The POLDER instrument is a wide field ofview radiometer designed to measure the polarization and the directionality
of the solar radiation reflected by the Earth-atmosphere system, in the visible and near infrared spectrum.
The original instrument concept of POLDER results in the capability of observing over a singlepass any target within the
instrument swath under up to 13 different viewing angles. For each viewing angle, the target is imaged in 8 narrowspectral
bands, and for 3 of these channels at three different polarization angles.
The multi-mission scientific objectives of POLDER lead to severe radiometric and geometrical requirements; thispaper
describes the POLDER instrument characteristics and the pre-flight performances measured on the flight model.
Developed by CNES, the French space agency, POLDER is installed on the ADEOS platform developed by NASDA, the
Japanese space agency. It will be launched in August 1996.
Keywords: Earth observation, multispectral camera, polarimeter, radiometer, POLDER, ADEOS.
1. INTRODUCTION
POLDER (POLarization and Directionality of the Earth Reflectance) is an instrument dedicated to the observation of
the polarized and directional solar radiation reflected by the Earth-atmosphere system. CNES, the Frenchspace agency, is
responsible for the POLDER system and the development of the instrument, that is to say performs the design, the
integration, the calibrations and the tests. POLDER will fly aboard the ADEOS (Advanced Earth Observation Satellite)
NASDA's satellite in 1996. ADEOS is a sun-synchronous orbiting satellite at an altitude of 797 km with an inclinaison of
98° and a local solar time at descending node between 10: 15 and 10:45 AM. Its design life time is threeyears of on-orbit
operations.
POLDER instrument concept is based on a CCD (Charged Coupled Device) matrix array detector, a rotating filter
wheel and a wide field of view optics. It can observe a target from 13 different viewing angles during the same orbit. The
instrument has no in-flight radiometric calibration device. Actually, a vicarious radiometric calibration will be performed
after the launch to assess the on-ground calibration and performances ofthe instrument.
After the main objectives of the mission, this paper presents the instrumental concept and describes the instrument
characteristics. Finally, the pre-flight calibration and performances, which measurements are described in acompanion
paper [1], are summarised.
2. MISSION SCIENTIFIC OBJECTIVES
The POLDER instrument will collect, from space, global observation of the polarization and directionality of the solar
radiation reflected by the earth-atmosphere system. The main objectives of the mission, referring to the International
Geosphere Biosphere program (IPGB) are to [2]:
O-8194-1931-1/95/$6.OO SPIE Vol. 2572 / 79
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
- Determine the physical and optical properties of the aerosols in the troposphere in order to improve their
classification and to study their variability and cycle.
- Improve the climate relevant knowledge of physical, chemical and radiative properties of clouds (clouds top
altitude, structure, aibedo ...).
- Study cloud-aerosol radiation interactions and the impact of clouds, water vapour and aerosols on the Earth
Radiation Budget (EBB).
- Study the variability of the primary production of carbon by the marine phytoplankton in order to improve the
understanding of the role of oceans in the carbon cycle.
- Improve the physical characterisation of the reflectance of vegetated surfaces in order to derive indices and other
parameters to be used in modelling the dynamics of the continental biosphere.
These objectives will be achieved through the sensor's unique capability to measure polarized reflectances in visible
and near-infrared spectrum, to observe a target within a single pass from 13 directions, and to operate in two dynamics
modes in order to achieve both high signal to noise ratio and wide dynamic range.
3. INSTRUMENTAL CONCEPT
The FOLDER instrumental concept is based on a wide field ofview telecentric optics, a rotating wheel bearing spectral
and polarized filters and a two-dimensional (2D) CCD detector. The 86° along-track by 102° cross-track instantaneous
field ofview intercepts a 2D Earth scene that is imaged on the 242 x 274 pixels CCD (see figure 1). This produces a swath
width of about 2400 km that allows a near complete daily coverage of Earth (see figure 2).
Filter'
Figure 1: POLDER instrument concept (artist's view, up)
The geometrical characteristics of the instrument are:
The size of the instrument field of view is: 42.3° along-track
50.7 ° across-track
The size of a ground pixel at nadir is 6.0 km along-track
7.1 km across-track
This size increases from nadir to swath edges, due to the earth curvature. as shown in figure 2
80 / SPIE Vol. 2572
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
Pixel (121.137)
I 7142m
6026 m
WHOLE FIELD OF VIEW: 2447 x 1809 KM2
Figure 2: Division ofthe footprint ofPOLDER's instantaneous field ofview into pixels. The pixel co-ordinates and
horizontal resolution are indicated for various pixels
Multi-angle viewing is achieved by the along-track migration at the spacecraft velocity. Thus, during a single orbit, the
same target can be observed under 13 different viewing angles (see figure 3). Combining multi-pass observations allows a
more complete sampling of the bi-directional reflectance and polarization distribution functions .For example, for a 40°N
target the observation sample of the Bidirectionality Reflectance Disfribution Function (BRDF) and Bidirectionality
Polarization Distribution Function (BPDF) is composed of 65 different viewing zenith angles and azimuths relative to the
sun direction in five days only. As the BRDF is generally symmetrical with respect to the principal plane, the density of
points can actually be doubled leading to a good analysis of the BRDF and the BPDF.
The selected spectral bands are: 443 nm polarized and 443 nm non polarized, 490 rim, 565 urn, 670 nm polarized, 763
nm, 765 nm, 865 nm polarized and 910 nm. The 443 non polarized band has been added to obtain a high signal to noise
ratio necessary for ocean colour measurements. The link between the different wavelengths and the mission objectives is
given in table 1.
443 nm 490 nm 565 nm 670 nm 763 nm 765 rim 865 nm 910 rim
Ocean,
Aerosols,
ERB
Ocean Ocean Vegetation, Clouds aerosols vegetation,
aerosols, clouds aerosols,
ERB ERB
water vapour
amount
Table 1: Main mission objectives of each spectral band
The exposure time of the CCD and the video electronics gain can be changed by telecommand in order to optimise the
video signal dynamic range with respect to the in-flight observed reflectances.
SPIE Vol. 2572 / 81
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
;//
Figure 3 : Principle of measurement. The CCD matrix is shown for two successive snapshots [2]. Theviewing angle of
the surface target has changed between the two snapshots ( thisfigure is not fit to scale).
POLDER has two nominal operational modes : the "imaging" mode and the "standby" one. The instrument will be set
in the "imaging mode" on the illuminated part of an orbit corresponding to solar zenithalangle e of less than 75°, which
leads to an operating time of about 50 mn per orbit, including the initial phase without image acquisition of about 10
minutes. POLDER will be set in the "standby" mode nominally anywhere else. The nominal operational sequence is an
alternance of the "imaging" and "standby" modes, repeated on the successive orbital cycles,leading to a nearly complete
coverage of the earth within one day.
The imaging mode consists in image cycles repeated all along each orbit. An image cycle is constituted by an image
sequence followed by a time interval during which no image is acquired. The image cycle duration (20 s) is four times the
image sequence duration (5 s). An image sequence is constituted by the 16 successive images, and corresponds to one turn
ofthe wheel.
82 I SPIE Vol. 2572
cci, Ary
w;d. FOY
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
Image cycle 19,6 s
<
<
image
>
sequence 4 9 s
>0 1 2 3 4 5 6 7 8 9
image window 306 ms
.111
exposure transition
24ms duration
;/,//J//7//;;///; ,/ long
exposure transition
lO5ms duralion
Figure 4: Nominal operational sequence
The exposure time of the signal of the CCD matrix may have two values depending on the objective of measurement:
Long exposure time corresponding to a low dynamic range observation, to improve the signal to noise ratio for the
ocean and aerosols mission objectives.
Short exposure time corresponding to a high dynamic range observation, in particular for clouds and vegetation
objectives.
Two types of sequences are defined , according to the allocation of a long or short exposure times to each spectral
channel. The combination of the two types of sequences is repetitive along the orbit. For one ground pixel, about 4
measurements are made in the low dynamic range mode and 1 1 in the high one for each wheel turn.
4. INSTRUMENT CHARACTERISTICS
4.1. General description
The external dimensions ofthe instrument are 800 x 500 x 250 mm3 for a weight of 32 kg. Its power consumption is
about 50 W during the imaging mode.
The main sub-systems ofPOLDER are:
- the optomechanical block,
- three electronics blocks composed of:
- the video electronics dedicated to the analogical acquisition and the 12 bits digital conversion of the video
signals,
- a buffer memory with a capacity of 16 images storage, associated to a telemetry generator interfacing with
ADEOS,
SPIE Vol. 2572 183
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
- the instrument manager which ensures the general control and scheduling of the instrument andmanages
the interface with ADEOS platform with regard to the commands and the housekeepingtelemetiy.
- the command electronics that controls the rotating wheel mechanism,
- the power supply converters,
- the CCD electronics thermal control,
- the thermal control (active for the optomechanical block, passive for the whole instrument),
- the mechanical structure.
The different sub-systems, shown in figure 5, are fixed on a carbon skin aluminium honeycomb plate. The instrument
is covered by a secondary structure, including the radiators. It hasveiy few thermal exchange with the ADEOS satellite and
has its own thennal control system.
Power distribution unit
Thermal control electronics
drive electronics
Figure 5: Internal architecture ofthe POLDER instrument. POLDER has been developed by CNES with, for the
realisation of the sub-systems, the participation of the following sub-contractors : ADR,ALTEN, CERCO, EREMS,
LATECOERE, SAGEM, SERESO. SEXTAN, SOPELEMJSOFRETEC, THOMSON, and ZODIAC.
4.2. The optomechanical block
The optomechanical block of POLDER gathers in the same structure:
- the objective,
- the filters fitted on the rotating wheel,
- the detection unit, composed of the CCD, its thermal control and the CCD electronics,
- the detector head electronic board that amplifies the output signal of the CCD,
- the rotating wheel, with a motor and ball bearings
- the wheel turn sensor,
84 / SPIE Vol. 2572
Buffer memory
Instrument manager
Video electronics
CCD sequencer
Video power supply converter
Connection board
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
4.2.1. The objective
Ten lenses compose the wide field ofview (1 14 degrees), telecentric objective ofPOLDER (see figure 6).
- The first one, made of synthetic silica has a good resistance to space radiation and ensures the protection of the whole
objective.
- The second one, corrects the distortion and compensates the theoretical variation of the pupil size in the field of view.
As a result, the light intensity distribution of an image on the CCD is the same that the distribution of the incident object.
- The third and fourth lenses constitute an afocal system with the two first lenses.
- The remaining lenses, forming three stuck doublets, focalise the image on the CCD and correct the optical defaults of
the objective such as aberrations or chromatism.
In order to reduce the reflection effects, the light interferences and the polarization rate of the objective, the lenses have
received an anti-reflection coatings.
The whole mechanical structure is made of a titanium alloy, chosen for its rigidity and its expansion coefficient close to
the lenses one.
The objective, with a 3.57 mm spectral average focal length at ff4.6, has a total distortion, defined as the gap to the
focal multiplied by the tangent of the field angle law less than 12 im at the focal plane.
DETECTOR
4.2.2. Rotating wheel and filters
Sixteen images are taken during one wheel turn, the first one being the darkness image which correspond to the CCD
dark current reference, the 15 others corresponding to the 15 spectral and polarized bands. As shown in table 2, three
spectral bands are equipped with polarizors : 443 nm, 670 nm and 865nm. For each of these wavelengths, three similar
spectral filters are associated with three polarizors, the direction of the first and the third one being at + 600 and -60° from
the direction of the central one. The three polarization directions are necessary to determine polarization properties of the
incident light, by inversion of the radiometric calibration model [l}.
During the time gap between two images, the satellite displacement is equal to one third of a pixel. As the mission
objectives need a good co-registration between the three polarized images of a same wavelength, the images are directly
superposed on the CCD matrix three by three by means of prisms fitted with the filters, with a gap of one pixel between two
SPIE Vol. 2572 / 85
SUN SHIELD
LENS
FIELD
DIAPII RAG M
APERTURE
DIAPIIRAGME
FILTER
Figure 6: optical design of the POLDER optomechanical block
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
successive groups of three superposed images during on-orbit operations. The distribution of prisms and filters, and the
characteristics of the filters are detailed in the table 2.
Channel Wavelength (nm) Band width (nm) Prism Polarization
1 443 20 + yes (+600)
2 443 20 no yes (0°)
3 443 . 20 - yes (-60°)
4 443 20 + no
5 490 20 no no
6 565 20 - no
7 670 20 + yes (+60°)
8 670 20 no yes (0°)
9 670 20 - yes (-60°)
10 763 10 + no
11 765 40 no no
12 910 20 - no
13 865 40 + yes (+60°)
14 865 40 no yes (0°)
15 865 40 - yes (-60°)
Table 2: Filters characteristics
A prism is noted + (-) when it registers the image with the next (previous) one.
4.2.3. Detector
The CCD is a TH7866 THOMSON-TMS detector including:
-a photosensitive zone
- a memory zone
- a lecture zone
- a output amplifier
The image zone is made of 242 x 548effective photo-elements. The photo-elements signals are binned on by two along
the detector lines after digital conversion in order to constitute the POLDER pixels.
The effective image zone consists then in 242 x 274 pixels, with a 27 x 32 ptm pixel size.
The CCD is also equipped with anti blooming drains in order to limit the radiometric disturbance induced by a saturated
zone due to the sun glint or the clouds.
Volume 800x500x250 mm3
Mass 32kg
Power consumption 50 W
Field ofview 86° along-track x 102° cross-track
Pixel coding 12 bits
Image size 242 x 274 pixels
Resolution at nadir 6 x 7 km2
Data rate 882 kbps continuous
Table 3: Main characteristics of the POLDER instrument
86 1 SPIE Vol. 2572
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
5. POLDER PilE-FLIGHT CALiBRATIONS ANDPERFORMANCES
5. 1. Geometrical calib ration and performances
The geometrical calibration consists in determining for each spectral band the mathematical model [1} relating the
viewing direction of an object with the location of the pixel on the CCD where it is imaged. The characteristics of this
model are :
- The Euler angles between the optical reference and the optical cube stuck on the optomechanical block.
- The co-ordinates on the CCD of the centre of the chromatic distortion law for each wavelength.
- The coefficients of the chromatic distortion law for each wavelength.
This distortion law is a fifth degree odd polynomial function of the field ofview angle tangent, taking into account the
effect of the lenses temperatures that are measured on-board for each image sequence.
With the help of a dedicated ground device and theodolite measurements all these characteristics are computed [1] and
included in the on-ground data processing to correct the images geometrical distortions.
The on-ground Euler angles between the optics reference frame and the POLDER/ADEOS interface reference are less
0.036° for a ground specification of 0. 15°. The on-orbit estimation is less than 0.04 °for a specification of± 0.2°.
The geometrical model accuracy is shown in figure 7: the line represents the theoretical specification for the
instrument, and the dots show the gap between the model and the measurements for the different bands at various view
angles.
The accuracy of the images co-registration has been evaluated, by measurements and calculations, for different points
of the field of view and time intervals between the images corresponding to the different needs: multi-polarization, multi-
spectral and multi-directional co-registration.
For a good estimation of the polarization rate of the target, the multi-polarization co-registration must by very
accurate. The results computed from the measurements are given in table 4.
Spectral
band
Centre of field For a view angle of 50°
443 nm 0.09 0.075
670nm 0.065 0.105
865 nm 0.045 0.145
Table 4: Accuracy of the multi-polarization co-registration given in POLDER pixel.
The multispectral studies need an accurate multi-band co-registration. The multi-band co-registration is obtained by
the superposition of the images, taken during the same wheel rotation, after the geometric processing. The on-orbit
accuracy, estimated using the pre-flight calibration results, is better than 0.06 of a pixel (RMS value) and 0. 1 of a pixel
(maximum value).
A good accuracy of the BRDF and BPDF measurements requires a precise co-registration of the images taken both
within one orbit (13 images of a same target in 250 s) and within different orbits (up to one month). The currenterror
budget is about one pixel. The platform attitude, determined by the AOCS of the satellite, constitutes the main part of this
SPIE Vol. 2572 I 87
Downloaded from SPIE Digital Library on 22 Dec 2011 to 211.68.2.156. Terms of Use: http://spiedl.org/terms
budget, and the instrument contribution is less than 0.22 of a pixel. As, the main part of the total error budget is constituted
by bias, it is expected to be reduced to 0.2 of a pixel by means of in-flight geometric calibration.
V
0
V
V
.0
•0
V
V
0
I.
5.2. Radiometric calibration and performances
The radiometric calibration has four aims:
- to determine the radiometric response of each pixel of the CCD,
- to determine the polarization rate ofthe optics,
- to measure precisely the direction of the 9 polarizers,
- to determine the spectral profile of each band.
The use of dedicated ground devices al