Mainly built around two dioptric optics arranged with a 40° angle, which optimises the elevation performances, the HRS camera was developed in one shot, direct qualification and validation on flight model.
After a description of HRS instrument architecture, design and performances, this paper describes its development and on-ground verification results. Then, the elevation in-orbit accuracy performances and the DEM end-products are presented.
The HRS mission is to generate Digital Elevation Model (DEM), geographical maps that include the altitude information, from stereoscopic pairs of images acquired sequentially by the High Resolution Stereoscopic HRS instrument mounted on SPOT5 satellite.
To satisfy defence, institutional cartography, geographical information system, telecommunication markets1, the HRS mission is sized to acquire 30 millions km2 over 5 years with 10 m elevation accuracy.
Instrument principle and dimensioning
Based on a push-broom concept with a wide optical field of view, the HRS instrument acquires each image of the stereo pair from:
▪ A row scanning obtained by the reading of a 12000 pixels CCD detector with a 10 m pixel resolution
▪ A column scanning obtained by the satellite velocity over its orbit. Thanks to the frequency sampling, the resolution is 5 m on this axis.
Stereo pairs are acquired along track in panchromatic spectral band [0.48μm - 0.7μm] by two cameras (fore and aft cameras), tilted by ± 20° with respect to Nadir about the track direction to ensure a 0.8 B/H ratio that optimises the elevation accuracy. SPOT5 being located at 830 km altitude, 600 km length of stereo strips is ensured, two consecutive stereo strips being spaced by 600 km as illustrated on Fig.1. To access any point on the earth surface within the orbital cycle without out-of-track sighting capability, selected for design simplicity reasons, the track width is greater than 109 km.
Getting excellent elevation accuracy is only possible with good radiometric and image finesse performances. Therefore, Modulation Transfer Function (MTF) characterizing the image quality and signal over noise ratio are of prime importance for the DEM accuracy and drive the instrument design and performances.
HRS imaging characteristics
|Nb of pixels/row||12000|
|Ground sampling||Row, 10 mColumn, 5 m|
|Swath Width||120 km|
|Viewing angle||-20° (fore camera)+20° (aft camera)|
|Spectral panchromatic range||0.48 μm - 0.70 μm|
|Min observable luminance||L1 = 19 (W.m-1.sr-1.μm-1)|
|Max observable luminance||L4 = 379 (W.m-1.sr-1.μm-1)|
Architecture and functions
The architecture of HRS instrument, briefly illustrated on the diagram hereunder, provides the following functional subassemblies:
Mounted on the instrument structure via quasi-isostatic mounts, 2 cameras, oriented at +20°/-20° with respect to Nadir, ensure the optical imaging and spectral filtering, the signal detection and video generation.
Each camera is composed of a detection unit coupled to a refractive telescope that provides the image quality and the focusing on the CCD detector thanks to a 11 lenses optical combination. Two parallel blades on the telescopes front side realize the spectral filtering and the radiation shielding (SUPRASIL blade). The telescope focus, key optical quality parameter, is thermally compensated and controlled. The telescope characteristics are:
|▪ Focal distance:||580 mm|
|▪ Useful optical diameter:||150 mm|
Based on a Thomson TH 7834 CCD (12000 pixels of 6,5 x 6,5μm2) and its proximity electronics, the detection unit is directly mounted on the rear side of the telescopes to ensure the focus stability performance. In this configuration, the CCD, controlled at 24 °C, is thermally and mechanically coupled to the telescopes. The proximity electronics dissipated power is evacuated thanks to specific radiators mounted on the top of the detection units. The detection unit sequencing and the acquisition of the video signal is performed by the MVS (Module Video Stereo).
Ultra light and highly stable, the main structure, made of Al honeycomb / CFRP sandwich, featuring new cyanate material (low thermal and moisture expansion figures), ensures a stable accommodation of the instrument units and decoupled interfaces with the satellite. In addition, secondary structures are implemented to support the active thermal control (MVS and Optical Assembly thermal hoods) and to avoid sun entrance in the cameras (sun cap).
The Optical Assembly accurate thermal control (+ 0,5°C temperature and thermal gradient stabilities), avoiding the use of any refocusing mechanism, is achieved thanks to a thermal enclosure concept that radiatively controls the high thermal inertia assembly via a combination of active and passive thermal control. Instrument control electronics
Accommodated inside a single unit, the instrument control electronics (MVS) provides management, housekeeping, detection sequencing and video electronics.
The management and housekeeping electronics realize the implementation of instrument modes and monitoring, the active heat control of the instrument and the standard SPOT5 interfaces with the satellite (DC/DC conversion of the satellite power bus and OBDH data handling bus). A CCD sequencing electronics, a Video electronics at 4,1Mpixels.s-1, capable of up to 10 Mpixels.s-1 and a 12 bits analog to digital conversion complete the detection chain and provide HRS images digital information to the satellite.
Accomodation on SPOT5
HRS instrument accommodation on SPOT5 is driven by two main considerations: on one hand, the maximization of interfaces decoupling between SPOT5 and HRS instrument and on the other hand, the minimization of the mechanical path between HRG and HRS instruments to optimise their relative stability.
Thermal isolation (MLI, thermal washers) is implemented at instrument interfaces. Mechanical decoupling is achieved via isostatic mounts used on sensitive instrument units (cameras).
Main performance data
HRS performances deal with:
▪ Geometrical performances:
▪ Radiometric performances
▪ Optical performances
Main instrument performances requirements
|Alignment stability||< 150 μrad||Between cameras|
|< 500 μrad||Between each camera and optical reference|
|S/B ratio||> 130 @ L2||L2= 118 W.m-1.sr-1.μm-1|
|MTF||> 0,18||Across Track|
|> 0,20||Along Track|
|Mass||< 90 kg|
|Power||< 75 W||Consumption in imaging mode|
|Reliability / availability||> 0,98||For the nominal mission|
DEVELOPMENT AND VERIFICATION
Decision to implement HRS mission on SPOT5 was finalised in the very beginning of year 1999 with the major condition that HRS development should not delay the overall satellite timescale. This resulted in an overall timescale for HRS instrument development of 2 years.
Given the available time, it was necessary to start in parallel units and instrument studies during the early engineering phase and to select a mono model philosophy, direct manufacturing of proto-flight model. Therefore, the instrument development was split in three phases, as illustrated on Fig.6:
▪ Architecture and design phase (from T0 to T0+5), concluded by a Preliminary Design Review
▪ Consolidation phase (from T0+5 to T0+15), concluded by a Manufacturing Release Review
▪ Qualification phase (from T0+15 to T0+25), concluded by a Qualification Review
The early engineering phase was an essential step for the success of the HRS challenging development. Indeed, to speed up the instrument definition, concurrent engineering was implemented with joint teams working in parallel to early identify design drivers and freeze the architecture, the design and the interfaces of the overall instrument. This phase has led to develop in parallel all mathematical models, to run simulations and establish budgets, each industrial team being fed in quasi real time by the others’ study outputs. For the same schedule constraints reasons, bulk of HRS instrument electronics are made of Mil standard parts, with adequate screening. Extensive use of Field Programmable Gate Array (FPGA) components for digital electronics allowed simple, late and quick design evolution. At satellite interface level, full « High-Rel » standard components are used.
Combination of analyses, simulations and tests, HRS verification and qualification programme was successfully held in 2000. However, the full qualification and the instrument flight worthy ability was only pronounced in 2001 after successful EMC qualification performed at S/L level.
The verification programme, done at instrument level, followed the following sequence:
▪ Instrument alignment: fore wrt aft camera alignment and alignment of cameras wrt alignment reference
▪ Functional and performances tests at ambient: modes management, thermal control, geometrical, radiometric and optical performances
▪ Reference test: sub-set of functional and performances parameters verification to trend the instrument health up to launch
▪ Mechanical qualification: mass and design qualification under mechanical environment (QSL, sine and acoustic environment)
▪ Thermal qualification: thermal power, thermal control and instrument performances verification under TB/TV environment. It is to be mentioned that a specific defocus sequence was applied during this test to characterise the defocus model
After a characterization of reference frames transfer matrices, the alignment stabilities (geometrical performances) were trend during the instrument verification sequence (before and after mechanical and thermal environment).
The results, illustrated on Table3, highlighted excellent stability performances that allow for automatic data reduction at customer level.
Alignment stabilities verification
|Aft camera / optical ref||Fore camera / optical ref||Between cameras|
|After vibration||< 85 μrad||< 55 μrad||< 125 μrad|
|After TB/TV||< 115 μrad||< 30 μrad||< 85 μrad|
After video chain calibration, mainly dedicated to gain and phase settings, the functional radiometric behaviour was first measured, especially:
▪ CCD offset correction
▪ Video chain behaviour in case a instrument detector saturation
▪ Recovery period (less than one CCD line) when switching from one camera to the other one or when changing the gain of the video chain
Then, the radiometric coefficients (equalization and dark coefficients) necessary to compensate for telescope and detection chain non-uniformity were characterized and delivered to the customer. Finally, the radiometric performances measured at ambient and under vacuum, exhibited excellent results in all domains.
Radiometric performances verification results
|Dispersion||Global||< 15 %||< 5 %|
|Local||< 8 %||< 2 %|
|Noise||S/N @ L2||> 120||> 180|
|Linearity||Relative @ G3||< 2 %||< 1 %|
As illustrated on Fig.9, the noise under different brightness, main instrument radiometric performance, is well within the specifications
HRS detection chain is linear over the large dynamic range (see Fig.10).
Spectral response, straylight, polarisation, distortion and Modulation Transfer Function (MTF) are the main optical performances, which have driven the instrument design. Measured at ambient and under vacuum (especially MTF), these performances are illustrated on Table4.
Optical performances verification synthesis
|Rejection||< 0.1 %||Rejection: transmission outside spectral range|
|Straylight||< 10 %||Worst case figure when light is rasante|
|Polarisation||< 0.3 %|
|Distortion||< 15 μm||Without modelling correction|
|MTF||> 0.19> 0.26||Fore cameraAft camera|
HRS mission and elevation accuracy are very dependent of instrument MTF performances. Therefore, during the instrument verification sequence, a specific attention was paid to this critical parameter that includes the MTF over the Field of View at best focus and the defocus performances. Although well within the specification for both cameras, a better MTF performance is observed on camera2 (Fig.12) compared to camera1 (Fig.11). This is only due to the dioptric telescope lenses alignment procedure, largely improved on the second camera to reduce the residual astigmatism. Therefore, complying with MTF requirement means a defocus stability of + 30 μm on camera 1 (fore camera) and + 45 μm on camera2 (aft camera).
So, it is clear that focus alignment and stability and, consequently the thermal control of the cameras, are key issues for the HRS mission performances. Therefore, large efforts were spent to master the defocus stability performance.
Thanks to optical simulations, a defocus modelling was first established:
Δfi: Initial focus adjustment @ 0°C
Sθo: Sensitivity to camera mean temperature
SGl: Sensitivity to camera longitudinal gradient
SθB: Sensitivity to baffle temperature
Sθhs: Sensitivity to heat sink of the observed scene
Sθhsw: Sensitivity to orbital variation of heat sink
θo: Camera mean temperature
Gl: Camera longitudinal gradient
θB: Baffle temperature
θhs: Heat sink of the observed scene
θhsw: Orbital variation of heat sink
Then, to correlate the model and predict flight performances, defocus sensitivity cases were introduced during TB/TV test. These specific technological cases performed on both cameras allowed estimating the model coefficients with defocus accuracy better than 4 μm over the whole campaign:
HRS IN-ORBIT RESULTS
HRS technological monitoring
After one year of successful mission in orbit, HRS instrument presents stable and excellent performances on observable parameters:
Voltage, current and power
In-orbit HRS voltages, currents and power consumptions, identical to predictions and on-ground measurements, exhibit stable behaviour, demonstrating the good health of the instrument
The in-orbit thermal control is performing as expected. The only unexpected behaviour concerns the CCD proximity electronics slightly hotter than predicted due to a higher satellite interface on –Xs side. Nevertheless this result has no performances impact.
As expected, no long-term thermal evolution is observed on cameras, the main contributor remaining the orbital environment:
In-orbit defocus estimation
Considering the correlated defocus model established during on-ground verification, the estimated in-orbit defocus, illustrated on Fig.17, is:
The orbital stability, better than expected (+ 7 μm compared to + 14 μm), can be explained by the inaccuracy of the defocus model heat sink coefficients estimation, inaccuracy due to the impossibility to simulate on-ground the real heat sink (thermal shroud and optical ground support equipment disturbing the representativity of the observed scene during TB/TV).
In-orbit MTF estimation
From the estimated in-orbit defocus, the worst-case MTF over the field of view is:
much better than the required 0.18 specification.
Over the 1.5-year in-orbit lifetime, all parameters are stable and exhibit excellent performances.
Since May 2002 systematic HRS acquisition is planned over pre-determined area that are classified depending of strategic and commercial priorities. HRS resource is shared between defence and commercial requirements on the basis of the initial public private partnership risksharing scheme:
Acquisition is being processed successfully and overall acquired area exceeds by a factor 3 the original budgets. Data validation is a several steps process:
▪ Cloud coverage check quickly after acquisition, in order to reallocate satellite resource
▪ Fine correlation verifications in order to qualify the stereo pairs and to archive them for further potential DEM generation.
Last September 2003, HRS acquisition resulted in 42000000 km2 of cloud free covered area and 35000000 km2 of correlated stereo pairs ready for DEM manufacturing.
HRS DEM PRODUCTS
Reference3D, the HRS Digital Elevation Model(DEM) commercial product, is produced under a co edition agreement between France’s Institut Géographique National (IGN) and Spot Image. It is a tilled product, complete geographical information source, made up of one-degree square tiles containing 7 geographically superposable layers:
HRS DEM Performances(2)
Location performances resulting from heavy work of French space agency (CNES) and mapping agency (IGN), has been conducted on more than 20 test sites. Reference3D production process is based on three steps:
▪ Registration of stereo pairs
▪ Generation of a DEM by automatic correlation on the stereo pairs
▪ Overall calculation of absolute site geometry by block triangulation
First level accuracy is mainly affected by orbital performances. Nevertheless, elevation accuracy (z) is already better than 15m for more than 90% of points.
Raw location performance
|Cross track||Along track||z|
|Mean||-0.7 m||-3.7 m||-5.6 m|
|Standard deviation||18.7 m||34.3 m||7.4 m|
|90% threshold||62.4 m||14.5 m|
Modelling of thermo elastic effect resulting from orbital position reduces significantly location dispersion. Block triangulation, most effective part of correction, increases planimetric accuracy by more than a factor 2 on planimetry and than a factor 3 on elevation accuracy.
Location performance after after thermo elastic effects modelling and correction and block triangulation
|Cross track||Along track||z|
|Mean||0.5 m||0.5 m||-0.7 m|
|Standard deviation||6.4 m||5.2 m||2.6 m|
|90% threshold||12.3 m||4.0 m|
Finally, elevation accuracy can still be improved, using elevation data of known points in the block triangulation process, that has sense as coastlines provide simple accurate potential points at zero elevation.
Location performance after block triangulation with Z control points
|Cross track||Along track||z|
|Mean||-0.3 m||0,5 m||0,0 m|
|Standard deviation||7,2 m||5,1 m||0,7 m|
|90% threshold||13,4 m||1,2 m|
This evaluation campaign and the continuous in orbit monitoring have confirmed that the Reference3D users needs are satisfied and somewhat exceeded. This will allow providing satellite data with high location performances all over the world avoiding necessary use of Geographical Control Points (GCP).
The HRS instrument, funded via an innovative private/public partnership, was developed in two years as originally planned with a mono model philosophy. The on-ground and in-orbit performances exceed the instrument requirements and satisfy the HRS mission.
Indeed, the HRS acquisition is today in advance with respect to the original acquisition plan and the DEM products exhibit outstanding performances after compensation (1 m compared to 10 m requirement).
The HRS mission success is the result of efforts spent to conduct the instrument development and then produce the Digital Elevation Models.
The authors wish to thank all people at CNES, ASTRIUM, SPOTIMAGE, IGN who contributed by their daily work to the success of this mission.
Laurent Maggiori, Christian Rouziès, Marc Bernard, “HRS: an enhanced mission capability for SPOT5”, AIAA, 2000Google Scholar
Reference3D location performance review and prospects ISPRS & EARSEL joint Workshop “High-Resolution Mapping from Space 2003”, Hanover, GermanyGoogle Scholar
Laurent Maggiori, Michel Duran, Gilles Planche, “HRS: a first assessment”, IAF, 2003Google Scholar