RADARSAT-1: An End-of-Mission Review of the Imaging and Calibration Performance of a Magnificent Canadian Instrument S. Cote, S. Srivastava Canadian Space Agency S. Muir Calian Technologies Ltd 1
RADARSAT-1 Canada's first operational EO satellite Launched November 1995 - Operational April 1996 HH, C-Band, right-looking Mission Lifetime: 5.25 yrs Actual Lifetime: 17.3 yrs Sun-synchronous, Dawn Dusk LEO Altitude: 798 km Period: 100.7 minutes Repeat Cycle: 24 days Orbits per day: 14 2
Outline Calibration Activities Closeout Image Quality (point target) review Radiometric performance review Internal calibration history Conclusion 3
Recent events affecting the spacecraft February 6, 2013: First occurrence of an anomaly related to a power degradation of the low power transmitter (LPT) induced by failure to load chirp information needed for imagery. March 5: After several occurrences (22) of LPT anomaly, decision was taken to reconfigure the payload in order to use the redundant unit. March 6: Redundant LPT was successfully activated. This did not immediately stop the anomaly from occurring: two other instances happened less than 24 hours after reconfiguration. March 7-8: Images acquired with the redundant LPT were of good general IQ. Plan had begun to assess the calibration and imaging performance of the reconfigured payload using point target and distributed target scenes. March 9-26: No anomalies, nominal payload imaging. March 27-28: Five planned outages for battery reconditioning were executed during that period (routine outages performed each spring). Spacecraft was back to nominal after procedures. March 29, between 06:05 and 07:28 UTC: Last complete communication with spacecraft: RADARSAT-1 was assumed to be in safe-hold mode. May 2013 4
Closeout of IQ and Calibration Operations Nominal IQ and Calibration operations have ceased shortly after final loss of communication with the spacecraft (29 Mar 2013): Acquisition planning, image analysis, trend analysis, transponder scheduling; Longer-term projects such as comparative radiometric level validations with R2, and prototyping of new beam pattern corrections (payload update 33); Close-out and final issuance of cal-val reports, procedures, databases, SW. Over the last couple of years, the Calibration Operations had already begun consolidating its activities for RADARSAT-1: Decommissioning of Resolute and Prince Albert transponders; Upgrade of Fredericton transponder and re-deployment in St Hubert. 5
Last Boreal Forest scene acquired Standard 1 6
Last transponder scene acquired Extended High 6 Transponder response TT&C antenna reflection CSA 7
IQ Performance: Impulse Response Indicators Range and Azimuth 3 db Impulse Response Width (30, 17.28, 11.58 MHz) 30 30 Range IRW (m) 25 20 15 10 30.0 MHz 17.28 MHz 11.58 MHz Azimuth IRW (m) 25 20 15 10 30.0 MHz 17.28 MHz 11.58 MHz 5 5 2D Integrated Side Lobe Ratio Range and Azimuth Peak Side Lobe Ratio 35 30 30 25 2D ISLR (db) 25 20 15 10 5 20 15 10 5 0 0 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 PSLR (db) Azimuth Range Specification 8
Geolocation accuracy After last range bias adjustment performed in Jan. 2007, directional location error bias was on average: +0.8 m. Average of absolute location error since Jan. 2007: 60 m. Requirement: 750 m. Absolute Location Error (m) 400 350 300 250 200 150 100 50 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Azimuth Range 2D 2009 2010 2011 2012 2013 Beam Absolute Location Accuracy (m) (predicted orbit) Wide 60 Standard 52 Fine 56 Low Incidence 72 High Incidence 56 Geometric distortion since Jan. 2007: 39.0 m (design goal: < 40 m) Based on Ottawa scenes with transponder and Gatineau antenna. 9
Image footprint offset relative to SPA (Swath Planning Application) Northern latitudes Boreal Forest and transponder sites Image Footprint Offset relative to SPA Swaths (km, East positive) 0-5 -10-15 -20 Jul-98 Jul-99 Manoeuvre overburn anomaly Apr. 2004 Jul-00 Jul-01 Jul-02 Jul-03 Jul-04 SPA orbit model update, Dec. 2004 Jul-05 Jul-06 Jul-07 Collision avoidance manoeuvre Aug. 2010 Jul-08 Jul-09 Jul-10 Jul-11 F S Jul-12 At end of mission, the centre of a delivered product was around 4 km to the West of the intended swath in SPA at the latitude of the calibration sites. Current trend towards a 0 km offset was part of a long-term oscillation between the simplified orbit model of the SPA and the actual RADARSAT-1 orbit propagator. Image footprint offset slowly continued its decrease of 1 km/year, which would have led to a 0 km offset by Aug 2016. 10
Radiometric stability from transponder data DGCF, 2000-present, from all transponders DGCF (db) = RCS true - RCS est 11 DGCF (db) 10 9 8 7 6 5 4 3 2 Ice buildup on antennas (NT0) R2 upgrade of Ottawa unit, deployed before recalibration Damaged radome horn covers (SK0) Azimuth positioner issue on ascending passes only (NT0) St Hubert not calibrated for R1 1 0-1 -2-3 - 0.084 db - 0.029 db 11
Relative radiometric accuracy Two-Way Gain (db) Elevation Angle (deg.) Current reference pattern In-orbit measurements of elevation beam pattern: averaging of image lines, then subtracting the backscatter profile of the area Difference pattern Radiometric deviation (rel. radiometric accuracy (maintained within 1 db) 12
Relative radiometric accuracy Since late 2008, relative radiometric accuracy measurements (elevation beam pattern shape monitoring) was performed using an area in the Canadian boreal forest. This site was exploited following the end of On-Board Recorder operations in early 2009, which prevented routine data acquisition at the Amazon. Anticipating On-Board Recorder limitations, Canadian areas had been studied since 2003 for beam pattern shape monitoring potential. To overcome seasonal backscatter variations, monthly backscatter models of the area have been derived and updated over the years. Acquisitions were increased during summer season, a period of higher confidence level of the measurements. Statistics at the area since 2003 indicate that certain in-scene deviations, as characterized, cannot originate from antenna pattern deviations, but from short-term fluctuations of the area s backscatter. These could therefore be excluded from beam pattern measurements, refining radiometric accuracy estimates. 13
Relative radiometric accuracy Measured at the boreal forest since since the last Payload update (March 2009). Rel. Rad. Acc. (db) µ EL1 1.00 S1 0.55 S2 0.53 S3 0.45 S4 0.44 S5 0.60 S6 0.50 S7 0.65 W1 1.19 W2 0.74 W3 0.74 EH3 0.44 EH4 0.45 EH6 0.28 F1N-F1-F1F 0.65 F2N-F2-F2F 0.83 F3N-F3-F3F 0.97 F4N-F4-F4F 0.76 F5N-F5-F5F 0.78 Fine beam patterns were not systematically tested at the boreal forest : swath sizes were not sufficiently large to overcome physical non-uniformities through averaging. Fine beam pattern levels were verified with transponder data, which showed no noticeable variation. Amazon data acquired between 1997 and 2008 showed Fine beams to be very stable radiometrically. Data acquired before F3 recalibration (Payload 32 ) 14
Relative radiometric accuracy Highest confidence level and best accuracy for elevation beam pattern measurements at the boreal forest: June to September 1.4 Relative radiometric accuracy measurements at the boreal forest site since last Payload update in March 2009 1.2 Relative Radiometric Accuracy (db) 1 0.8 0.6 0.4 0.2 15
Internal calibration (1999-2009) Non-imaging data acquisitions were used to test receiver linearity, and gain variation in limiter and Low Noise Amplifier Cal attenuator settings + AGC attenuator: 40 db range Results very similar year to year, no noticeable cyclic variations, but small signal decrease over 10 years: SAR Antenna Low Power Transmitter Cal. attenuator 0.5 db (30.0 MHz BW) 0.7 db (17.3 MHz BW) 1.0 db (11.6 MHz BW) Limiter & LNA A/D and AGC 102 100 30.0 MHz 1999 2001 Mean signal Strength (db) 98 96 94 92 Noise dominates Decrease of cal. attenuator setting Low relative noise 2002-1 2002-2 2003 2004 2005 2006-1 2006-2 90 2007 88 0 5 10 15 20 25 30 35 40 AGC Attenuation Setting (db) 2009 16
Internal calibration (1999-2009) On-line tracking of gain variation for in-processor compensation (every 8 th pulse, excludes limiter and Low Noise Amplifier) SAR Antenna Low Power Transmitter Pulse replica energy was slowly degrading: < 0.1 db/year Decays are different for each bandwidth, in close agreement with the long-term internal cal results. Signal decrease over 10 years: 0.5 db (30.0 MHz BW) 0.8 db (17.3 MHz BW) 0.9 db (11.6 MHz BW) Limiter & LNA A/D and AGC 30.0 MHz 9000 samples -0.05 db/yr 65.68 db in 1996 17.3 MHz 3000 samples -0.08 db/yr 65.61 db in 1996 11.6 MHz 7500 samples -0.09 db/yr 65.57 db in 1996 17
Imaging and cal-val performance: 1997-2013 Until end of RADARSAT-1 operations, imaging and calibration performance showed no sign of degradation, except for the very slow decrease of transmitted power: Through the nominal mission (1997-2001), and successive extended missions (2001-2013), image quality parameters remained very stable (resolution, focusing, location accuracy), and within initial specifications and design goals; Relative radiometric accuracy, monitored at the Amazon (1997-2008) and then at the boreal forest, (2008-2013), was maintained within the 1 db design goal; Long-term transponder DGCF data indicate good stability of the end-to-end SAR system gain over the years. The R1 IQ Operations represent a prime example of comprehensive, consistent SAR calibration monitoring and maintenance over a long duration (16+ years): Cal Plan was maintained while the IQ Ops System (including transponders) and image processor underwent various upgrades; RADARSAT-1 has set a standard for operational provision of stable and calibrated data, including ScanSAR; Consistent use of a single area of the Amazon (CEOS 2004), then of the Canadian Boreal Forest, permitted long-term characterization of the areas, for the benefit of present and future cal-val operations. 18
The RADARSAT-1 Image Quality Working Group Robert Hawkins (CCRS) Peter Hoang (CCRS then CSA) Tom Lukowski (CCRS then DND) Tony Luscombe (MDA) Dean Flett (CIS) Rob Gray and Nick Shepherd (Altrix) René Périard, Gordon Fitzgerald (RSI) Celine Fabi (RSI, then CIS) Pierre Le Dantec (RSI) 19
20