ADEN ALOS PALSAR PRODUCT VERIFICATION Patricia Wright 1 , Peter Meadows 1 , Gordon Mack 2 , Nuno Miranda 3 & Marco Lavalle 4 1 BAE SYSTEMS Advanced Technology Centre, Chelmsford, UK 2 VEGA Group PLC, Welwyn Garden City, UK 3 ESA/ESRIN, Frascati, Italy 4 University of Rome “Tor Vergata”, Italy Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Introduction - Within the ALOS Data European Node (ADEN) the verification of PALSAR products is an important and continuing activity, to ensure data utility for the users. - This is performed for ESA-ESRIN by the PALSAR team within Instrument Data Quality Evaluation and Analysis Service (IDEAS). - There are three levels of PALSAR products: L0 (raw data), L1.1 (complex data) and L1.5 (detected data). - We examine four product types: fine mode single polarisation (FBS), fine mode dual polarisation (FBD), fine mode quad polarisation (PLR) and wideswath (WS). Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Routine Analysis - Header analysis of all PALSAR products distributed electronically by ADEN (Level 400 Doppler Centroid Frequency (Hz) 1.0, 1.1 and 1.5). -100 - Parameters examined include: -600 - Data Coverage - Doppler frequency -1100 - Spacecraft attitude - Pulse repetition frequency -1600 0 1000 2000 3000 4000 5000 6000 7000 8000 - Orbit state vectors Frame Doppler centroid frequency monitoring – note the data points at -1600Hz due to suspension of yaw steering. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Product Visualisation Interference - Visual inspection performed on a sample set of PALSAR products. - Several types of artefacts observed: - Interference - Ambiguities - Wideswath sub-swath boundaries - Wideswath scalloping - Wideswath azimuth fluctuation Range Ambiguities Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Product Type P1.1 P1.5W1.5H1.1H1.5 Sub-Swath Boundaries 0 0 20 0 0 Azimuth Ambiguity 0 1 7 0 5 Scalloping 0 0 10 0 1 Incorrect Antenna Pattern Correction 0 0 1 1 0 Range Ambiguity 1 2 6 5 12 Azimuth Focusing 0 0 0 1 1 Sub-Swath Range Intensity Fluctuation 0 0 3 0 0 Azimuth Intensity Fluctuation 0 0 7 0 0 Interference 0 0 6 1 0 Total Number Of Products Inspected 19 28 46 34 68 Wideswath sub-swath boundaries and P is quad-pol, W is wideswath and H is single or dual pol azimuth intensity fluctuations - The majority of the artefacts are seen in wideswath imagery, some in single/dual polarisation imagery and very few in quad pol data. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Impulse Response Function (IRF) Analysis - IRF analysis assesses the spatial resolution and sidelobe properties of the data (for comparison with theoretical results). - PALSAR products acquired over ground stations at Maspalomas, Spain, Matera, Italy and Tromsø, Norway and the DLR corner reflectors, Germany have been used. Maspalomas Ground Station IRF in wideswath imagery (rcs too high for fine mode imagery) Resampled Image Slice through IRF Peak Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
- IRF measurements include azimuth and range spatial resolution, integrated sidelobe, peak sidelobe and spurious sidelobe ratios. Acq Date Inc Azi Res Range ISLR PSLR SSLR Ang (m) Res (m) (dB) (dB) (dB) 16/04/07 39.2 169.1 120.9 -15.1 -8.5 -11.9 10/07/07 20.2 131.7 135.3 0.5 -8.9 -9.0 25/08/07 40.7 166.8 187.6 -5.0 -10.9 -12.1 10/10/07 40.7 128.5 129.4 -7.7 -8.6 -11.5 25/11/07 40.8 165.7 123.0 -12.4 -6.5 -11.7 02/12/07 23.0 170.3 121.3 -0.9 -9.1 -9.0 11/04/08 40.8 165.3 147.5 -2.6 -11.2 -9.6 03/09/08 22.7 133.5 125.8 6.3 -4.6 -6.4 Maspalomas WS HH IRF Measurements - Variability in the spatial resolution is due to the data being undersampled (pixel size is 100m). The ISLR is also quite variable while the PSLR and SSLR are acceptable. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Matera Ground Station IRF in Level 1.1 Single Polarisation Imagery Acq Date Inc Azi Range ISLR PSLR SSLR Acq Date Inc Azi Range ISLR PSLR SSLR Ang Res Res (dB) (dB) (dB) Ang Res Res (dB) (dB) (dB) (m) (m) (m) (m) 23/08/07 38.6 21.2 21.2 11.5 -0.8 10.9 20/02/07 38.7 8.5 7.7 -9.4 -7.9 -10.7 Matera Level 1.5 FBS and FBD HH IRF Measurements - Spatial resolution indicates that the data is undersampled (pixel size is 6.25m for FBS and 12.5m for FBD). The FBD sidelobes high. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Acq Date Inc Azi Range ISLR PSLR SSLR Acq Date Inc Azi Range ISLR PSLR SSLR Ang Res Res (m) (dB) (dB) (dB) Ang Res Res (m) (dB) (dB) (dB) (m) (m) 05/01/07 38.9 4.47 4.59 -3.47 - - 08/07/07 38.7 4.75 9.33 0.93 - - 20/02/07 38.8 4.47 4.47 -2.10 - - 23/08/07 38.7 4.61 9.20 1.15 - - 08/01/08 38.7 4.59 4.61 -3.78 - - 25/05/08 38.6 5.71 9.56 1.79 - - 23/02/08 38.6 4.28 4.55 -1.34 - - 10/07/08 38.8 4.37 10.00 2.61 - - 09/04/08 38.6 4.23 4.44 -0.00 - - 25/08/08 38.9 4.71 9.08 4.79 - - 25/08/08 38.9 5.46 9.91 2.68 - - Matera Level 1.1 FBS and FBD HH IRF Measurements - Spatial resolutions comparable with theoretical values (data also adequately sampled). ISLR high due to ground station rcs being low (~30dbm 2 ) and consequently the background is quite high relative to the IRF. This is the reason why PSLR and SSLR is not calculated. - Tromsø ground station results similar to Matera (also FBS and FBD). Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
- During the ADEN commissioning phase six corner reflectors deployed by DLR, Germany as suitable point targets for IRF analysis. Acq Date Pol Azi Range ISLR PSLR SSLR Acq Date Pol Azi Res Range ISLR PSLR SSLR Res (m) Res (m) (m) (m) (m) (m) Res (m) (m) (m) (m) 06/09/06 HH 16.7 22.3 -1.1 -6.0 -12.7 06/09/06 HH 4.48 9.63 -5.2 -12.7 -17.2 06/09/06 VV 16.5 26.4 -1.2 -8.4 -12.7 06/09/06 VV 4.44 9.85 -5.3 -14.0 -14.7 15/11/06 HH 18.0 27.3 -0.1 -7.7 -10.0 15/11/06 HH 4.58 9.77 -5.4 -11.6 -16.7 15/11/06 VV 17.1 26.7 0.9 -7.5 -10.6 15/11/06 VV 4.66 9.78 -5.7 -11.2 -17.4 DLR Level 1.1 PLR IRF Measurements DLR Level 1.5 PLR IRF Measurements - The Level 1.5 spatial resolution measurements indicate that the products are undersampled in azimuth (the PLR pixel size is 12.5m) but not in range. The ISLR values are high while the PSLR and SSLR values are reasonable. - The Level 1.1 spatial resolution measurements are comparable with their theoretical values and the sidelobe ratios are lower than the corresponding Level 1.5 measurements. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Radiometric Stability - A successful method of assessing the stability the SAR instruments such as ERS-2 SAR and Envisat ASAR has been to be the Amazon rainforest by calculating and monitoring gamma ( σ 0/cos(i)). - Since Amazon rainforest data is not available for PALSAR data within the ADEN node, an alternative site within the African rainforest has been selected. -5.4 Gamma (dB) -5.6 -5.8 -6.0 -6.2 36 37 38 39 40 41 Incidence Angle (deg) Gamma range profile Fine Mode Level 1.5 African Rainforest Image Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Acq Date Orbit Mean Trend Gamma 21/02/07 5745 -5.75 dB -0.02 dB 09/10/07 9100 -5.81 dB -0.10 dB 24/11/07 9771 -5.82 dB -0.23 dB 09/01/08 10442 -5.84 dB 0.27 dB 24/02/08 11113 -5.70 dB 0.37 dB 10/04/08 11784 -5.66 dB 0.00 dB Mean gamma and trend for Fine Mode African Rainforest Imagery (HH polarisation) - There is some variability in the mean gamma (max – min = 0.18dB) and in the slope of gamma from near to far range. The mean gamma is -5.76±0.07dB. - This indicates an excellent PALSAR radiometric stability of 0.07dB. Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Noise Equivalent Sigma0 - The upper limit to the noise equivalent radar cross-section (NESigma0) of an image can be estimated by measuring the radar cross-section of low intensity regions (usually ocean region under low wind speed conditions). - Measurements compare favourably with the JAXA specifications of -21dB for HH & VV polarisations and -25dB for HV & VH polarisations. -20 HV Polarisation NESigma0 VH NESigma0 (dB) -25 VV HH -24.84 ± 2.89 HH -30 HV -30.61 ± 1.96 VH -27.44 ± 3.23 -35 -40 Mean NESigma0 15 20 25 30 35 40 45 Incidence angle (deg) NESigma0 Measurements Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
ENL and Radiometric Resolution - The equivalent number of looks and the radiometric resolution can be measured directly from PALSAR imagery by using large uniform distributed targets. - The measured ENL are close to the actual ENL used in the processing: Product Actual ENL ENL Rad Res No. Results (dB) P1.1 1 0.87 3.17 9 P1.5 4 4.23 1.72 4 H1.1 (FBS) 1 0.98 3.04 1 H1.5 (FBS) 2 1.28 2.79 5 H1.1 (FBD) 1 0.82 3.24 2 H1.5 (FBD) 4 3.35 1.89 6 W1.5 8 7.21 1.41 7 ENL and RR Measurements Advanced Technology Centre ALOS PI Symposium, 3rd November 2008
Recommend
More recommend
