ASCAT RADIOMETRIC PERFORMANCE

Transcription

1 ASCAT RADIOMETRIC PERFORMANCE J. J. W. Wilson Slide 1

2 BASIC ERROR CONTRIBUTIONS Static Calibration Bias Error Gain Error ε = ± db Corresponding Cross-Section Error e ( ε ) = ± db Quasi-static Bias Error Around Orbit Gain Error δ = db Worst Case LA δ = db Best Case RM Corresponding Cross-Section Error e ( δ ) = db WC e ( δ ) = db BC Calibration Noise-like Error (1 sigma ) Gain Error = ± db Corresponding Cross-Section Error σ ( ) = ± db Slide 2

3 TOTAL ERROR (WORST CASE) ON SINGLE MEASUREMENT OF IDEAL POINT TARGET Total Cross-Section Error Estimate ( 95.4 % confidence ) e ( ε ) + e ( δ ) + 2 σ ( ) = ± 0.35 db Part of this error comes from the transponder stability. The budget transponder radiometric stability error ( 1 sigma ) σ ( TS ) = ± 0.07 db ASCAT only Radiometric Accuracy Error ( 95.4 % confidence ) e ( ε ) + e ( δ ) + 2 Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± 0.30 db ASCAT only Radiometric Accuracy Error ( 99.7 % confidence ) e ( ε ) + e ( δ ) + 3 Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± db Slide 3

4 TOTAL ERROR (BEST CASE) ON SINGLE MEASUREMENT OF IDEAL POINT TARGET Total Cross-Section Error Estimate ( 95.4 % confidence ) e ( ε ) + e ( δ ) + 2 σ ( ) = ± db Part of this error comes from the transponder stability. The budget transponder radiometric stability error ( 1 sigma ) σ ( TS ) = ± 0.07 db ASCAT only Radiometric Accuracy Error ( 95.4 % confidence ) e ( ε ) + e ( δ ) + 2 Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± db ASCAT only Radiometric Accuracy Error ( 99.7 % confidence ) e ( ε ) + e ( δ ) + 3 Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± db Slide 4

5 TOTAL ERROR (WORST CASE) AFTER MANY MEASUREMENTS OF IDEAL POINT TARGET Error on Mean of Sample for Sample Size N=90: Error on Final Estimated Gain Pattern due to Calibration Noise-like Error (1 sigma) Gain Error M = ± db / Sqrt( N ) Corresponding Cross-Section Error σ ( M ) = ± db / Sqrt ( 90 ) σ ( M ) = ± db Total Cross-Section Error Estimate ( 95.4 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 2 σ ( M ) = ± db Total Cross-Section Error Estimate ( 99.7 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 3 σ ( M ) = ± db Part of this error comes from the transponder stability. The budget transponder radiometric stability error ( 1 sigma ) σ ( TSM ) = ± 0.07 db / Sqrt( 90 ) σ ( TSM ) = ± db ASCAT only Radiometric Accuracy Error ( 95.4 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 2 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db ASCAT only Radiometric Accuracy Error ( 99.7 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 3 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db Slide 5

6 TOTAL ERROR (BEST CASE) AFTER MANY MEASUREMENTS OF IDEAL POINT TARGET Error on Mean of Sample for Sample Size N=90: Error on Final Estimated Gain Pattern due to Calibration Noise-like Error (1 sigma ) Gain Error M = ± db / Sqrt( N ) Corresponding Cross-Section Error σ ( M ) = ± db / Sqrt ( 90 ) σ ( M ) = ± db Total Cross-Section Error Estimate ( 95.4 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 2 σ ( M ) = ± db Total Cross-Section Error Estimate ( 99.7 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 3 σ ( M M ) = ± db Part of this error comes from the transponder stability. The budget transponder radiometric stability error ( 1 sigma ) σ ( TSM ) = ± 0.07 db / Sqrt( 90 ) σ ( TSM ) = ± db ASCAT only Radiometric Accuracy Error ( 95.4 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 2 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db ASCAT only Radiometric Accuracy Error ( 99.7 % confidence ) e ( a ) + e ( ε ) + e ( δ ) + 3 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db Slide 6

7 ALGORITHM ERROR e(a) & SATELLITE REFLECTION LM BEAM Slide 7

8 ERROR (WORST CASE) ON A SINGLE MEASUREMENT OF A PREFECT DISTRIBUTED TARGET ASCAT only Calibration Noise-like Error on Cross-Section (1 sigma ) σ ( MEAS ) = Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± db = ± Distributed Target Noise-like Statistical Measurement Error (1 sigma) σ ( Kp ) = ± Kp. SZO Total Cross-Section Error on Single Measurement of Distributed Target E(DT, p) = e ( ε ) + e ( δ ) + p Sqrt ( σ 2 ( MEAS ) + ( Kp. SZO ) 2 ) Typically Kp is between 2% and 3.5%. For Kp = 3%: 95.4% Confidence: For SZO = 0 db, E(DT, 2) = ± db For SZO = - 10 db, E(DT, 2) = ± db For SZO = - 20 db, E(DT, 2) = ± db 99.7% Confidence: For SZO = 0 db, E(DT, 3) = ± db For SZO = - 10 db, E(DT, 3) = ± db For SZO = - 20 db, E(DT, 3) = ± db Slide 8

9 ERROR (BEST CASE) ON A SINGLE MEASUREMENT OF A PREFECT DISTRIBUTED TARGET ASCAT only Calibration Noise-like Error on Cross-Section (1 sigma ) σ ( MEAS ) = Sqrt ( σ 2 ( ) - σ 2 ( TS ) ) = ± db = ± Distributed Target Noise-like Statistical Measurement Error (1 sigma) σ ( Kp ) = ± Kp. SZO Total Cross-Section Error on Single Measurement of Distributed Target E(DT, p) = e ( ε ) + e ( δ ) + p Sqrt ( σ 2 ( MEAS ) + ( Kp. SZO ) 2 ) Typically Kp is between 2% and 3.5%. For Kp = 3%: 95.4% Confidence: For SZO = 0 db, E(DT, 2) = ± db For SZO = - 10 db, E(DT, 2) = ± db For SZO = - 20 db, E(DT, 2) = ± db 99.7% Confidence: For SZO = 0 db, E(DT, 3) = ± db For SZO = - 10 db, E(DT, 3) = ± db For SZO = - 20 db, E(DT, 3) = ± db Slide 9

10 RE-CALIBRATION ERROR ASSUMING STABILITY Re-Calibration Error Assuming Stability (ASCAT + TRANSPONDERS) 95.4% Confidence: e ( a ) + e ( ε ) + 2 σ ( M ) = ± db 99.7% Confidence: e ( a ) + e ( ε ) + 3 σ ( M ) = ± db Re-Calibration Error Assuming Stability (ASCAT ONLY) 95.4% Confidence: e ( a ) + e ( ε ) + 2 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db 99.7% Confidence: e ( a ) + e ( ε ) + 3 Sqrt ( σ 2 ( M ) - σ 2 ( TSM ) ) = ± db Slide 10

11 SUMMARY OF RADIOMETRIC PERFORMANCE Point Target Cross-Section Single Measurement Best Case Worst Case ± 0.20 db (2 sigma) ± 0.30 db (2 sigma) ± 0.28 db (3 sigma) ± 0.38 db (3 sigma) Distributed Target Normalised Cross-Section Single Measurement for Kp = 3% Best Case Worst Case ± 0.20 db (2 sigma) ± 0.30 db (2 sigma) ± 0.28 db (3 sigma) ± 0.38 db (3 sigma) Residual Bias Contribution Best Case Worst Case ± 0.06 db (2 sigma) ± 0.15 db (2 sigma) ± 0.07 db (3 sigma) ± 0.15 db (3 sigma) Re-Calibration Error Assuming Stability ± 0.06 db (2 sigma) ± 0.07 db (3 sigma) Slide 11

12 RAIN FOREST GAMMA ZERO 2007: ASC & DES Slide 12

13 RAIN FOREST GAMMA ZERO ASC+DES: YEAR ON YEAR Slide 13

14 Kp (Approx.) versus Sigma Zero One week / Left Mid Slide 14

15 SUMMARY & RECOMMENDATIONS ASCAT is a very accurate and stable radar Make use of these characteristics to improve models / develop new applications Future Instrument Engineering Work: Future Work on Kp: alg & stats. Future Work on Rain Forest, Stable Ice & Ocean Calibration Periodic Calibration Campaigns to monitor / correct antenna gain pattern aging if any Implications for Future Radar Wind Scatterometers Slide 15