GOES Imager Visible Channel Calibration using Lunar and
GOES Imager Visible Channel Calibration using Lunar and Stellar Observation data Xiangqian Wu and Fangfang Yu 04/22/2010
GOES Imager Lunar Calibration • Methodology: – Measured Moon irradiance compared with the USGS lunar model – Absolute calibration – Wu, X et al. 2006. Vicarious calibration of GOES Imager visible channel using the Moon, Proc. SPIE 6296. EGOES: GOES measured irradiance Ri: Radiance from pixel I ω: Solid angle N: number of moon observation pixels 04/22/2010 S: Prelaunch slope (reciprocal of instrument gain) Ci, R: Raw count of pixel I CS: Space count
Error Sources • Source of Uncertainties, assuming invariable SRF – illuminated moon area • Edge at sub-pixels – Detector responsivity variations • Reference detector – Scan angle effect • ~2. 2% based on MIT Lincoln Lab – USGS model • • 1% for geometric uncertainty 5 -10% for absolute uncertainty – Others (e. g. straylight/earth shine/? ) 04/22/2010
Performance Evaluation Independ ence Stabili ty Traceabi lity Precision Availabili ty Latenc y Cost Total Moon 5 5 3 (? ) 2 3 3 3. 5 3. 45 Star 5 5 2 3 1 3 3 3. 15 3. 30 Weighting reanalysis operation 15% 10% 20% 15% 20% 04/22/2010 Scores are comparative. 1 = Worst, 5 = Best
GEO-GEO/LEO Inter-calibration • Moon – Common and unmatched stable reference to all the spacecrafts • How often does the satellite look at the moon? • Measure the Moon at the similar phase angle? – What is needed? • USGS model results • Satellite measurements of the Moon Sat 1 - > Moon: Sat 2 - > Moon: Sat 1 - > Sat 2: 04/22/2010 R 1 = a * R 2 + b
Comments on Error Analysis (1/2) • Counted Fully – e. g. , if target selection (clear, aerosol) is a step in the algorithm give it an error budget, say 10 -9 • Reassure that this has been considered • Provide a baseline for future revision or refinement – Depends on season – Variation among target • Attention to correlation – e. g. , wind speed and surface pressure may be negatively correlated • Experience with star calibration – Results inconsistent • Star 1: degradation 6%± 0. 01% • Star 2: degradation 4%± 0. 01% – Suspect that error and/or correlation has not been not fully accounted. 04/22/2010
Comments on Error Analysis (2/2) • Similar error analysis has been widely used in engineering – Currently best / worst / most probable estimate – Can do better or worse, in connection with budget and schedule • Why do we do it? – Tells us what is good enough – Help to choose validation methodology and / or evaluate validation results 04/22/2010
- Slides: 7