1. 6 Moon glint in AMSU Nigel Atkinson and Roger Saunders Met Office (UK)

2. 7 Moon glint in AMSU Moon passes through space calibration view in AMSU- A and AMSU-B at certain seasons of year Typically affects ~10 minutes per orbit for ~10 consecutive orbits Tends to affect tropics (inclination of moon's orbit is low) Magnitude: 2K for AMSU-A; 30K for AMSU-B Errors maximum at cold scenes; zero for ambient scenes. Various strategies have been devised to correct for this.

3. 8 Impact on Earth view radiances Insert cardiogram here No data Moon in space view sdev bias AMSU-A channel 8 Global O-B stats for 2001 degK

4. 9 AMSU-B 90 Earth views 4 space views (within SV3) Hot Target View Edge of Earth’s disk Platform AMSU-B has 4 space view samples, each separated by 1º. Normally only 1 or 2 views are contaminated (occasionally 3) Moon

5. 10 AMSU-B moon-glint corrections NESDIS 1B files: A moon glint correction was introduced in December 2000 (following Met Office recommendation). If the space views fail the normal consistency checks, the 2 or 3 samples with the lowest space view counts are used. AAPP local 1B files: Contaminated data fail the consistency checks and are flagged 'do not use' Code is available for AAPP to implement the NESDIS correction. However, at mid-latitudes the problem is rare.

6. 11 AMSU-A AMSU-A has only one space view. Therefore normal quality checks are passed. Moon glint events are detectable as a degradation in NWP model statistics. Met Office has implemented prototype 'patch' to AAPP: –Implemented at 1B-1C converter, thus works on global and local datasets, and also on historical data. –Position of moon is calculated using standard astronomical formula, and hence angle between moon and space view –Space view data rejected if angle <3.5º. Gain interpolated between end points. Use offset from hot black body. –Zero error at ambient scenes; <0.1K at 200K.

7. 12 Met Office interpolation scheme 290K 3K Raw Corrected Plots show brightness temperature for fixed raw count, i.e. calibration variation around an orbit Hot target Space view

8. 13 NESDIS scheme NESDIS are working on a correction scheme at 1B level Involves correcting the measured space view readings on the basis of pre-launch antenna patterns and known moon surface temperature No interpolation, so consistent with the architecture of the NESDIS pre-processor NESDIS are using the method to derive post-launch antenna patterns

9. 14 Results Correction Algorithm Works Well on A1-2 using pre-launch antenna pattern. Ripples on A1-1 and A2 A2 A1-2

10. 15 Post-launch antenna pattern measurement NESDIS interpolation using Smooth Curve

11. 16 Preliminary NESDIS Results NOAA-16 AMSU-A Pre-launchPost-launch

12. 17 Conclusions The moon can affect AMSU calibration and hence Earth view radiances for NWP assimilation For AMSU-B correction code is ready to be included in AAPP. Used at Met Office. For AMSU-A Met Office interpolation scheme implemented operationally and tested with several events. Could easily be included in AAPP. The NESDIS scheme could in principle be implemented in AAPP also, but would require more effort.