ASL Cal L. Strow UMBC Introduction AIRS L1C Frequency - PowerPoint PPT Presentation

ASL ν Cal L. Strow UMBC Introduction AIRS L1C Frequency Calibration Raw Data Model Fit M3 versus M10 Summary L. Larrabee Strow and Scott Hannon Physics Department and Joint Center for Earth Systems Technology University of Maryland Baltimore County (UMBC) Airs Science Team Meeting - Passadena - CA May 2, 2009 1 / 27

ASL Overview ν Cal Update from last October Science Team Meeting L. Strow UMBC Cross-correlation technique works well for several modules, except over poles. Introduction Raw Data Fall 2008: calibrated whole mission using M12 and Model Fit cloud-cleared radiances M3 versus M10 M12 failed, to some degree, over poles (only Q-branch had Summary contrast). Did not use internal QA. This presentation: Calibrated whole mission using M3 (water, 1400 cm − 1 ) and M10 (CO 2 , 750 cm − 1 ), retained and used internal QA (B(T) contrast). Goal: Make frequency calibration a “non-issue” for AIRS climate applications. 2 / 27

ASL Frequency Calibration ν Cal Use a cross-correlation technique on M3 and M10 for ν L. Strow calibration UMBC Cross-correlate L2CC radiances with Calc radiances. Calcs Introduction Raw Data done using AIRS L2 retrievals. Model Fit Careful selection of channels M3 versus M10 One ν calibration per granule. Summary Units: using “micron” shift in focal plane. Shifts referred to as “yoffsets” 1 micron ∼ 1% of an SRF FWHM Yoffsets measured relative to TVAC values, generally had ∼ -14 µ m shift at launch, with about 1 µ m during rest of mission. 3 / 27

∆ B(T) for a dx = 1 µ m ASL 1 µ m Equal to Mission Shift, Orbital ∼ 0.4 µ m ν Cal L. Strow UMBC 0.024 0.022 Introduction 0.02 Raw Data 0.018 Δ ν for yoffset = 1 μ m Model Fit 0.016 M3 versus M10 0.014 0.012 Summary 0.01 0.008 0.006 0.004 500 1000 1500 2000 2500 3000 Wavenumber (cm − 1 ) 4 / 27

ASL Frequency Calibration Model ν Cal Raw ν (Yoffset) calibrations were binned by 2 deg. in orbit L. Strow phase, giving 180 data sets, each one is fit to the following UMBC expression: Introduction Raw Data 3 Model Fit � y ( t ) = y o − b 1 exp ( − t /τ) + [ a i sin ( 2 π t + φ i )] M3 versus M10 i = 1 Summary Fast time behavior is orbit phase (latitude), parameterized by 180 values for b 1 , τ , three a i harmonic terms, and their phases, φ i . Slow time behavior is due to seasonal and slower effects. Data first averaged over 16 day time periods for fits to the above equation. 5 / 27

M3 Raw Calibration (per granule) ASL With, and without B(T) Contrast Filter ν Cal L. Strow No QA UMBC BT Contrast > 10K − 13 Introduction Raw Data Model Fit M3 versus M10 − 13.5 Summary Yoffset ( μ m) − 14 − 14.5 − 15 2004 2005 2006 2007 2008 2009 Time 6 / 27

M10 Observations (per granule) ASL With, and without B(T) Contrast Filter ν Cal L. Strow UMBC Introduction − 13 Raw Data Model Fit Yoffset Obs − Cal ( μ m) M3 versus M10 − 13.5 Summary − 14 − 14.5 − 15 2004 2005 2006 2007 2008 2009 Time 7 / 27

OLD ν Calibration ASL Binned by 2 deg in latitude, 16 days in time ν Cal L. Strow UMBC Introduction Raw Data Model Fit M3 versus M10 Summary 8 / 27

M3 Yoffset vs Orbit Phase ASL Binned by 2 deg in latitude, 16 days in time ν Cal yoff ( μ m) L. Strow 350 UMBC − 13.3 Introduction − 13.4 300 Raw Data − 13.5 Model Fit 250 M3 versus M10 Orbit Phase (deg) − 13.6 Summary − 13.7 200 − 13.8 150 − 13.9 − 14 100 − 14.1 50 − 14.2 − 14.3 2005 2007 Time 9 / 27

Tropical M3 Calibration ASL Further binning to ± 30 deg ν Cal − 13.2 L. Strow Decending UMBC Ascending Introduction Difference − 13.4 Raw Data Model Fit − 13.6 M3 versus M10 Summary Yoffset ( μ m) − 13.8 − 14 − 14.2 − 14.4 2005 2007 Time 10 / 27

ASL Polar M3 Calibration ν Cal L. Strow − 13.3 Decending UMBC Ascending − 13.4 Introduction Difference Raw Data − 13.5 Model Fit − 13.6 M3 versus M10 Summary − 13.7 Yoffset ( μ m) − 13.8 − 13.9 − 14 − 14.1 − 14.2 − 14.3 2005 2007 Time 11 / 27

Yoffset versus Latitude ASL Averaged over 5 Years ν Cal − 13.7 L. Strow Descending UMBC Ascending Introduction − 13.75 Raw Data Model Fit − 13.8 M3 versus M10 Summary Yoffset ( μ m) − 13.85 − 13.9 − 13.95 − 14 − 14.05 − 100 − 80 − 60 − 40 − 20 0 20 40 60 80 Latitude 12 / 27

M3 Monthly Avg. Yoffset vs Orbit Phase ASL Back to raw binned data but aggregated by month (for all years). ν Cal L. Strow North Pole South Pole yoff ( μ m) UMBC 12 − 13.6 Introduction 11 − 13.65 Raw Data 10 Model Fit − 13.7 M3 versus M10 9 − 13.75 Summary 8 − 13.8 Month 7 − 13.85 6 − 13.9 5 − 13.95 4 − 14 3 − 14.05 2 − 14.1 1 50 100 150 200 250 300 350 Orbit Phase (deg) 13 / 27

ASL Fit to our Parameterization Equation ν Cal Will will now examine reasonableness of the fitted parameters. L. Strow Reminder: UMBC Introduction 3 Raw Data � y ( t ) = y o − b 1 exp ( − t /τ) + [ a i sin ( 2 π t + φ i )] Model Fit i = 1 M3 versus M10 Summary 14 / 27

Amplitude of Sinusoidal Terms ASL These are the a 1 , a 2 , a 3 terms for each 2-deg bin of orbit phase. ν Cal 0.14 L. Strow UMBC ω Introduction 0.12 2* ω Raw Data 3* ω Model Fit 0.1 M3 versus M10 Summary Amplitude ( μ m) 0.08 0.06 0.04 0.02 0 0 50 100 150 200 250 300 350 Orbit Phase (deg) 15 / 27

Yoffset Decay Time Constant vs Latitude ASL τ for both M3 and M10 versus Latitude ν Cal 3 L. Strow UMBC 2.8 Introduction Raw Data 2.6 Model Fit τ (years) M3 versus M10 2.4 Summary 2.2 Yoffset Decay Rate 2 1.8 M3 M10 1.6 1.4 Δ τ of 1 year ≡ 0.2 μ m after 5 years 1.2 1 − 100 − 80 − 60 − 40 − 20 0 20 40 60 80 Latitude 16 / 27

Yoffset Decay Amplitude vs Latitude ( b 1 Terms) ASL � 3 y ( t ) = y o − b 1 exp ( − t /τ) + i = 1 [ a i sin ( 2 π t + φ i )] ν Cal L. Strow M3 UMBC − 1.8 M10 Introduction Raw Data − 2 Model Fit μ m) M3 versus M10 Yoffset Decay Amplitude ( − 2.2 Summary − 2.4 − 2.6 − 2.8 − 3 − 3.2 − 100 − 80 − 60 − 40 − 20 0 20 40 60 80 Latitude 17 / 27

Obs - Fitted Calibration, M3 ASL Mission through 2009+ ν Cal L. Strow − 13.2 Obs UMBC Calc Introduction − 13.4 Raw Data Model Fit M3 versus M10 − 13.6 Summary Yoffset ( μ m) − 13.8 − 14 − 14.2 − 14.4 2004 2005 2006 2007 2008 2009 Time 18 / 27

Obs - Fitted Calibration, M3: 2007 ASL 1-year ν Cal L. Strow − 13.5 Obs UMBC Calc − 13.6 Introduction Raw Data − 13.7 Model Fit M3 versus M10 − 13.8 Summary Yoffset ( μ m) − 13.9 − 14 − 14.1 − 14.2 − 14.3 − 14.4 Q1 − 07 Q2 − 07 Q3 − 07 Q4 − 07 Q1 − 08 Time 19 / 27

Obs - Fitted Calibration, M3: Jan. 08, 2007 ASL 1-day ν Cal L. Strow − 13.6 UMBC Introduction − 13.7 Raw Data Model Fit M3 versus M10 − 13.8 Summary Yoffset ( μ m) − 13.9 − 14 − 14.1 Obs − 14.2 Calc 00:00 06:00 12:00 18:00 00:00 Time 20 / 27

Histogram of M3 Obs-Calc Good FOVs ASL Errors are Gaussian; Bias = 0.00 µ m, Std = 0.06 µ m ν Cal 6000 L. Strow UMBC Introduction 5000 Raw Data Model Fit M3 versus M10 4000 \# of Observations Summary 3000 2000 1000 0 − 0.5 0 0.5 Yoffset Obs − Cal ( μ m) 21 / 27

ASL Histogram of M3 Obs-Calc for Outlier FOVs ν Cal 1500 L. Strow UMBC Introduction Raw Data Model Fit M3 versus M10 1000 \# of Observations Summary 500 0 − 1.5 − 1 − 0.5 0 0.5 1 Yoffset Obs − Cal ( μ m) 22 / 27

Difference Between M3 and M10: ASL Raw Data and Fit ν Cal L. Strow M3 − M10 Obs UMBC 0.5 M3 − M10 Calc Introduction 0.4 Raw Data Model Fit 0.3 M3 versus M10 μ m) 0.2 Summary Δ Yoffset Obs − Cal ( 0.1 0 − 0.1 − 0.2 − 0.3 − 0.4 − 0.5 2004 2005 2006 2007 2008 2009 Time 23 / 27

ASL Histogram of M3 minus M10 Obs ν Cal 4 3 x 10 L. Strow UMBC Introduction 2.5 Raw Data Model Fit M3 versus M10 2 \# of Observations Summary 1.5 1 0.5 0 − 0.4 − 0.2 0 0.2 0.4 0.6 M3 − M10 Yoffset ( μ m) 24 / 27

ASL Variation of M3 and M10 Yoffset: Tropics ν Cal L. Strow M3 − Des UMBC − 13.5 M3 − Asc Introduction M10 − Des − 13.6 M10 − Asc Raw Data Model Fit − 13.7 M3 versus M10 − 13.8 Summary Yoffset μ m − 13.9 − 14 − 14.1 − 14.2 − 14.3 − 14.4 2004 2005 2006 2007 2008 2009 Time 25 / 27

Other Arrays ASL Use tropics for offsets ν Cal L. Strow UMBC Introduction Raw Data Model Fit M3 versus M10 Summary 26 / 27

ASL Conclusions ν Cal M3 appears to be the best module for frequency L. Strow calibration UMBC Behavior is somewhat complicated, but reasonable Introduction Raw Data Use tropics to determine static offset between M3 and Model Fit other modules. Higher errors in some modules reflect M3 versus M10 lower requirement for knowlege of the module frequency. Summary Some concern that some modules may move differently. Will evaluate fitting parameters in tropics for almost all modules. Almost ready for V6 implementation. τ term should predict future drifts unless instrument is stressed. 27 / 27