1. Aquarius/SAC-D Mission Error Validation and Early Orbit Corrections Gary Lagerloef 6 th Science Meeting; Seattle, WA, USA 19-21 July 2010

2. 2 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Three Phases 1.Pre-Launch 2.In Orbit Checkout (Launch + 45 days) 3.IOC + 6 months

3. 3 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Phase 1 – Pre-Launch What: Generate a set of special case 7-day cycles –Specific cases: biases, drifts, etc, in following charts How: Use the operational mission simulator. –Start with daily files from RSS –Add spurious signal for specific case study (see below) –Input these to the mission simulator as test cases. –Distribute and analyze (science team) Timetable: –September: Produce a subset of scenarios (L2 files) –September-November: Analysis and assessment –December: Operational Readiness Review (ORR)

4. 4 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Scenario Flow Chart Daily simulated files (RSS & JPL) Mission Simulator Daily Mission Simulator L1, L2, L3 files uploaded to web site Telemetry files to/from Cordoba Level 1 > Level 2 > Level 3 Routine Processing Apply spurious signals for special case studies Generate special case telemetry files Level 1 > Level 2 > Level 3 Special Case Processing Special Case Simulator L1, L2, L3 files uploaded to evaluation products web site with documentation

5. 5 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Potential scenarios / rehearsals Simulate certain special situations to test and analyze Develop and test analysis tools Demonstrate we can address these issues at ORR 1.Insert arbitrary biases in all channels 2.Simulate arbitrary drifts in all channels 3.Orbit harmonics calibration drifts (1-4 harmonics, e.g.) 4.Simulate a complete cold sky maneuver 5.Solar flare 6.RFI 7.Attitude offsets & time tag offsets 8.Channel failure a.H or V channel b.P or M channel 9.Other suggestions … lunar, bi-static solar reflection, diffuse galaxy refl.

6. 6 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Arbitrary biases in all channels Generate test data: 1.Select a 7-day segment from the mission simulator input. 2.Add an arbitrary bias to each of the radiometer and scatterometer channels on each horn. 3.Process to L2 science data files, L3 bin-average and L3 smooth files. 4.Place the modified L1A, L2 and L3 files in an evaluation folder. Science Team Analysis: 1.Assess effects on L2 science data file TAs, U, Faraday, wind speed and SSS 2.Develop and evaluate bias detection/removal techniques, and assess uncertainties. 3.Inspect the impact on L3 files as a diagnostic tool. 4.Set up follow-on blind test with unknown biases.

7. 7 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Linear 7-day calibration drifts Generate test data: 1.Select a 7-day segment from the mission simulator input 2.Add arbitrary drift (~0.1K/ 7-days) to each of the radiometer and scatterometer channels on each horn. 3.Process to L2 science data files, L3 bin-average and L3 smooth files. 4.Set aside the modified L1A, L2 and L3 files in a special folder. Science Team Analysis: 1.Assess effects on L2 science data file TAs, U, Faraday, wind speed and SSS 2.Develop and evaluate drift detection/removal techniques, and assess uncertainties. 3.Inspect the impact on L3 files as a diagnostic tool. 4.Set up follow-on blind test with unknown drifts.

8. 8 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Harmonic orbital calibration drifts Generate test data: 1.Select a 7-day segment from the mission simulator input 2.Add arbitrary harmonic to each of the radiometer and scatterometer channels on each horn. (N cycles per orbit, N=1..3, with amplitudes ~0.1K) 3.Process to L2 science data files, L3 bin-average and L3 smooth files. 4.Set aside the modified L1A, L2 and L3 files in a special folder. Science Team Analysis: 1.Assess effects on L2 science data file TAs, U, Faraday, wind speed and SSS 2.Develop and evaluate drift detection/removal techniques, and assess uncertainties. 3.Inspect the impact on L3 files as a diagnostic tool. 4.Set up follow-on blind test with unknown drifts.

9. 9 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Phase 2 – In Orbit Checkout Aquarius Commissioning phase timeline Science Commissioning approach Critical events and Science Tasks Preliminary Acceptance Criteria

10. 10 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Early Orbit and Commissioning C-10

11. 11 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Aquarius Instrument Commissioning Timeline

12. 12 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Post-launch in-orbit checkout simulation: During the period from launch to L+25 days, the science team will compute simulated Tb and σ 0 based on the final orbit maneuvers (Science Task 1). –These data will provide “Expected Values” for each beam along-track to compare quantitatively with observations for both the engineering and science acceptance analyses. The science team will carry out an analysis sequence (Science Tasks 2-7) at each stage of the instrument turn-on sequence. Acceptance criteria are limited: –The timeline only allows for one 7-day cycle after the instrument is fully turned on. –Assess whether the data are “as expected” in qualitative terms and the sensor is “calibrate-able”. –Gross geographical and geophysical features are as expected, biases can be removed, stability is reasonable, polarization differences are appropriate, etc. (details below) Science Commissioning Approach

13. 13 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Aquarius Commissioning Science Tasks

14. 14 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Land versus ocean features are clearly evident, both in terms of brightness temperature (Tb) contrast and location. V-H polarization differences for each incidence angle are consistent with the emissivity model The 3 rd Stokes is small. Incidence angle effects such as the V-pol (H-pol) Tb increasing (decreasing) from the inner to outer beams consistent with emissivity model. Relative stability is seen over known Earth targets such as Antarctic ice, rain forests, and ocean. Scat σ 0 sensitivity to wind speed within expectations for each channel RFI detection and filtering are functional No detectable scat-radiometer interference No detectable radiometer interference from other SAC-D instruments SSS biases will be removed via ground data matchups and crossover difference analysis; initial “first-look” 7-day map will be produced We will also look for any large anomalous features that cannot be easily explained by expected calibration issues that usually occur early in a mission. Preliminary Acceptance Criteria

15. 15 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Phase 3 - IOC + 6 months Evaluate match-up data between Aquarius and in situ observing system Provide preliminary SSS bias removal and un-validated science data Monitor and analyze calibration drifts and biases Asses systematic and geographically correlated errors Analyze Scatterometer wind speed algorithm Analyze roughness effects on TH, TV and salinity algorithm And more….. Plan a science meeting for approximately IOC + 6 months Asses calibration and validation results Approve algorithm updates Reprocess the first six months of data.

16. 16 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Approaches to De-biasing Each radiometer needs to be independently calibrated First try to remove the gross offsets in the initial retrieved SSS –Use reference salinity field globally –AVDS matchups – tabulate for each beam –Cross-over analysis to remove residual inter-beam biases With time, work through the re-calibration of retrieval coefficients Buoy Obs. Search Radius Match-up processing One match-up file per buoy observation contains all the satellite data in the search radius Subsequent processing to generate the optimal weighted average satellite observation per buoy Global tabulations Match-up processing One match-up file per buoy observation contains all the satellite data in the search radius Subsequent processing to generate the optimal weighted average satellite observation per buoy Global tabulations

17. 17 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Remove orbit errors and biases Beam 2 SSS (with constant and harmonic errors added) Adjusted Beam 2 SSS (with constant and harmonic errors removed) Beam 1 SSS (with harmonic errors added) Adjusted Beam 1 SSS (with harmonic errors removed) Cross over difference analysis to remove systematic errors between different beams

18. 18 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Validation Working Group Review activities for these three phases –Pre-lauch –IOC –IOC + 6 months Focus in ideas of the third phase

19. 19 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Validation - Theory The satellite salinity measurement S S and the in situ validation measurement S V are defined by: S S = S ± ε S S V = S ± ε V where S is the true surface salinity averaged over the Aquarius footprint area and microwave optical depth in sea water (~ 1 cm). ε S and ε V are the respective satellite and in situ measurement errors relative to S. The mean square of the difference ∆S between S S and S V is given by: = + where denotes the average over a given set of paired satellite and in situ measurements, and =0. Our objective is to validate that = − ≤ ε R 2 where ε R is the allocated rms error requirement. The satellite measurement error is the difference between and mean square in situ error.

20. 20 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Validation Errors ε V is the rss of several terms: ε P is the difference error between a point salinity measurement and the area average over the instantaneous satellite footprint (log-normal distribution, median ~0.05 psu, extremes ~0.5). ε O is an error due to the temporal or spatial offset between the satellite and in situ samples (many buoys surface once every 10 days and will be paired with the nearest satellite pass). ε Z is the difference error between the skin depth (~1-2 cm) salinity and in situ instrument measurement generally at 0.5 m to 5 m depth (can be >1 psu in rain). ε C is the in situ sensor calibration error, usually very small (<0.05 psu) ε V is the rss of several terms: ε P is the difference error between a point salinity measurement and the area average over the instantaneous satellite footprint (log-normal distribution, median ~0.05 psu, extremes ~0.5). ε O is an error due to the temporal or spatial offset between the satellite and in situ samples (many buoys surface once every 10 days and will be paired with the nearest satellite pass). ε Z is the difference error between the skin depth (~1-2 cm) salinity and in situ instrument measurement generally at 0.5 m to 5 m depth (can be >1 psu in rain). ε C is the in situ sensor calibration error, usually very small (<0.05 psu)

21. 21 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA ε P = difference error between a point measurement and the area average Ship tracks spatially filtered and differenced with point measurements

22. 22 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA ε P = difference error between a point measurement and the area average Log normal ε P ~ 0.05 psu Log normal ε P ~ 0.05 psu

23. 23 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA ε O = error due to temporal or spatial offset between the satellite and in situ samples Spatial structure function from ship track data Distance from footprint center Nomin ally ~0.2 psu

24. 24 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA ε Z = difference error between the skin depth (~1- 2 cm) salinity and in situ instrument TOGA/COARE R/V Wecoma Skin (2cm) vs 2m and 5m depths during heavy rain TOGA/COARE R/V Wecoma Skin (2cm) vs 2m and 5m depths during heavy rain 100 km

25. 25 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Argo Enhanced SSS Float Trials Purpose: To obtain “skin” salinity and upper 5m gradient statistics Argo CTD nominally shuts off at ~5m Steve Riser and Gary Lagerloef are testing experimental Argo floats each with a secondary CTD sensor to profile to the surface. The primary CTD will shut off at ~5 m per normal operations. Sea-Bird developed a specialized “Surface Temperature Salinity” (STS) sensor which is programmed to profile the upper ~30 m and is inter-calibrated with the primary CTD We deployed the first at the HOT site near Hawaii late summer 2007. 20 more have been funded by NASA and deployed in the past 2 years. (S.Riser) New version funded including acoustic rain and wind measurements (S.Riser)

26. 26 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA Validation Approach Match co-located buoy and satellite observations globally. Account for various surface measurement errors. Sort match-ups by latitude (SST) zones. –Validate that the error allocations are met for the appropriate mean number of samples within the zone, or –Calculate global rms over monthly interval The Current Best Estimate (CBE) includes instrument errors plus all geophysical corrections such as surface roughness, atmosphere, rain, galaxy, solar, …

27. 27 19-21 July 2010 Error Validation & Early Orbit Corrections – Lagerloef 6rh Science Meeting, Seattle, WA, USA