EOS Validation Program

Institution: College of Oceanic and Atmospheric Sciences
                    Oregon State University
                    104 Ocean Administration Bldg
                    Corvallis, OR 97331-5503
Phone: (541) 737-2748
E-mail: mhf@coas.oregonstate.edu

February 2003

An investigation is proposed to conduct detailed validation of the AMSR-E and ADEOS-AMSR wind speed data sets. The analyses will involve comparisons between AMSR wind speed estimates and those from operational ocean moored buoys (National Data Buoy Center and TOGA/TAO), the SeaWinds microwave scatterometer instruments on QuikSCAT and ADEOS-II, and operational surface wind analyses produced by ECMWF and NCEP. The random component error model of Freilich (1997) and the statistical approach of Freilich and Vanhoff (2001) will form the basis for the validation analysis -- although derived originally for scatterometer validation, the approach uses only wind speed information and is thus applicable to AMSR wind data. The formalism explicitly accounts for (and solves for the magnitudes of) equivalent random component errors in the comparison data sets as well as the AMSR measurements to be validated; it is thus well suited for the use of satellite and operational surface wind analyses as comparison data. The effects of compatibility errors between the satellite and comparison data will be quantified and minimized for both the buoy and the ECMWF/NCEP data sets.

The proposed investigation will build on advanced validation analysis techniques, data acquisition and reduction tools, and archives developed by the PI in previous studies of spaceborne microwave wind data sets; complement the baseline AMSR wind validation (conducted by the AMSR Oceans Team) by using an alternate error model and a formalism that explicitly accounts for random errors in all data sets; extend previous studies by using temporal averaging of buoy measurements and spatial averaging of AMSR measurements (when compared with ECMWF/NCEP analyses) to minimize, and quantify the effects of, resolution-induced compatibility errors; and provide regional and global ocean estimates of systematic gain, offset, and random uncertainty for AMSR wind speeds both at the 25 km resolution of the raw data and at the 100-400 km resolution of operational surface wind analyses.

Initial analyses and algorithm refinements will be conducted using existing available radiometer data from SSM/I and scatterometer data from SeaWinds on QuikSCAT. AMSR Level 2 data will be used as soon as preliminary wind speed data sets become available. Results will be made available to the AMSR Instrument Team for their use in post-launch geophysical algorithm refinement.