Assessment of Enivsat ASAR wind speed retrieval performance

The retrieval of high-resolution (less than 1 km) wind speed and direction from spaceborne synthetic aperture radar (SAR) imagery is critically dependent upon having an accurate measure of normalized radar cross section (NRCS) and a well-understood geophysical model function that relates wind speed and direction to NRCS. Indeed, the requirements of NRCS accuracy and precision for wind field retrievals are greater than for most other geophysical applications. The Envisat C-band dual-polarization Advanced SAR (ASAR) requires an in-orbit performance assessment of retrieval skill. It also provides an important opportunity to compare competing C-band geophysical model functions. Thus far, we have accumulated over 200 Envisat SAR images in both wide-swath and dual-polarization modes. These images are processed in a quasi-operational environment using the APL/NOAA SAR Wind Speed Retrieval System (ANSWRS). Here, we systematically compare SAR wind speed retrievals with model-based wind speed estimates. Although for any particular situation there may be substantial differences between the skill of the model and actual SAR wind field retrievals, the large number of comparisons available permit us to identify and quantify systematic differences between the performances of the two key candidate model functions: CMOD4 and CMOD5 and their polarization dependencies. We find that for wind speeds less than about 15 m/s, wind speed retrievals are very similar using both model functions, with perhaps CMOD4 agreeing best with model estimate at low wind speeds. At higher wind speeds, the CMOD5 performs substantially better. We discuss the implications of these findings for the application of SAR for operational wind speed retrieval.