Toward an optimal inversion method for synthetic aperture radar wind retrieval

Article English OPEN
Portabella, M. ; Stoffelen, A. ; Johannessen, Johnny A. (2002)
  • Publisher: American Geophysical Union
  • Related identifiers: doi: 10.1029/2001jc000925
  • Subject: Meteorology | Remote sensing | Mathematical Geophysics
    arxiv: Physics::Atmospheric and Oceanic Physics

In recent years, particular efforts have been made to derive wind fields over the oceans from synthetic aperture radar (SAR) images. In contrast with the scatterometer, the SAR has a higher spatial resolution and therefore has the potential to provide higher resolution wind information. Since there are at least two geophysical parameters (wind speed and wind direction) modulating the single SAR backscatter measurements, the inversion of wind fields from SAR observations has an inherent problem of underdetermination. Moreover, this modulation is highly nonlinear, further complicating the inversion. Lorenc [1986] presented a general statistical approach to solve inversion problems (including underdetermined problems) in meteorological analysis. We propose a SAR wind retrieval method based on this general approach. This simplified method combines the SAR information with some background information coming from high-resolution limited area model to retrieve the most probable wind vector, assuming that all sources of information contain errors and that these are well characterized. We then evaluate two different SAR wind retrieval methods. The first one is commonly used by the SAR community and is based on a combination of a wind streak detection algorithm (wind direction retrieval) and a C band model inversion (wind speed retrieval). The second one is the new method we propose, based on the general statistical approach. We show the potential problems and limitations of using any of these methods and show how the second method can potentially contribute to a significant improvement in SAR wind retrieval. The new method prepares the ground for the assimilation of SAR data in high-resolution numerical weather prediction models.
Share - Bookmark