Towards extracting fine-scale winds from synthetic aperture radar images

Tower- and aircraft-based research over the last three decades has shown that the spatial or temporal averages of radar backscatter from the ocean surface at radar wavelengths that sample the gravity-capillary part of the ocean´s surface-wave spectrum are proportional to nearly the square of the mean wind speed. During the last 15 years, analyses of satellite-based scatterometer images of the ocean surface (that is, radar images whose pixels measure 25 km to a side) show atmospheric frontal features and other, synoptic atmospheric structures. With these images, researchers have created images of the spatial distribution of wind (with extra work, they get both speed and direction) associated with those synoptic structures. Going down in scale, recent research has shown that synthetic aperture radar (SAR) images of the oceans (snapshots of the radar-backscatter patterns of the ocean surface with pixels measuring 100 m or 12.5 m to a side) manifest signatures of atmospheric turbulence-scales of motion on the order of hundreds of meters to tens of kilometers. These facts raise the following question: “(How) can one use SAR images of the ocean surface to infer fine-scale winds, that is wind patterns on scales of atmospheric turbulence?” The present work represents a step towards answering that question, by asking a related question, namely: “How much spatial averaging of backscatter is required to define a stable, convergent backscatter value for a given area, so that that value can then be translated into a wind speed?” The answer is based on the hypothesis that the statistics of radar backscatter patterns created by atmospheric turbulence should be related to the statistics of the atmospheric turbulence that generated the roughness patterns on the ocean surface