Spaceborne wind scatterometers are an important element in future remote sensing
systems because of their proven ability to make all-weather measurements of
vector winds over the ocean, a capability first demonstrated by the
Ku-band (14.6 GHz) Seasat scatterometer (SASS) in 1978 [14].
Scatterometers measure the normalized radar backscatter coefficient ()
of the earth's surface. Over the ocean, the
measurements can be used to
estimate the near-surface wind [14].
The coarse resolution (nominally 50 km) of the scatterometer measurements, while
suitable for ocean wind measurement, is a significant limitation in the
application of scatterometer data to land and ice studies. The principle
application of land measurements has been in the calibration of the
scatterometer using tropical forests as homogeneous targets
[10][3]. However, global studies (e.g.,
[13][5][4]) have indicated that the
measurements
are very sensitive to vegetation and physiographic type, and may provide
valuable information for discriminating between land cover classes.
In a separate paper [7] (hereafter LHW) we introduced a new method for obtaining enhanced resolution radar images from low resolution scatterometer measurements. In this paper we use the method to generate enhanced resolution images of the extended Amazon basin. Using these medium-scale SASS-derived images, a series of experiments in vegetation classification were conducted. We conclude that the medium-scale enhanced resolution images can be very useful in studies of tropical vegetation.
This paper is organized as follows: We first briefly provide some background in scatterometry theory, a description of the SASS data used, and an overview of our enhanced resolution imaging technique. (A detailed description of the enhanced resolution imaging technique is presented in LHW). Data considerations and assumptions required to apply the technique to SASS data are then discussed. We present enhanced resolution radar images of the extended Amazon basin generated from the SASS data and then provide the results of several vegetation classification experiments based on the images. We consider both vertically and horizontally polarized measurements. Finally, we summarize our results.