Vegetation classification using seasonal variations of sigma-0

摘要

Abstract: Scatterometers were originally designed and employed based on their proven ability to measure near-surface winds over the ocean. However, they are also providing useful in global and regional studies of vegetation and soil moisture. In this paper we examine the C-band and Ku-band radar signatures of vegetation over South America using data for the ERS-1/2 and NSCAT scatterometers. We compare the seasonal response of various types of vegetation based on both Matthew's classification and the University of Maryland AVHRR-based classification and use the seasonal response in a simple vegetation classification experiment. A time-series of enhanced resolution $sigma degree images are generated for both sensors. The classifier is used to classify the vegetation coverage of each pixel in the image frame into broad classes of vegetation type. Considering the accuracy and resolution limitations of the reference vegetation maps, the classifications result exhibit a high degree of accuracy and consistency with the primary confusion observed between related vegetation classes of similar vegetation canopy density. !12