Sub-pixel Mapping of Croplands in China: Using Multi-temporal VEGETATION Sensor Data and Spectral Mixture Analysis
Over the last few decades large-scale mapping and monitoring of agriculture has primarily been dependent upon the remotely sensed data from the NOAA AVHRR sensors at 1-km spatial resolution. At 1-km spatial resolution, however, the land surface is mostly a mixture of various land cover types, including vegetation, soil and water. Early studies that explored uses of spectral mixture analysis and AVHRR data for quantification of percent fractional cover within a pixel have had limited success, primarily because AVHRR has too few spectral bands. In this study, we explored the potential of the VEGETATION (VGT) sensor on the SPOT 4 satellite for sub-pixel mapping of agriculture in China. VGT represents one of a new generation of optical sensors designed for enhanced observation of land and vegetation. VGT has 4 spectral bands: blue (430-470 nm), red (610-680 nm), near-infrared (780-890 nm), and mid-infrared (1580-1750 nm); and it provides daily global coverage at 1-km spatial resolution. Multi-temporal 10-day composites of VGT data from March 1999 to May 1999 were used for spectral mixture analysis to derive four endmembers (cropland, forest, water and urban/soil) in eastern China. Fractional cropland estimates from the VGT-based spectral mixture analysis were evaluated using (1) point-level land cover data from field surveys in 2000, (2) landscape-level cropland data derived from digital classification of a Landsat TM image acquired in 1996; (3) province-level agricultural field survey data from the early 1980s; and (4) province-level cropland data derived from visual interpretation of Landsat TM images acquired in 1995/1996. The preliminary results show that VGT data are very useful for sub-pixel mapping of agriculture in China. This study provides a sound basis to explore daily VGT data in 2000 for sub-pixel mapping of agriculture at the country to continental scales, as part of our contribution to the VEGA2000 Initiative effort.
Joint Assembly Meeting, American Geophysical Union
American Geophysical Union Publications
Xiao, X., Boles, S., Liu, J., Frolking, S., Salas, W., Li, C. and Moore, B., Sub-pixel Mapping of Croplands in China: Using Multi-temporal VEGETATION Sensor Data and Spectral Mixture Analysis, EOS Trans. Suppl. 82, Spring meeting. AGU Abstract B51A-10.