EVALUATION OF RESOUCESAT-1 AWIFS DATA FOR PRODUCING AN AGRICULTURAL CROP INVENTORY FOR CANADA

摘要

Agriculture and Agri-Food Canada (AAFC) is currently developing sensor-independent methods to produce an annual agricultural crop inventory from earth observation data. Classification of croplands requires satellite data that is both spatially extensive, and provides sufficient spatial resolution to discern field-scale information. These two needs are generally in conflict since the choice of input data from commercial satellites represents a trade-off between spatial resolution and swath coverage. The Indian Remote Sensing (IRS) satellite Resourcesat-1 (IRS-P6) collects multi-spectral data from its Advanced Wide Field Sensor (AWiFS) instrument with a nominal spatial resolution of 56 meters at nadir, and a swath size of 740 kilometers. AWiFS data has been utilized successfully over several growing seasons by the United States Department of Agriculture (USDA), but this sensor has not been widely used in a Canadian context. This study will compare the use of multiple multi-spectral optical satellites, including SPOT, Landsat-TM and AWiFS, for crop mapping over three Canadian agricultural sites: an agricultural site in Eastern Ontario, a site in the Red River valley, Manitoba, and a site near Swift Current in southwestern Saskatchewan. These locations represent different crop mixes, as well as differences in the relative size of agricultural fields. The impact of differences in spatial and spectral resolution on classification accuracy will be examined. In addition, the study will consider the use of each sensor type as well as the potential to use these sensors in conjunction with microwave sensors, such as Envisat-ASAR. Landsat-TM has historically provided reliable information at a sufficient scale for most land cover/land use applications. With expected gaps in the Landsat program, it is essential that alternative sensors be explored for this type of application.