Diagnosing Cotton Farmland Quality Using Multi-Temporal Remotely Sensed Data

摘要

Farmland quality is a comprehensive indicator for soil, environmental, and health quality. Using remotely sensed imagery, this study explored a method of evaluating farmland quality for cotton. Determining cotton growth conditions with multi-temporal images at the flower-boll stages, the reflectance value from LANDSAT-5 TM_4 appropriately classified cotton fields into three ranks of productivity. Our methods successfully classified 417 blocks of approximately 11 705.3 ha of fields using multi-temporal images. On Farm 148, 36.4% of the cotton fields were most productive, 34.1% were moderately productive, and 29.5% were least productive. These classifications were validated with synchronization-based soil and LAI analysis in eight cotton fields of approximately 426.5 ha. The validation showed that the main causes of low land productivity were salinity, soil texture, and soil topography. These results promote the application of remotely sensed imagery to improve the quality of cotton-growing soils and increase the efficiency of managing cotton farmlands.