Plastic-mulched Farmland Recognition in Loess Plateau Based on Sentinel-2 Remote-sensing Images

Abstract

Identification of mature crops plays an important role in crop area estimation, agricultural production, and yield statistics. To provide a simple method for identification of crop at mature stage with remote sensing, the method based on normalized difference vegetation index (NDVI)-normalized spectral separation index (NSSI) space and Hue saturation value (HSV) transformation was proposed by using Sentinel-2A data. Fengyang County located in Chuzhou City, Anhui Province, was selected as study area. The pure pixel of photosynthetic vegetation, non-phototrophic vegetation and bare soil were estimated by using the NDVI-NSSI space firstly. Coverage of non-phototrophic vegetation and the distribution of mature crops were estimated by using spectral mixture analysis method. As wheat and rape had the same maturity period in the study area, HSV transformation was used to estimate the distribution of rape by using the Sentinel-2A data at the flowering stage of rape, and the mature wheat and rape was identified. 

Keywords: mature crops, remote sensing recognition, non-photosynthetic vegetation, NDVI-NSSI space, mixe pixed decomposition