Cotton area estimation using muti-sensor RS data and big plot survey in Xinjiang

An accurate agricultural statistics act as an important source of decision making to decision makers. Cotton area is very important especially to Xinjiang, which had 3.2 million tons in cotton output for 2012, accounting for half of the national yield. Although the potential of remote sensing to facilitate agriculture management has been explored but still there are grounds of improvement which can be explored to add some tools and process to a more accurate remote sensing based agriculture management system in any area. This research presents a method to estimate cotton area in Xinjiang. Many field data had been accumulated including each main crop through the Big Plot Survey from 2011 to 2013. Meanwhile the MODIS NDVI product were used to build reference NDVI time series, according to the Big Plot Survey result. Higher spatial resolution RS data (Landsat-8 and HuanJing-1) merged NDVI series in 2013. Through the relative NDVI correction, the 30m merged muti-source NDVI series could be classified with sub-region classification method. Gaofen-1 satellite data were employed to validate the cotton distinguish result, which has 2m spatial resolution. Result shown our workflow is efficiency, and the total area of cotton is 1.692 million hectare in Xinjiang 2013.