Application of GEE in cotton monitoring of the 7th division of Xinjiang Production and Construction Corps

Abstract

Cotton is a very important economic crop, which is not only an important strategic material related to the national livelihood, but also brings a lot of economic benefits and occupies an important position in the national economy. Cotton production in Xinjiang accounts for more than 80% of China’s cotton production, accounting for nearly one quarter of global cotton production. therefore, to do a good job in different growth cycles of cotton planting area, planting areas, cotton growth time changes and other basic information, for agricultural production departments and cotton farmers to make economic decisions and cotton planting structure adjustment is essential. At the same time, do a good job including cotton census and important crop growth monitoring, disaster warning, disaster assessment and post-disaster recovery work is also very important. And in recent years, with the continuous expansion of cotton cultivation area, scope and scale in Xinjiang, cotton is becoming increasingly important in the region’s economic and social development. Timely, accurate, and efficient acquisition of real-time cotton growth, monitoring of disaster conditions, acquisition and display of the extent and scope of damage, and post-disaster assessment in Xinjiang are of great importance for maintaining regional economic and provincial development and ensuring people’s well-being. Based on Google Earth Engine remote sensing big data cloud computing platform and high spatial and temporal resolution remote sensing image data Sentinel-2, this study takes Kuitun Reclamation Area of the Seventh Division of Xinjiang Production and Construction Corps, a typical cotton growing area in Xinjiang, as an example, and uses continuous time series image data combined with various indices (such as commonly used normalized vegetation index NDVI, normalized moisture index NDWI and cotton chlorophyll data SPAD), a time series analysis model was constructed to analyze the cotton growth conditions in different growing seasons in the study area. Based on the time series remote sensing data, the cotton change patterns and key temporal phase data of cotton in different months were extracted by the time series analysis method, and combined with the unique spectral information and texture features of cotton, the current mainstream high-precision threshold segmentation classification method was used to accurately and quickly extract the cotton planting area, planting range and cotton growth conditions in different growing stages, and the analysis of each feature information on the The contribution of each feature information to cotton extraction was analyzed. Combined with the ground real sampling data, the accuracy of cotton planting area and planting range extraction based on remote sensing images was verified, and the overall classification accuracy was above 90%, which is of great practical significance for the extraction and analysis of cotton information in this region. The online real-time big data processing engine, combined with a large number of remote sensing images with high spatial and temporal resolution, quickly and accurately identifies and extracts typical cotton planting areas in Xinjiang, which is of great significance and value to regional social development and crop extraction.