| In this study,Aksu Prefecture and Bortala Mongol Autonomous Prefecture were selected as examples.Through the comparison and analysis of different data sources,the Landsat 8 and GF-1 satellite images were selected to study the crop identification in Xinjiang agricultural region,and then the crop distribution in the region was obtained.Crop identification research zoning using meteorological data.According to the statistical data,the main crops and the sown area were obtained,and the planting area of cotton,corn,wheat and rice were determined.The key phenological periods of crops in different regions were obtained.The Normalized Difference Vegetation Index(NDVI)sequences were obtained from remote sensing images to determine the optimal interpretation time for each crop.The results showed that:(1)Landsat 8 data and GF-16 m data can complement each other because of different return visit cycle,and can participate in the research and analysis together.The results show that the Landsat 8 and GF-16 m data can be used to identify the crop planting structure in county and large area.(2)In the research of crop classification in large area,the process of crop classification is optimized based on meteorological data and crop phenological information.(3)Obtaining the NDVI time series of the key phenological period to get the difference among crops.The best times to identify Aksu Prefecture crops were:July for cotton,April for wheat,and July for corn and rice.The best time to identify crops in Bozhou is late August for cotton,July for corn,late April,mid May for wheat and June for rice.(4)Using the requirement of crop accumulated temperature,crop spectral characteristics and Ndvi time series,at the same time,using the masking film extracted from the map of land use status to construct decision-making,to classify crops.The precision of classification is over 75%,and the precision of total area is up to 95%.The decision tree can be used to classify many crops in a large area. |
PDF Full Text Request