Testing the Efficiency of Using High-Resolution Data From GF-1 in Land Cover Classifications

High-resolution remote sensing plays an important role in the study of subtle changes on the Earth's surface. The newly orbiting Chinese GF-1 satellites are designed to observe the Earth surface on a regional scale; however, the satellite efficiency requires further investigation. In this paper, the efficiency of using GF-1 01 satellite images to monitor a complex surface is tested by considering supplementary information and different land cover classification methods. Our work revealed that the GF-1 satellite observations can efficiently detect land cover fragments. When the support vector machine method is applied, the overall classification accuracy based on multisource data reaches 90.5%. The “salt and pepper phenomenon” is effectively reduced in classification images. These results also indicate that the accuracy of the GF-1 image classification is superior to the results when using the same method with the Landsat 8 and Sentinel-2A images, with the overall classification accuracy increasing by 23.6% and 13.6%. Our study suggests that GF-1 satellite observations are suitable for land cover studies on complex land surfaces. This approach can benefit various related fields such as land resource surveys, ecological assessments, environmental evaluations, and so forth.