Monitoring Suspected Pollution on Unutilized Land Using Gray-level Co-occurrence Matrices

To strengthen the supervision of unutilized land in large areas, this paper identified the unutilized land of potential pollution through remote sensing technology. Taking the northern part of Gansu Province as the research area, a remote sensing interpretation of the land use and the cover type based on Landsat satellite data was first conducted to determine the scope of the unused land in the area. Second, the principal component analysis (PCA) of the image was carried out, and the first principal component was used as the data source of the gray level co-occurrence matrix. The energy, entropy, moment of inertia and correlation were selected as the feature quantities. The absolute value of the gray value change of the corresponding image was combined to extract the greatly-changed area. Finally, by comparing the feature quantity changes of Landsat remote sensing images in 2010 and 2015, areas with obvious texture or grayscale changes were extracted. Combining with the high-resolution images of Google Earth and point of interest (POI) data containing industrial and mining enterprise location information, it was concluded that there were 40 suspected soil pollution sites in this area from 2010 to 2015, with a total area of about 10 square kilometers. Twenty-one of the results were conducted field-test, and 19 suspected contamination sites were confirmed, with an accuracy of approximately 90%. Compared with the traditional manual interpretation method, the method based on the gray level co-occurrence matrix method to identify the suspected pollution in unused land is more effective. It can save manpower and material resources significantly, improve the monitoring efficiency and have better precision.