Abstract: Based on the seven phases multi-source remote sensing images of the 1987, 1994, 2000, 2006, 2009, 2013 and 2017 of the Nihe watershed in Huainan Coal Mine, the texture and the spectral features have been used to classify the remote sensing images by support vector machine (SVM). Based on this, the influence of coal mining on land kuse and landscape pattern change in Nihe watershed in recent 30 years is analyzed. The results show that the mining of coal mine has changed the land use pattern and type of the watershed, the most significant change is the mutual transformation of cultivated land and construction land and the transformation of cultivated land to the subsidence water. Over the past 30 years, the area of subsided water increases by 2 281.05 hm², the area of construction land increases by 14 741.73 hm², and the area of cultivated land decreases by 15 044.67 hm². Driven by coal mining, the overall landscape pattern of the Nihe watershed is turned into a turning point in 2006. It is characterized by fragmentation, irregularity, heterogeneity and low connectivity in the early stage, and then it is continuous, regularized, balanced and highly connected changing. The construction land, cultivated land and subsidence water have undergone the process of continuous fragmentation to the gradual improvement of the landscape, the plaque shape of cultivated land, construction land and natural water body is complex. The shape of forest land, garden land and subsidence water is relatively regular, cultivated land has always been in a dominant position, but its superiority is weakened. The maximum plaque index is reduced from 79.706 0 to 38.745 5, and the advantages of subsidence water and construction land are enhanced.

Key words: remote sensing, support vector machine, Nihe watershed, land use, landscape pattern, coal mining