Abstract: In view of the problem of reduced yield and ecological damage caused by coal mining in the high phreatic mining area in the east of China，takes the maize crop of coal mining subsidence in Dongtan Coal Mine as an example，based on the multi-spectral image of the UAV （unmanned aerial vehicle） and the field measured data，and combing with the empirical model method，the inversion model of maize leaf area index of coal mining subsidence farmland is established and the quality evaluation of farmland carried out.The results show that the power function model established by the RRENDVI （red-red edge normalized difference vegetation index） is the optimal model with R2=0.756 and RMSE=1.125，and this model has features of high precision and reliability.Based on the model，LAI of maize from the study area was inverted to obtain the distribution map of maize LAI .According to the spatial distribution of maize LAI in the study area and the average value of local normal growing maize LAI，combing with the actual subsidence of water accumulation，the quality evaluation rules for subsided farmland are constructed，and the quality of coal mining subsidence is classified into five grades.According to the degree of damage of various grades of farmland，suggestions for reclamation are proposed，such as flattening，digging，deepening and padding.The above discussion results has important guiding significance for monitoring the damage to farmland in small subsidence area，evaluation the farmland quality and land reclamation.

Key words: UAV remote sensing, Mining subsidence farmland, Empirical model method, Red edge NDVI, LAI, Quality evaluation of framland, Land reclamation.