| Summary: | Identifying the spatial range of mining disturbance on vegetation is of significant importance for the plan of environmental rehabilitation in mining areas. This paper proposes a method to identify the spatial range of mining disturbance (SRMD). First, a non-linear and quantitative relationship between driving factors and fractional vegetation cover (FVC) was constructed by geographically weighted artificial neural network (GWANN). The driving factors include precipitation, temperature, topography, urban activities, and mining activities. Second, the contribution of mining activities (Wmine) to FVC was quantified using the differential method. Third, the virtual contribution of mining activities (V-Wmine) to FVC during the period without mining activity was calculated, which was taken as the noise in the contribution of mining activities. Finally, the SRMD in 2020 was identified by the significance test based on the Wmine and noise. The results show that: (1) the mean RMSE and MRE for the 11 years of the GWANN in the whole study area are 0.0526 and 0.1029, which illustrates the successful construction of the relationship between driving factors and FVC; (2) the noise in the contribution of mining activities obeys normal distribution, and the critical value is 0.085 for the significance test; (3) most of the SRMD are inside the 3 km buffer with an average disturbance distance of 2.25 km for the whole SRMD, and significant directional heterogeneity is possessed by the SRMD. In conclusion, the usability of the proposed method for identifying SRMD has been demonstrated, with the advantages of elimination of coupling impact, spatial continuity, and threshold stability. This study can serve as an early environmental warning by identifying SRMD and also provide scientific data for developing plans of environmental rehabilitation in mining areas. © 2022 by the authors. Licensee MDPI, Basel, Switzerland.
|
|---|
