Using neural network modeling to improve the detection accuracy of land subsidence due to groundwater withdrawal  

作　　者：Ali M.RAJABI Ali EDALAT Yasaman ABOLGHASEMI Mahdi KHODAPARAST 

机构地区：[1]Department of Engineering Geology,University of Tehran,Tehran 1417614411,Iran [2]Department of Civil Engineering,University of Qom,Qom 3716146611,Iran

出　　处：《Journal of Mountain Science》2024年第7期2320-2333,共14页山地科学学报（英文）

摘　　要：Despite the high efficiency of remote sensing methods for rapid and large-scale detection of subsidence phenomena,this technique has limitations such as atmospheric impact and temporal and spatial decorrelation that affect the accuracy of the results.This paper proposes a method based on an artificial neural network to improve the results of monitoring land subsidence due to groundwater overexploitation by radar interferometry in the Aliabad plain(Central Iran).In this regard,vertical ground deformations were monitored over 18 months using the Sentinel-1A SAR images.To model the land subsidence by a multilayer perceptron(MLP)artificial neural network,four parameters,including groundwater level,alluvial thickness,elastic modulus,and transmissivity have been applied.The model's generalizability was assessed using data derived for 144 days.According to the results,the neural network estimates the land subsidence at each ground point with an accuracy of 6.8 mm.A comparison between the predicted and actual values indicated a significant agreement.The MLP model can be used to improve the results of subsidence detection in the study area or other areas with similar characteristics.

关 键 词：DINSAR Land subsidence Groundwater withdrawal Aliabad plain Artificial neural network 

分 类 号：P641.8[天文地球—地质矿产勘探] P642.26[天文地球—地质学]