Optimization of the ALPRIFT method using a support vector machine (SVM) to assess the subsidence Vulnerability of the southwestern plain of Tehran

Considering the increase in population and the development of agricultural activities in the south-western plain of Tehran, which has led to an increase in underground water drainage, the assessment of the vulnerability of subsidence areas is of great importance. In this research, the vulnerability of south-western plain of Tehran to subsidence has been investigated using ALPRIFT method in ArcGIS and optimization of the ALPRIFT method has been performed using SVM. To implement the ALPRIFT method, effective parameters have been used to assess the subsidence susceptibility including aquifer media, land use, groundwater pumping, recharge, aquifer thickness, faults distance and groundwater decline, which are ranked in seven separate layers and Weighted the ALPRIFT index from the combination of these seven layers, which was estimated to be 173-77. In order to optimize the ALPRIFT method, the SVM model was used. For this purpose, the input data (ALPRIFT parameters) and output (vulnerability index) and the subsidence amount were related to the two groups of training and testing. After training the model, Using the subsidence amount, the model results were evaluated at the experimental stage. The results showed that the SVM model was able to improve the results of the original ALPRIFT method. In order to verify the results, the InSAR radar map and its correlation coefficient (R^2) with the vulnerability index and correlation index (CI) of the piezometers in the plain were used.