Modeling and zoning of land subsidence in the southwest of Tehran using artificial neural networks

Abstract

The earth's surface, due to its natural conditions and its structure is always changing and reshaping. One of the created deformations is the land subsidence. This is the most dangerous events which can be seen in most urban areas especially in the agricultural plains today. This study aims at zoning land subsidence and recognition of geometrical factors in southwest of Tehran. To estimate and predict land subsidence, all the effective subsidence factors were identified. Among the factors, nine most important factors including, downfall of groundwater, thickness of clay, depth of groundwater, annual discharge of water from wells, the distance of well to each other, slop, elevation, land use and geology were evaluated. Ultimately, three variables were selected as the most important variables. For modeling and zoning these factors, artificial neural network using Matlab software and Arc-GIS software for creating primary layers were used. The results indicate that the main cause of subsidence is excessive removal of underground water resources. Since the use of water resources in agriculture is accounted for the highest percentage of consumption and also because a large part of the study area have an agriculture land use, therefore the underground water drop and agricultural land uses are the most susceptible areas of land subsidence occurrence.