Urban environmental and land cover change analysis using the scatter plot, kernel, and neural network methods

Abstract

This study simulates and predicts the urban environment growth in the Tehran, capital of Iran, using the remote sensing data, multi-layer perceptron neural network, zonal, trend, and profile modeling. A spatial-temporal modeling was used for the analysis and prediction of the urban environment development. After building the probability map of the land changes, random points scatter and kernel analysis (RPSKA) was used. The pixel values of all the maps was extracted to the random points for the scatter plot and kernel analysis. The results obtained by developing the change transition model using multi-layer perceptron neural network showed high accuracy in most of the sub-models. The area of the open lands and green spaces was reduced, and urban areas, agricultural lands, and clay plains were increased. Most of the land use and land cover (LULC) changes during the period 1990–2000 were observed in the north, while the most land use and land cover changes during the period 2000–2016 were observed in the west. The results of RPSKA were shown the direct and inverse relationship between the probability of land changes and the other factor maps. Sever changes have occurred from the open lands to the urban areas. The slope and the population density had more effect on the changes. Modeling of future LULC change showed that the urban areas would be increased, while open lands and green spaces would be decreased. These land changes have taken place in the north and west of the city that these regions were most popular and had suitable infrastructures for developments.