Multi Logistic Regression Model for Simulating Multiple Land Use Changes; a Case Study: Tehran, Iran

In recent years due to increasing the population and migration to cities, land use changes have become the most important issues. For a better planning for the future, a vision of the future would beneeded. Because of this reason, we have to simulate the land use changes for the future years. Themain objective of this paper is simulating multi land use change for Tehran Metropolis. To access the trend of changes, TM and ETM+ images for the years 2000 and 2010 were used to produce the land use maps. In this regard, the Maximum Likelihood (ML) classification method was used to producethe land use maps. Four types of land uses have been found in study area, such as open lands, buildings, parks, and agricultural lands. Multi Logistic Regression (MLR) method was used tosimulate land use changes in Tehran. Accordingly, Logistic Regression (LR) was applied for each landuse separately and four predicted map integrated to each other to produce final land use map. Different variable such as distance to buildings, distance to parks, distance to open lands, distance to agriculture lands, distance to roads, distance to city exit, elevation, slope, and population density were used tocreate suitability maps. These factors were extracted from various existing maps and remotely senseddata using the ArcGIS and ENVI softwares. Relative Operating Characteristic (ROC) was used to validate the method. Finally, the approach was applied to produce land use maps for the year 2020.

Keywords:

Multi Logistic Regression , Land use changes , ROC , Remote sensing