Integration of Cellular Automata -Markov (CA-Markov) Model and Logistic Regression to Land-Use Change Prediction: (A Case Study of Gamasiab Basin)

Land use modeling is very vital for decision makers and plays an important role in environmental planning and management. The CA-Markov model have strong ability to project the spatial pattern and to evaluate land use and land cover changes. In This Study, a series of satellite images of Landsat TM, ETM+ and OLI data of 1987, 2002, 2016 were used to produce classified land use maps. Land use maps of Gamasiab basin were prepared using maximum likelihood classification. Area change and spatial distribution of land use were calculated using GIS technology. The transition area among different land use types were analyzed to obtain the transformation matrices. The transition probability matrix shows from 1987 to 2002, barren land, grassland and urban expansion are the most stable classes. In other hand, the most dynamic classes are water and cultivated land. Based on the success of the models for 2016 using 1987 and 2002 maps, simulated future land use map for 2030. Suitability image collection were prepared by using logistic regression and then its results were used in CA-Markov model. The high agreement between predicted and the real map demonstrated that the suitability image collection derived from logistic regressions which have high precision‬. It also proved that the selected factors could represent the influencing processes of land use changes adequately. The coefficients of the distance to current land use classes had the highest values among all impact factors. The distance from the road and river shows a high impact on urban development.The results show that the greatest change in grassland has turned them into farmland and then barren land. The barren land has also turned into farmland.