Modelling land use/cover change in Seyhan basin, Turkey using remote sensing and geographic information system

Land use and land cover (LULC) changes affect many environmental factors such as soil, water, biodiversity, climate, which all of them usually affect human lives. Therefore, LULC change has become an important topic to manage land use and environment. One of the effective way to analyses LULC change is using of modeling approaches. In this study, CA-Markov model that combines the Cellular Automata and Markov chain models, was used to predict LULC changes in the Seyhan Basin for the 20-year horizon, i.e., to 2036. Remote sensing multispectral imagery acquired in years 1995, 2006 and 2016 were classified using the object-oriented classification approach and fed into the CA-Markov model. Subsequently, post-classification comparison technique was used to detect historic change. Markov Chain analyses and Multi-Criteria Evaluation (MCE) were used to produce transition probability matrixes and suitability future maps, respectively. In validation process, the model was validated using Kappa index of agreement which resulted in overall accuracy 77 %. Finally, the future land use/cover changes of 2036 based on transition rules and transition area matrix was predicted and mapped. LULC prediction for year 2036 showed 50% increase of built-up area, 7% decrease of open spaces with little or no vegetation in comparing with year 2016. About 8% increase of agricultural land was found in 2036. No significant changes were found for wetland and water body category. About 4% increase of shrub land and 5% decrease of forest were found when compared with 2016 data. Keywords: Land use/cover changes, CA-Markov model, Seyhan Basin