Integrating logistic regression and cellular automata-Markov models with the experts' perceptions for detecting and simulating land use changes and their driving forces.

Modeling spatial-temporal dynamic of land use change is of great necessity for understanding the status of the past, causes of the change, and prediction of the future. This study aims to objectify three topics which include identifying the past land use changes, modeling the future changes, and subsequently considering their driving forces. The change detection analysis has shown that about 12,081.8 ha of the study area has changed since 1984 to 2014. Moreover, the models of cellular automata (CA) and Markov chain were applied in order to predict the land use changes of 2024 and 2034. The simulated transition matrix showed that about 6780 ha and 10,835 ha would change during the periods of 2014-2024 and 2014-2034, respectively. Furthermore, the results of the logistic regression model showed that the human driving forces of distance to roads, distance to wells, distance to streams, and distance to residential areas have had a negative effect on the process land use changes. Additionally, a questionnaire was used to obtain information considering the management factors of preventing land use changes, the perception of the natural resources' experts and in turn finding some socioeconomic and policy forces on land use changes. The Friedman's test analysis indicates that the factors of the official rules of government, economy, weakness of regulatory systems, and development activities, e.g., infrastructure and industrial projects, were identified as the leading causes of converting natural ecosystems to other land uses, particularly to cropland. Therefore, the decision-makers and managers should be assigned comprehensive planning for the protection, restoration, and development of natural resources, especially in this region.