The Consequences of Following Two Different Land Use Change Scenarios in 2040 in the Tajan Watershed

The purpose of this research was to model and predict land use/cover changes using the LCM model based on two management scenarios, namely the sustainability scenario and the agricultural increase scenario in the Tajan Watershed, the Province of Mazandaran. Landsat satellite images and the data collected by the TM and OLI sensors were analyzed for 1991, 2010 and 2019. Images from the three-time periods were categorized into five categories: residential, farmland agricultural, forest, orchard, and rangeland. Land use forecasting for 2019 was performed using the 1991 and 2010 land use maps in the LCM model. The LCM model uses the perceptron multilayer neural network to construct transmission sub-models, map the probability of transmission, and ultimately predict the land use changes. Spatial variables such as distance from a road, distance from a river, slope, altitude, distance from a village, and distance from farmland were used as the influencing factors on the neural network changes. The land use map of 2019 was used to validate the model, which resulted in the Kappa and ROC coefficients of 0.84 and 0.89, respectively. The modelling results based on the sustainability scenario show a 5.74 percent decrease in the rangeland area and a 9.17 percent growth in the orchard area over the next 20 years, while the second scenario showed an increase of 8.61% in farmlands and a decrease of 5.59% in rangeland areas. Therefore, both rangeland and forest areas are declining and farmland and residential areas are increasing. A better land use management will require conservation scenarios in different parts of the watershed.