Simulation of land use changes using CLUE-s model (Case study: Mahdishahr Township, Semnan province)

Introduction
Land use is the most important factor affecting biodiversity on a global scale, the availability of water and climate condition. A large part of the land is used for agriculture, forestry, residential, and industrial areas, which has a great impact on the environment. Agricultural products, preparation of wood from forests, medicinal plants, animal products, air conditioning and purification are biological uses related to land use. Land use changes and conversion of natural resources to agricultural and residential areas have become a major problem for many countries, because it directly affects human life. One of the important reasons for land use changes is human interaction with nature and the use of natural resources to improve the quality of life. Land use changes affect hydrological processes such as infiltration, groundwater recharge, base flow and surface runoff. Unfortunately, the increase in population, the development of technology and the change in human lifestyle have had destructive effects on land use, especially in recent decades. Land use classification is often the first step in land use studies and thus forms the basis for many earth science studies. Up-to-date land use maps are very important to scientists, planners, natural resources managers and policy makers. Land use simulation is one of the most important tools to manage land use, so various methods were developed for land use change simulation. CLUE-s is one of this model that use to simulate land use change of Mahdishahr Township. Mehdishahr Township and especially Shahmirzad city has favorable weather conditions compared to Semnan and is always the focus of the people of Semnan and other cities of the province. This condition has caused an excessive increase in demand for residential areas and the destruction of natural resources, agricultural lands and their conversion into residential areas.
Methodology
The aim of this research was land use mapping using remote sensing technique, land use change detection, and finally land use change simulation for year 2040 in Mahdishar Township. At first, Landsat satellite images related to different seasons were prepared for 1992 and 2017. In the next step using field survey and also using Google Earth software, 128 training samples were prepared for supervised classification. The 85 training samples were used for classification and 43 samples were used to evaluate the classification method. There is multi-temporal agriculture in the case study, and there are spectral mixes between agriculture and other land use types. As a result, traditional methods did not have enough accuracy for land use classification in this region, and land use map of 1992 and 2017 was prepared using synthetic methods. In Synthetic methods, different types of land use are prepared using different methods. Synthetic methods use additional maps next to satellite images to separate land use types with the same reflectance. This integration of remotely sensed data with other data sources can result in higher classification accuracy. In the synthetic method, supervised and unsupervised methods and ancillary data were used simultaneously. Unsupervised classification method and false color combination were used to extract agricultural land use. Also, the slope map was used, in such a way that the agricultural use was compared with the slope map and the areas with a slope above 30% were removed from the map. Forest, residential and barren land use types were prepared using the supervised maximum likelihood method. By combining these land use types, the remaining areas were considered as rangeland. For 1992, the method was relatively similar, but training samples were taken from the images themselves and in specific areas of each land use types. For accuracy assessment, overall accuracy and kappa coefficients were calculated for the map created with the synthetic approaches. Training samples were entered into ENVI software and their compatibility with land use classes was checked. In the next step area of each land use types were calculated and land use change was identified in this period. Finally based on land use maps of years 1992 and 2017, land use change simulation of the Mahdishar Township was done using CLUE-s model. The simulation is such that susceptible areas for each type of land use are assigned to the respective land use based on the maximum probability. This is done based on the regression relationship for all land use types in all pixels. Accuracy assessment of regression method (a part of simulation process) was carried out using ROC curve for each land use types. To evaluate the accuracy of the land use map, the training samples that were not used in the classification are used. Accuracy assessment showed that overall accuracy and kappa coefficient of synthetic methods was 0.93 and 91.3% respectively. Overall accuracy of the synthetic approach (0.93) is over the 85 % level that is considered satisfactory for planning and management purposes. This shows that integration of remote sensing data, ancillary data and decision rules provides better classification accuracy than traditional methods. Results of land use change detection showed that the main land use change in Mahdishahr Township is degradation of natural resources areas and conversion to agriculture and residential land. The degradation rate of forest and range is 24 and 17.6 percentage respectively. The value of ROC method was achieved 0.91, 0.86, 0.92, 0.96 and 0.89 for agricultural area, bare land, forest, residential area and rangelands respectively.
Conclusion
In many studies land use map in topography mas are used that are no very accurate. Using these maps is not logical because of land use change in Iran. In this research land use map was created using synthetic method with acceptable accuracy. In general, Mahdishahr city has better climate conditions than its surroundings, so is very susceptible to land use changes. Due to the drought in recent years and the decrease in employment, migration from other provinces to Semnan province and consequently Mehdishahr city has increased, which has led to an increase in land use change in the region. Considering this condition, it seems very necessary to pay more attention to land use management and monitoring land use changes in this area. Simulated map of land use change using CLUE-s model can be a useful tool to better management of land use in the study area.