Determination of temporal change using surface temperature and climatic data (Kabul city example)

As the world's population grows day by day, the needs of people are also naturally increasing. Nonetheless, due to different factors the migration rate to urban centers also increases. These are the most important reasons for land use land cover change (LULC) and urban areas expansion. Kabul City, the capital of Afghanistan with approximately more than 5 million population, is also not an exception in this case and the rate of urban growth and spatial changes is increasing day by day. Therefore, the aim of the current study is to analyze LULC spatial changes of Kabul city for thirty years, to investigate climatic elements and their relationship with surface temperature, and to simulate urban areas for the future. For the purpose of generating LULC map, Landsat satellite images were classified using three different image classification methods after pre-processing operation. LULC map was obtained using decision tree method by combining the results of Support Vector Machine (SVM), Neural Network Classification (ANN) and Mahalanobis Distance (MD) satellite image classification methods. After the accuracy analysis of the map was done, a 30-year LCLU changes analysis was performed using the Land Change Modeler (LCM) model. The results show that the most affected classes by the thirty-year changes are urban areas and barren lands. For the purpose of studying the relationship of LULC change with climatic and surface temperature, surface temperature maps were generated by the Planck function method using NDVI maps derived from Landsat satellite images. The results of comparing surface temperature and LCLU maps suggests that changes in LCLU, especially the growth of urban areas and the increase and decrease of vegetation, influences and a relationship with surface temperature. In addition, temperature, precipitation, humidity, drought, and evapotranspiration temporal change graphs were formed using climatic data from the meteorological department and by using Google earth engine online platform considering MODIS, IDAHO and TRMM raster data which contains climatic values; and subsequently their relationship with LCLU changes was investigated. In the last part of the thesis, for years 1989, 2004 and 2019, new maps with two classes of urban and other areas, were produced from LCLU maps. Maps with future change probability values are generated from Vector, DEM and LULC data. The generated maps are combined according to the general suitability and probability of changes resulting from AHP analysis, by using the weighted linear combination (WLC) method in the multi-criterion evaluation (MCE) module. The urban areas map for the year 2019 was then modeled using Markov chain and Cellular Automata-Markov models. Since the Kappa accuracy (0.7034) of the map modeled according to the 2019 urban areas map is acceptable, urban areas were simulated by the same method for the future 2034 and 2050 years.