Earth is a limited and non-renewable natural resource that is directly affected by the population growth pressures. In order to make the optimal use of land, it is necessary to be aware of bad land use/land cover (LULC) changes and the ways in which human beings make use of the land, which is possible by detecting LULC change. In this study, remote sensing (RS)/geographic information systems (GIS) were used, and Landsat 5 and Landsat 8 images from the years 1986, 2000, and 2016 were analyzed for changes. The aim was using multilayer perceptron (MLP) neural network and systematic points statistical analysis (SPSA) for predicting the trend of LULC changes in RS/GIS. The satellite images of three different years were classified into five classes. Variables such as proximity to the road network were considered as effective parameters in growth and development. The SPSA with scattering point trends and points kernel shape also showed the effect of changes on each factor and urban zone. According to the results, during the 30 years, 10.6% of agricultural lands were destroyed and urban areas increased by 23.4%. Agricultural lands and open lands have changed more than other LULCs and have become urban areas with the highest rates of change in the southern parts of the river on the southern and northern margin of the city. These results were shown some layers had more effective on changes, and some region according to desirable for urban developments had more changes that should be considered in urban planning.
Kalkhajeh, Reza Ghorbani,Jamali, Ali Akbar. Analysis and Predicting the Trend of Land Use/Cover Changes Using Neural Network and Systematic Points Statistical Analysis (SPSA)[J]. JOURNAL OF THE INDIAN SOCIETY OF REMOTE SENSING,2019-01-01,47(9):1471-1485