Land allocation based on spatial analysis using artificial neural networks and GIS in Ramsar, Iran

Abstract

The purpose of the current study is to compare two kinds of allocation maps. In this investigation, the first map is taken from a supervised classification of the advanced spaceborne thermal emission and reflection radiometer imagery, and the other map is adopted from topo-climatic data assessment; the self-organizing map (SOM) and multi-layer perception (MLP). Topo-climatic data were analysed through artificial neural network (ANN) techniques as it has allowed not only to detect to distinct between low, moderate and high allocation zones. A new computational framework was developed in this research to compare results between two different methods including SOM and MLP. In this field, geographic information system (GIS) is applied due to the ability of GIS databases to integrate and work with information from heterogeneous and uncertain data into a geospatial context. The results show that the MLP was significantly close to current cultivation. Yet, it has provided better insights compared to the SOM in safe regions with regard to citrus allocation maps (CAMs). An accuracy assessment of 99.8% demonstrated the allocation of the proposed approach. Consequently, the comparison and differences of SOM and MLP algorithm of ANN method along with the topo-climatic factors could help policymakers and planners to improve the accuracy of CAMs.