Evaluation of Land Use Suitability for Wheat Cultivation Considering Geo-Environmental Factors by Data Dependent Approaches

Abstract

Two techniques were investigated to be standard deviation based weighting Multi Criteria Assessment (MCA), and Artificial Neural Network (ANN) considering base environmental factors to define wheat cultivation suitability in Van region. Climate data (long term annual, maximum and minimum temperature, total mean precipitation and solar radiation), physical factors such as elevation, hillshade, slope, soil depth, accessibility to the fields and land use cover were used to produce wheat suitability map. All inputs were weighted with reference to existing wheat areas. MCA and ANN approaches were applied using same dataset to compare the performance of the two techniques. In total, 228 wheat parcels were used as training (171 parcels) and testing (57 parcels) data. Relative Operational Characteristic (ROC) was applied for accuracy assessment. ROC values of the MCA technique which was depended on existing wheat lands, and ANN techniques were derived to be 0.875 and 0.71 respectively. Results showed that 15% of the research area was very suitable for wheat farm, and today, only 67% of very suitable areas were used to be agriculture. Other areas were currently used as grassland (28%), bare ground (4%), and other (1%).

Keywords

• agricultural land use suitability
• wheat
• multi-criteria assessment
• artificial neural network
• gis