Zoning Suitable Land for the Cultivation of Rice, Wheat, and Barley by Integration of Artificial Intelligent Methods and Spatial Data

Abstract

Determination of suitable land for cultivating different crops is an essential action for increasing agricultural productivity. The present research is carried out to zone suitable lands for cultivating rice, wheat, and barley by integrating intelligent methods and spatial data in the Silakhor plane of Iran. Machine learning methods, including artificial neural network and random forest in integration with geographic information system data, are used for zoning. The nine agro ecological parameters were used as input layers. The suitability level as an output layer is classified into four classes: very suitable (S1), suitable (S2), relatively unsuitable (S3), and not suitable (N). The 720 samples were considered to evaluate the models. The 70% of sample data were used for training, 15% for testing, and 15% for validation. The coefficient of variation, root mean square error, and area under the ROC (Receiver Operating Characteristics) curve were obtained to evaluate the performance of the models. Based on the results, both methods have sufficient validity to assess the land suitability of rice, wheat, and barley in the studied area. By using the ANN model, 56.1%, 67.1%, and 80.7% of the studied areas were suitable for cultivating rice, wheat, and barley, respectively. By using the RF model, 58.6%, 58.3%, and 62.6% of the studied areas were suitable for cultivating rice, wheat, and barley, respectively.