Evaluation of the soil fertility for corn production (Zea Mays) using the multiple-criteria decision analysis (MCDA)

Abstract

The ongoing water deficiencies in arid and semi-arid regions in conjunction with certain nutritional requirements highlight the importance of pinpointing the most optimal locations for cultivating agricultural products with the highest yield. Given the importance of this issue, this study proceeds to prepare soil fertility maps for corn production (Zea Mays L.) in Fars province, Iran. Initially, fuzzy membership functions (FMs) are employed to prepare fuzzy maps for each layer in the geographic information system (GIS), after which feature selection algorithms are deployed to designate the most relevant layers. The layers are then assigned specific weights obtained using a combination of analytic network process (ANP) and analytic hierarchy process (AHP) methods to prepare soil fertility maps. The input data consist of organic content (OC), phosphorus (P), potassium (K), iron (Fe), zinc (Zn), manganese (Mn), and copper (Cu). Inverse distance weighting (IDW) is utilized to procure interpolation maps for each layer. Thereafter, zoning maps are obtained using FMs. Ultimately, ANP and AHP models are once again deployed to generate the final overlayered fertility maps for corn production. The results show that combining the ANP method with feature selection (OC, K, Fe, and P) results in higher accuracy than solely applying the AHP method. Thus, incorporating feature selection and ANP methods with both inter and intra-group pair-wise comparison would result in more accurate fertility maps for corn production with lower costs and time complexity.