Implementation of soil fertility detection and crop recommendation system based on Geographic Information System (GIS) for precision agriculture

Accurate soil fertility detection and intelligent crop recommendation systems are fundamental to the advancement of precision agriculture, enabling data-driven decision-making for sustainable food production. However, many smallholder farmers still rely on intuition or traditional practices for crop selection, which often leads to mismatches between soil characteristics and crop requirements, resulting in inefficient input use and reduced productivity. To address this issue, this study implements an integrated Soil Fertility Detection and Crop Recommendation System based on Geographic Information System (GIS) technology. The system integrates real-time soil sensors (pH, temperature, and moisture) with a Fuzzy Mamdani inference model to classify fertility levels and generate crop recommendations visualized through a web-based GIS interface. Field testing was conducted in Kamal Village, Bangkalan, Indonesia, using multiple soil sampling points. Experimental validation showed high computational precision, with an average fuzzy system deviation of 0.00397% compared to MATLAB simulations. Sensor calibration yielded MAPE values of 0.0209% (temperature), 0.0481% (pH), and 0.0929% (moisture), while GPS testing using the BN220 module produced an average positional error of 4.1 cm. The recommendation subsystem accurately classified field conditions, identifying peanut for Field 1 and maize for Fields 2 and 3. GIS visualization effectively mapped spatial variations of soil parameters and crop suitability, confirming the system’s reliability and applicability for data-driven precision agriculture.