In the heart of East Java, Indonesia, rice farmers are on the brink of a technological transformation that could redefine their approach to soil management. A recent study led by Novandi Rizky Prasetya from the Soil and Water Management Postgraduate Program at Brawijaya University sheds light on a cutting-edge method for predicting soil total nitrogen (STN) levels in rice fields using advanced Geo-AI techniques. This could be a game-changer for the agricultural sector, especially in regions where soil nutrient data is scarce.
The challenge of managing soil fertility is not a new one, but the absence of high-resolution STN data has long posed a significant hurdle for sustainable farming practices. As Prasetya explains, “Our research integrates remote sensing data with various environmental factors to create a more accurate picture of soil health.” By employing the random forest algorithm, a machine learning technique, the study harnesses a wealth of information—from topographic variables like elevation and slope to climate attributes such as temperature and precipitation.
The research utilized data collected from 318 sampling points across diverse landforms, including alluvial, karst, and volcanic regions within Malang Regency. This comprehensive dataset allowed the team to build a predictive model that not only demonstrates impressive accuracy—boasting an R² of 0.94 and a root mean square error (RMSE) of just 0.05—but also highlights the critical role remote sensing indices play in understanding soil dynamics.
Farmers stand to benefit immensely from these findings. With precise STN predictions, they can optimize their fertilization strategies, potentially increasing yields while minimizing waste and environmental impact. As Prasetya points out, “This approach offers a sophisticated solution for rice production, enabling farmers