Prediksi Kandungan Nitrogen Tanah Menggunakan GNDVI Landsat 8 OLI, DEM, dan Lengas Tanah dengan Metode Geographically Weighted Regression

Abstract

Nitrogen is a macronutrient that plays an essential role in supporting plant growth. Soil nitrogen content is dynamic and varies due to the influence of various environmental factors. This study aims to analyze the spatial distribution of soil nitrogen in Sumbersari District and to examine the influence of GNDVI, Digital Elevation Model (DEM), and soil moisture variables using the Geographically Weighted Regression (GWR) method. The study was conducted using 33 soil sample points and remote sensing imagery data. The results show that the GWR model has not been able to fully explain the variation in soil nitrogen, with a coefficient of determination (R²) of 0.4826. This indicates that the model explains only about 48% of the variation in soil nitrogen content, while the remaining variation is influenced by other factors. In addition, a comparison with the Ordinary Least Squares (OLS) method shows that the OLS model is more stable in predicting soil nitrogen content. This condition is likely influenced by the limited number of samples, which reduces the optimal performance of the GWR model. Therefore, the GWR model still has limitations in modeling soil nitrogen under limited sample conditions.