A LINEAR REGRESSION MODEL FOR SOIL SALINITY PREDICTION IN THE GREAT HUNGARIAN PLAIN USING SENTINEL 2 DATA

Salts occur naturally within soil and water. When exceeding the thresholds, salinity becomes a severe threat, damaging agricultural productivity, water and soil quality, biodiversity, and infrastructures. Multispectral data retrieved from Sentinel-2 MSI sensor were used in this study to predict soil salinity in the Great Hungarian Plain. For this purpose, samples were collected from the upper layer of soil between mid-September and mid-October in the Hungarian Soil Monitoring System framework. The application of multiple linear regression analysis between salt content (g/kg) and remotely sensed data revealed a highly moderate correlation with a coefficient of determination R² equals 0.52, a p-value equals 0.001198, and an RMSE equals 0.194 g/kg. The model can be employed to highlight soil salinity levels in the study area and understand the efficiency of land management strategies, considering its moderate predictive power.