Assessment of the Potential of PRISMA Hyperspectral Data to Estimate Soil Moisture

In this research the potential of the PRISMA hyperspectral sensor in comparison with multispectral data (Sentinel-2 MSI and Landsat 8 OLI) was assessed for predicting soil moisture. To this aim, PRISMA, Sentinel-2 and Landsat 8 spectra, resampled according to the spectral bands of each sensor, were simulated from a laboratory soil spectral library. The soil samples used to create the spectral library were collected from different agricultural areas in Central and Southern Italy. Partial Least Square Regression (PLSR), the Normalized Soil Moisture Index (NSMI) and the Soil Moisture Gaussian Model (SMGM) were employed to calibrate soil moisture (SM) estimation models from the resampled spectra. The prediction accuracy of SM estimation was assessed from statistical metrics. The best accuracies in retrieving SM were obtained by PLSR using data resampled at PRISMA spectral resolution. A preliminary test of the application of the calibrated models was carried out using real PRISMA and Sentinel-2 data.