HYPERSPECTRAL DATA AND CLOUD COMPUTING SUPPORTING AMAZON’S SOIL HEALTH MONITORING

Deforestation and unsustainable land use practices increasingly compromise Amazonian soil health (SH). This study investigates the application of geotechnologies to enhance soil monitoring by integrating reflectance spectroscopy and cloud computing. We analyzed the physicochemical properties of 211 topsoil samples from diverse land-use areas, focusing on 15 key attributes commonly considered as soil quality indicators (SQIs) for SH assessment. Utilizing the Vis-NIR-SWIR Fieldspec 4 Pro hyperspectral sensor, we processed reflectance data through the Brazilian Soil Spectra Service (BraSpecS), which employs machine learning algorithms for real-time modeling. Validation results demonstrated strong correlations between observed and predicted data. Those R² above 0.70 were considered as SQI. For instance, soil organic carbon exhibited an error of 5.53 g kg-1 and an R² of 0.72 in the validation step. This methodology enables rapid, non-invasive SQI assessments and highlights the potential of hyperspectral reflectance and cloud computing to improve SH monitoring.