Modeling water quality in the brazilian semiarid region using remote sensing: support for water management

Water management in semi-arid regions faces challenges due to water scarcity and the need for continuous quality monitoring. This study evaluates the use of remote sensing to analyze a reservoir's water quality status in Brazil's semi-arid region to support its management. Data from the Landsat-8 (Operational Land Imager) and Sentinel-2 (MultiSpectral Instrument) satellites were used to correlate spectral bands with water quality parameters such as Chlorophyll-a and Total Phosphorus. Using the Stepwise method, multiple regression models were developed to predict these parameters. Landsat-8 achieved determination coefficients (R²) of 0.81 for Chl-a and 0.72 for TP, outperforming Sentinel-2. Spectral analysis indicated that the higher signal-to-noise ratio of Landsat-8 in visible and near-infrared wavelengths contributed to the quality of the predictive models. Additionally, the assessment of land use along the reservoir margins revealed that the reduction of pasture areas is associated with the stability of TP levels. The trophic classification of the reservoir remained in an ultra-oligotrophic state during the analyzed period; however, seasonal episodes of TP increase exceeding established environmental limits were observed. These results highlight the need for continuous monitoring integrated with land use data. Expanding the collected database and adopting advanced methodologies, such as machine learning and hyperspectral remote sensing, is recommended to improve estimation accuracy. This study provides evidence supporting water management policies and environmental conservation in the Brazilian semi-arid region.