<![CDATA[The decline in water transparency in the Saguling Reservoir indicates deteriorating water quality due to increasing pollutant loads from domestic, industrial, agricultural, and aquaculture activities. This study aims to predict water transparency using an ANN model, integrating remote sensing data from Landsat 8 (2013-2024) and in-situ measurements from 12 sampling points. The analysis began with correlation tests, which revealed weak to moderate relationships between transparency and other variables. Subsequent simple and multiple linear and logistic regression analyses produced weak correlations, with the highest R² of 0.1853 observed in multiple logistic regression. The Random Forest algorithm was applied to identify the most influential variables. The selected predictors included Bands 3, 4, 5, and 7, as well as temperature, EC, and TSS. These variables were used as inputs for the ANN model, which demonstrated high performance with an R² of 0.8514, explaining 85.14% of the variability in water transparency. The prediction results were visualized in a distribution map, indicating a predominance of transparency class IV (0-2.5 m) across the reservoir. This suggests limited light penetration due to high pollutant loads. The study shows that integrating remote sensing and ML enables effective large-scale water quality monitoring and supports sustainable water resource management.]]>
Copyrights © 2025
