Quantifying turbidity dynamics in lake water using OLS regression: A landsat 8 OLI-based remote sensing approach

Abstract

Lake Tana, Ethiopia’s largest freshwater lake, has experienced a notable increase in water turbidity. This issue highlights the need for an in-depth understanding of how human activities and environmental changes are impacting its ecological balance. Addressing these turbidity challenges is crucial for safeguarding the sustainability of this vital resource. This research utilized Landsat 8 satellite imagery to examine turbidity levels in Lake Tana. Six bands from Landsat OLI—band 2, band 3, band 4, band 5, band 6, and band 7—were analyzed both individually and in combination. Ordinary least squares (OLS) regression modeling was applied to investigate the relationships between these bands and in-situ turbidity data. Our findings reveal that the combined use of specific bands—particularly band 2 + band 5 - band 6—accounted for 87 % of the variance in turbidity as explained by the OLS regression model. Additionally, the Koenker- (Breusch-Pagan) statistic indicated no conflicting relationships (p > 0.005) within the model, affirming its reliability. To further validate the model’s impartiality, the Jarque-Bera test was performed. Polynomial and exponential regression analyses were also conducted, leading to the identification of an optimal regression equation for predicting the spatial distribution of turbidity in Lake Tana. • Advanced remote sensing assessed turbidity with Landsat 8's six spectral bands. • OLS regression showed (band 2 + band 5 - band 6) explained 87 % turbidity variance. • Koenker (BP) and Jarque-Bera tests validated the model’s robustness and neutrality. • Combining spectral bands and regression advanced large-scale turbidity assessment.