| Abstrakt: |
This study was carried out to apply a cost-effective remote sensing-based methodology to predict water quality parameters of Manchar Lake in Pakistan. Water quality models were developed using level 2 Landsat 8 OLI satellite data for the lake, where regular water quality monitoring is limited in time and space. This study focused only on three water quality parameters (WQPs) with optical properties that can be directly captured using remote sensing tools including total dissolved solids (TDS), total suspended solids (TSS), and turbidity. Twenty-one water samples were collected at various times for pre-monsoon (April 2019), monsoon (August 2019), and post-monsoon (November 2019) seasons on or within 2–3 days the satellite overpass dates. Among them, six samples were randomly selected for models' validation. Regression analysis was performed on the remaining samples to obtain WQPs' empirical relationships with bands' surface reflectance employing single and different band combinations. The linear, logarithmic, and first- and second-degree polynomial regression models were developed and based on the highest R2 values, the "best" models were selected. TDS models performed comparatively well with R2 = 0.9731, 0.7359, 0.7969 in pre-monsoon, monsoon, and post-monsoon periods, respectively. The R2 for the "best" TSS models were 0.7721 (pre-monsoon), 0.8561 (monsoon), and 0.5868 (post-monsoon), whereas their values for turbidity were 0.4807 (pre-monsoon), 0.5212 (monsoon), and 0.6404 (post-monsoon) suggesting relatively weak model performances. Models' validations produced maximum/minimum root mean square errors of 515/246.9 mg/L, 9.53/8.78 mg/L, and 27.99/13.86 NTU, respectively, for TDS, TSS, and turbidity. Statistically significant values of R2 indicated the suitability of models to predict WQPs of Manchar Lake. However, the models' performance was reduced in the post-monsoon season, but R2 values were still greater than 0.5. From these results, it can be expected that GIS and remote sensing-based water quality modeling will eventually provide convenient solutions for Manchar Lake's management and long-term planning. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink