A complementary evaluation of spectral indices and an ICAMPFF–GA composite model for smart satellite-based water quality monitoring

Smart water quality monitoring using remote sensing is typically performed using either empirical spectral indices or the newly developed composite algorithms. This study conducts a complementary algorithmic evaluation of three classical empirical spectral indices: (i) NDWI - Normalized Difference Water Index, (ii) NDTI - Normalized Difference Turbidity Index, and (iii) NDCI - Normalized Difference Chlorophyll Index, and a genetic optimized composite algorithm (ICAMPFF–GA). We utilized Sentinel-2 imagery over two distinct regions (Souss-Massa, Morocco; Cartagena Bay, Colombia) for model validation against in-situ data using descriptive-statistical measures and quantitative error metrics (MAE, RMSE, MAPE). The findings indicate that NDWI and NDTI are effective at mapping water bodies and turbidity, while the NDCI supports estimation of chlorophyll-a concentrations. ICAMPFF–GA, as a composite model, has produced the best overall accuracy of MAE $$\approx$$ 1.84; RMSE $$\approx$$ 2.38; and MAPE $$\approx$$ 2.70 %. Therefore, the composite algorithm is the most robust and adaptable method for performing multi-parameter water quality assessments using remotely sensed data. Moreover, this comparative assessment has shown the potential for collaboration between empirical spectral indices and composite algorithms to create robust, satellite-based operational monitoring systems for aquatic resources.