A Machine Learning Approach for Estimating the Trophic State of Urban Waters Based on Remote Sensing and Environmental Factors

To improve the accuracy of remotely sensed estimates trophic state index (TSI) inland urban water bodies, key environmental factors (water temperature and wind field) were considered during modelling process. Such can be easily measured display a strong correlation with TSI. Then, backpropagation neural network (BP-NN) was applied to develop TSI estimation model using remote sensing factors. The trained validated quantified by five indicators obtained for period between 2018 2019, then we selected most appropriate combination input variables according performance BP-NN. Our results demonstrate that optimal combining single-band reflection values Sentinel-2 satellite imagery as (R2 = 0.922, RMSE 3.256, MAPE 2.494%, classification rate 86.364%). Finally, spatial temporal distribution aquatic over four months different levels mapped in Gongqingcheng City model. In general, predictive maps based on our proposed show significant seasonal changes characteristics state, indicating possibility performing cost-effective, RS-based studies complex bodies elsewhere.