Machine learning-based inversion of water quality parameters in typical reach of the urban river by UAV multispectral data
Urban rivers play an essential role in the human environment and urban development; because of their narrow and long characteristics, challenging for general remote sensing data sources to meet the monitoring requirements. In order to solve the problem of insufficient application of remote sensing w...
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Elsevier
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/51aee175aaa149f6a2b007657c26c64b |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Urban rivers play an essential role in the human environment and urban development; because of their narrow and long characteristics, challenging for general remote sensing data sources to meet the monitoring requirements. In order to solve the problem of insufficient application of remote sensing water quality monitoring in urban rivers. In this paper, based on unmanned aerial vehicles (UAV) images and measured water quality data, the genetic algorithm_extreme gradient boosting (GA_XGBoost) algorithm is used to model water quality parameters in the study area, combined with its characteristics of supporting urban river polymorphism learning and semantic feature analysis. The results show that the coefficient of determination (R2) of GA_XGBoost algorithm for chlorophyll a (Chla), total phosphorous (TP), total nitrogen (TN), ammonia–nitrogen (NH3-N) and turbidity (TUB) is 0.855, 0.699, 0.787, 0.694, and 0.597, respectively, indicating a high precision and the predicted results are consistent with the measured data. Meanwhile, this paper compares the GA_XGBoost model with other algorithms: Deep Neural Network (DNN), Random Forest, genetic algorithm_RandomForest (GA_RandomForest), adaptive boosting (AdaBoost) and genetic algorithm_adaptive boosting (GA_AdaBoost), and the performance of the GA_XGBoost model is better. At the same time, data from different periods have been added to verify the model’s applicability. Moreover, based on the inversion results, analyze from the point of view of point source pollution, non-point source pollution, etc., to further investigate the influencing factors that cause urban river pollution. The current method has important practical significance for promoting the intelligent and automatic level of water environment monitoring technology in ecological environmental protection and urban water resources protection. |
|---|