Monitoring of Urban Black-Odor Water Using UAV Multispectral Data Based on Extreme Gradient Boosting

During accelerated urbanization, the lack of attention to environmental protection and governance led formation black-odor water. The existence urban water not only affects cityscape, but also threatens human health damages ecosystems. bodies are small hidden, so they require large-scale high-resolution monitoring which offers a temporal spatial variation quality frequently, unmanned aerial vehicle (UAV) with multispectral instrument is up task. In this paper, Nemerow comprehensive pollution index (NCPI) was introduced assess degree in order avoid inaccurate identification based on single parameter. Based UAV-borne data NCPI sampling points, regression models for inverting parameter indicative were established using three artificial intelligence algorithms, namely extreme gradient boosting (XGBoost), random forest (RF), support vector (SVR). result shows that qualified evaluate level XGBoost (XGBR) model has highest fitting accuracy training dataset (R2 = 0.99) test 0.94), it achieved best retrieval effect image inversion shortest time, made best-fit compared RF (RFR) SVR model. According results XGBR model, there size mild study area, showed achievement treatment Guangzhou. research provides theoretical framework technical feasibility application combination algorithms images field inversion.