Spatial Analysis of Aquatic Ecological Health under Future Climate Change Using Extreme Gradient Boosting Tree (XGBoost) and SWAT

• Published in: Water (Basel) 2024-08, Vol.16 (15), p.2085
• Main Authors: Woo, Soyoung, Kim, Wonjin, Jung, Chunggil, Lee, Jiwan, Kim, Yongwon, Kim, Seongjoon
• Format: Article
• Language: English
• Subjects: Accuracy; Algorithms; Aquatic ecosystems; Climate change; Climatic changes; Data collection; Drought; Environmental impact; Environmental monitoring; Fish; Hydrology; Machine learning; Plankton; Precipitation; River ecology; Stream flow; Water quality; Water resources; Watersheds
• ISSN: 2073-4441; 2073-4441
• DOI: 10.3390/w16152085

Climate change not only affects the water resource system but also has a great impact on the aquatic ecosystem, which is complexly linked to various organic and inorganic matter. It is difficult to simulate the current aquatic ecosystem and predict the future system due to the immensity and complexity of aquatic ecosystems; however, a spatial analysis of future aquatic ecological health is necessary if we are to adapt and take action against future climate change. In this study, we evaluated the aquatic ecological health of the Han River basin under the future climate change RCP4.5 and RCP8.5 scenarios using three indices: fish assessment index (FAI), trophic diatom index (TDI), and benthic macroinvertebrate index (BMI). For this, we developed the SWAT-XGBoost linkage algorithm, and the algorithm accuracy for the FAI, TDI, and BMI was 89.3~95.2%. In the case of the FAI and BMI assessment of aquatic ecological health, the upstream Han River was classified as a hot spot. In the case of the TDI, the downstream area of the Han River was classified as a cold spot. However, as the current TDI downstream was classified as grades D and E, continuous management is needed.