Liu et al. (2025) Assess the impacts of climatic change and human activities on streamflow and floods by using a hybrid-physics-data (HPD) model: A case study in the Lancang-Mekong River Basin

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-09-10
• Authors: Binxiao Liu, Xiaobo Yun, Baoxiang Pan, Ximeng Xu, Paul P.J. Gaffney, Hui Lü, Lifeng Luo, Gaohu Sun, Qiuhong Tang
• DOI: 10.1016/j.ejrh.2025.102763

Research Groups

• Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
• University of Chinese Academy of Sciences, Beijing, China
• State Key Laboratory of Space Information System and Integrated Application, Beijing Institute of Satellite Information Engineering, Beijing, China
• National Meteorological Center, China Meteorological Administration, Beijing, China
• Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
• Department of Earth System Science, Institute for Global Change Studies, Tsinghua University, Beijing, China
• Department of Geography, Environment, and Spatial Sciences, Michigan State University, MI, USA
• International Economic & Technical Cooperation and Exchange Center, Ministry of Water Resources, Beijing, China

Short Summary

This study utilizes a hybrid-physics-data (HPD) model, combining VIC-CaMa-Flood with LSTM, to assess the relative contributions of climatic change and human activities to streamflow and floods in the Lancang-Mekong River Basin (LMRB) from 1966–2015. It reveals that climatic change primarily influenced streamflow and floods during 1993–2007, while human activities, mainly reservoir operations, became the dominant factor in the post-2008 period, significantly altering seasonal flow and mitigating flood magnitudes and frequencies.

Objective

• To assess the impacts of climatic change and human activities on streamflow and floods within the LMRB from 1966 to 2015 using a hybrid-physics-data (HPD) model.
• To utilize the HPD model to disentangle the separate roles of climatic and human-induced alterations to streamflow and floods at five stations along the LMRB.
• To quantify the influence of climatic and human factors on streamflow at both annual and seasonal scales.

Study Configuration

• Spatial Scale: Lancang-Mekong River Basin (LMRB), spanning from 9°60′ to 33°80′N and 93°50′ to 108°60′E, with a drainage area of 795,000 km². The study focused on five hydrological stations: Chiang Sean (CS), Luang Prabang (LP), Vientiane (VT), Mukdahan (MK), and Pakse (PK).
• Temporal Scale: 1966–2015 (50 years), divided into three periods: baseline (1966–1992), transition (1993–2007), and post-impact (2008–2015).

Methodology and Data

• Models used:
  • Hybrid-Physics-Data (HPD) model, which integrates:
    • Variable Infiltration Capacity (VIC) model (semi-distributed hydrological model)
    • Catchment-based Macro-scale Floodplain (CaMa-Flood) model (globally distributed river routing tool)
    • Long Short-Term Memory (LSTM) (recurrent neural network for post-processing and bias correction)
• Data sources:
  • Streamflow: Daily observations from five hydrological stations, provided by the Mekong River Commission.
  • Meteorological data: Daily maximum and minimum temperatures, and average wind speeds (1965–2015) from the Global Meteorological Forcing Dataset (0.25° spatial resolution).
  • Precipitation: Daily gridded dataset from the APHRODITE (Asian Precipitation-Highly Resolved Observational Data Integration Towards Evaluation of Water Resources) project (0.25° spatial resolution).
  • Land cover: University of Maryland’s global land cover dataset (1 km resolution).
  • Soil data: Harmonized World Soil Database.
  • Elevation data: Shuttle Radar Topography Mission (30 m resolution).

Main Results

• The HPD model demonstrated superior performance in simulating streamflow and flood magnitudes compared to standalone VIC-CaMa-Flood or LSTM models, with Nash–Sutcliffe efficiency (NSE) values between 0.90 and 0.94 and relative errors (RE) between -6.6% and 3.7% during the testing period (1980–1992).
• Transition Period (1993–2007): Streamflow and floods were primarily affected by climatic change. Climatic change contributed 42–89% to annual streamflow variations and 73–80% to wet season streamflow variations. Flood magnitudes (MAF, Q95, Q90) increased by 1.5–17.3%, and flood frequency increased by up to 1.67 events per year.
• Post-Impact Period (2008–2015): Human activities became the dominant factor.
  • Human activities led to a decrease in wet season streamflow by -6% to -24% (contributing 38–82% to changes) and an increase in dry season streamflow by 7–39% (contributing 42–91% to changes).
  • While climatic change alone would increase flood magnitude by up to 21% and flood frequency by up to 1.88 events per year, observed flood characteristics were significantly lower than HPD simulated natural changes, indicating substantial flood mitigation due to human activities (primarily reservoir construction and operation).
  • Upstream (e.g., CS station), flood reductions were attributed to a combination of climatic change and human activities.
• Streamflow predictability, as measured by the Colwell index, increased in the post-impact period, largely due to an 8–37% increase in the Constancy index, reflecting more stable interannual streamflow due to reservoir regulation.

Contributions

• This study presents the first application of a hybrid-physics-data (HPD) model, integrating VIC-CaMa-Flood with an LSTM network, to attribute streamflow and flood changes to climatic and human factors in the Lancang-Mekong River Basin.
• It demonstrates that the HPD approach provides more trustworthy results for attribution by effectively minimizing systematic errors inherent in purely process-based hydrological models.
• The research offers quantitative insights into the spatiotemporal shifts in the dominant drivers of hydrological changes, highlighting the increasing role of human activities in the post-2008 period.
• The findings provide valuable information and strategies for improved water resource management and flood control within the transboundary LMRB.

Funding

• National Natural Science Foundation of China (U2243226)
• NSFC-DFG Mobility (M-0468)
• CAS Interdisciplinary Innovation Team (xbzg-zdsys-202104)

Citation

@article{Liu2025Assess,
  author = {Liu, Binxiao and Yun, Xiaobo and Pan, Baoxiang and Xu, Ximeng and Gaffney, Paul P.J. and Lü, Hui and Luo, Lifeng and Sun, Gaohu and Tang, Qiuhong},
  title = {Assess the impacts of climatic change and human activities on streamflow and floods by using a hybrid-physics-data (HPD) model: A case study in the Lancang-Mekong River Basin},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.102763},
  url = {https://doi.org/10.1016/j.ejrh.2025.102763}
}