Li et al. (2025) Evaluating multi-source precipitation datasets for hydrological applications in ungauged alpine region of Tibetan Plateau

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-12-01
• Authors: H. Y. Li, Jie Chen, Lu Li
• DOI: 10.1016/j.ejrh.2025.103003

Research Groups

• State Key Laboratory of Water Resources Engineering and Management, Wuhan University, Wuhan, PR China
• Hubei Key Laboratory of Water System Science for Sponge City Construction, Wuhan University, Wuhan, PR China
• NORCE Norwegian Research Centre, Bjerkernes Centre for Climate Research, Bergen, Norway

Short Summary

This study evaluates the accuracy of five multi-source gridded precipitation datasets for hydrological applications in an ungauged alpine region of the Tibetan Plateau using the physically-based WRF-Hydro/Glacier model, finding that reanalysis datasets (ERA5-Land, TPReanalysis) provide the most realistic streamflow simulations.

Objective

• To evaluate the accuracy of multi-source gridded precipitation datasets (global reanalysis, regional reanalysis, merged, and satellite products) for hydrological applications in an ungauged alpine region of the Tibetan Plateau.
• To intercompare the spatial and temporal characteristics of these precipitation datasets.

Study Configuration

• Spatial Scale: PLZB river basin, southeast of the Tibetan Plateau, with an area of approximately 6858 km² and elevations ranging from 2725 m to 6260 m. The Land Surface Model (LSM) domain of WRF-Hydro/Glacier used a 5 km grid resolution, with sub-grid routing processes executed at a 500 m resolution.
• Temporal Scale: The study period for data and simulations was 2004–2018. The model was calibrated for 2010–2014 and validated for 2004–2007. Precipitation datasets varied in temporal resolution from 30 minutes to 3 hours.

Methodology and Data

• Models used:
  • Weather Research and Forecasting-Hydrological Glacier modelling system (WRF-Hydro/Glacier).
  • WRF-Hydro (distributed hydrological model) coupled with Noah-MP (Land Surface Model).
  • Crocus (multi-layer, physically-based snow model) for glacier grids.
• Data sources:
  • Precipitation datasets:
    • ERA5-Land (Global reanalysis, 0.1° spatial resolution, hourly temporal resolution).
    • TPReanalysis (Regional reanalysis, 9 km spatial resolution, 1 hour temporal resolution).
    • TPMFD (Merged, 1/30° spatial resolution, hourly temporal resolution).
    • CMFD (Merged, 0.1° spatial resolution, 3 hour temporal resolution).
    • GPM (Satellite-based, 0.1° spatial resolution, 30-minute temporal resolution).
  • Meteorological forcing (from CMFD for all simulations): Wind speed, specific humidity, near-surface air temperature, surface pressure, downward shortwave and longwave radiation.
  • Observed data: Daily streamflow at the watershed outlet for the 2004–2018 period.

Main Results

• All five precipitation datasets exhibited similar spatial distribution patterns (decreasing precipitation from southwest to northeast) and intra-annual variability (inverse-V-shape, peak in July for most), but showed large discrepancies in precipitation amounts.
• ERA5-Land and TPReanalysis had the largest watershed-averaged annual precipitation (>1450 mm), followed by TPMFD, while GPM and CMFD had the least (<725 mm).
• TPReanalysis showed the largest magnitude, longest duration, and greatest intensity of extreme precipitation, whereas CMFD had the smallest values across these metrics.
• For hydrological performance, streamflow simulated using ERA5-Land and TPReanalysis most closely matched observed streamflow, achieving Nash-Sutcliffe efficiency coefficients (NSE) above 0.86 and relative biases within 16% for the 2004–2018 period. These datasets also captured high-flow events well (mean annual biases < 8.0% during the wet season).
• TPMFD performed slightly worse than ERA5-Land and TPReanalysis (Kling-Gupta efficiency (KGE) of 0.74, relative bias of -21.29% for the whole period) but still better than CMFD and GPM.
• GPM and CMFD considerably underestimated streamflow time series (relative biases of -42.08% and -45.06% respectively for the whole period), with underestimation exceeding 70% during the dry season.
• ERA5-Land and TPMFD showed the smallest biases in peak flow and time to peak (e.g., ERA5-Land median peak flow bias of -0.64% and time to peak bias of 1 day).
• Seasonally, ERA5-Land had the smallest biases in spring (-9.79%) and summer (-1.16%) streamflow, while TPReanalysis performed best in autumn (1.22%) and winter (-30.03%) streamflow. All datasets underestimated winter streamflow.
• Runoff coefficients calculated using TPMFD, GPM, and CMFD were greater than 1, indicating precipitation underestimation, while TPReanalysis (0.80) and ERA5-Land (0.78) yielded more realistic values.

Contributions

• Proposed and demonstrated a physically-based hydrological modeling approach using WRF-Hydro/Glacier to evaluate gridded precipitation datasets in ungauged alpine regions, addressing the challenge of insufficient ground-based validation.
• Provided a comprehensive evaluation of multi-source precipitation datasets (global reanalysis, regional reanalysis, merged, and satellite products) in the Tibetan Plateau, an area where such integrated assessments were previously limited.
• Highlighted the superior performance and potential of high-resolution reanalysis datasets (ERA5-Land and TPReanalysis) for improving hydrological studies in Asian alpine regions, challenging traditional evaluations based on potentially underestimated gauge observations.

Funding

• National Natural Science Foundation of China (Grant Nos. 52479024, W2412158, and U2240201)
• Overseas Expertise Introduction Project for Discipline Innovation (111 Project) funded by Ministry of Education and State Administration of Foreign Experts Affairs P.R. China (Grant No. B18037)

Citation

@article{Li2025Evaluating,
  author = {Li, H. Y. and Chen, Jie and Li, Lu},
  title = {Evaluating multi-source precipitation datasets for hydrological applications in ungauged alpine region of Tibetan Plateau},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.103003},
  url = {https://doi.org/10.1016/j.ejrh.2025.103003}
}