Yu et al. (2026) Season-land-use heterogeneity Bayesian Three-Cornered Hat (SLH-BTCH) for precipitation fusion in ungauged and sparsely gauged regions

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2026
• Date: 2026-01-13
• Authors: Qiangwei Yu, Xiaohua Dong, Zengchuan Dong, Yaoming Ma, Xiang Cheng, Xue'er Hu, Chengqi Gong, Bob Su, Wenzhuo Wang
• DOI: 10.1016/j.ejrh.2026.103110

Research Groups

• College of Hydraulic & Environmental Engineering, China Three Gorges University, Yichang, China
• Engineering Research Center for the Ecological Environment of the Three Gorges Reservoir Area, Ministry of Education, Yichang, China
• College of Water Resources and Hydrology, Hohai University, Nanjing, China
• Yellow River Research Center, Hohai University, Nanjing, China
• Land-Atmosphere Interaction and its Climatic Effects Group, State Key Laboratory of Tibetan Plateau Earth System, Environment and Resources (TPESER), Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China
• Faculty of Geo-Information Science and Earth Observation, University of Twente, Enschede, The Netherlands

Short Summary

This study introduces the Season–Land-Use Heterogeneity Bayesian Three-Cornered Hat (SLH-BTCH) method for fusing multi-source precipitation products in ungauged regions. SLH-BTCH significantly improves precipitation estimation accuracy and hydrological applicability by modeling error covariance based on season and land-use type, outperforming traditional BTCH and equal-weight averaging.

Objective

• To develop and evaluate the Season–Land-Use Heterogeneity Bayesian Three-Cornered Hat (SLH-BTCH) method to improve multi-source precipitation product fusion accuracy in ungauged and sparsely gauged regions by accounting for spatiotemporal error heterogeneity without requiring prior reference data.

Study Configuration

• Spatial Scale: Source region of the Yangtze River (Tibetan Plateau), approximately 600 kilometers (east-west) by 200 kilometers (north-south), with an average elevation exceeding 4500 meters. Data were resampled to a 0.1° × 0.1° grid.
• Temporal Scale: A 38-year daily precipitation series from 1981 to 2018.

Methodology and Data

• Models used:
  • Season–Land-Use Heterogeneity Bayesian Three-Cornered Hat (SLH-BTCH)
  • Bayesian Three-Cornered Hat (BTCH)
  • Equal-Weight Average (EWA)
  • The core principle involves Bayesian estimation of error covariance matrices from pairwise differences of multiple precipitation products, followed by weighted averaging, with SLH-BTCH applying this within season- and land-use-defined groups.
• Data sources:
  • Precipitation Products: CHIRPS, CMFD, TPHiPr, CHM-PRE.
  • Ground Observations (for evaluation): China National Surface Weather Station Daily Dataset (Version 3.0) and Hydrological Yearbooks from six meteorological stations and two hydrological stations.
  • Land Use Data: University of Maryland Global Land Cover Classification (UMD GLCC).

Main Results

• SLH-BTCH consistently outperformed BTCH and Equal-Weight Averaging (EWA) across nearly all evaluation metrics (Pearson correlation coefficient (CC), Root Mean Square Error (RMSE), Relative Bias (RB), Probability of Detection (POD), modified Kling–Gupta Efficiency (KGE), Mean Absolute Error (MAE), Nash–Sutcliffe Efficiency (NSE)) at six representative stations.
• SLH-BTCH achieved the lowest RMSE and MAE, demonstrating effective suppression of systematic bias and random fluctuations, with RB significantly reduced compared to BTCH and EWA, particularly at Tuotuohe, Wudaoliang, and Zhiduo.
• The method improved event timing and seasonal precipitation consistency with gauge observations, reduced storm-intensity bias, and decreased day-to-day noise across diverse land-surface types.
• Hydrologically, SLH-BTCH sharpened damaging-storm signals and reduced false alarms around key headwater stations, leading to tighter basin water-balance closure and more reliable flood simulation and dry-season water-availability estimates.
• Error analysis revealed significant seasonal patterns (e.g., RMSE peaking in July–August) and land-use dependencies (e.g., RMSE ranging from 1.49 mm for Alpine meadow to 2.01 mm for Sparsely Vegetated Land), which SLH-BTCH effectively addressed through its grouping strategy.

Contributions

• Introduces SLH-BTCH, a novel precipitation fusion method that explicitly models error covariance based on season and land-use type, addressing the limitation of spatially homogeneous error assumptions in traditional BTCH.
• Retains the "prior-free" advantage of BTCH, making it highly suitable for ungauged or sparsely gauged regions by estimating error covariance and weights from mutual differences among products without requiring in-situ gauge data for calibration.
• Demonstrates significant and consistent improvements in precipitation estimation accuracy, temporal consistency, and bias reduction compared to traditional BTCH and equal-weight averaging in complex, high-altitude terrain.
• Provides a lightweight and interpretable fusion framework that integrates physical controls on product errors (seasonality, land-use) instead of assuming a single homogeneous error structure.
• Enhances hydrological applicability by producing fused fields that better represent storm timing, intensity, and seasonal evolution, leading to more reliable flood simulation, water balance closure, and dry-season water availability estimates.

Funding

• Second Tibetan Plateau Scientific Expedition and Research (STEP) Program (Grant No. 2019QZKK0103)
• Major research project at Power China Guiyang Engineering Corporation (Limited) "Research on Key Technologies for Smart Water Network Construction in Guizhou Province" (Grant No. YJZDZX240001)
• "Dragon 6" project jointly sponsored by the European Space Agency and the National Remote Sensing Center of China (Grant No. 95357)

Citation

@article{Yu2026Seasonlanduse,
  author = {Yu, Qiangwei and Dong, Xiaohua and Dong, Zengchuan and Ma, Yaoming and Cheng, Xiang and Hu, Xue'er and Gong, Chengqi and Su, Bob and Wang, Wenzhuo},
  title = {Season-land-use heterogeneity Bayesian Three-Cornered Hat (SLH-BTCH) for precipitation fusion in ungauged and sparsely gauged regions},
  journal = {Journal of Hydrology Regional Studies},
  year = {2026},
  doi = {10.1016/j.ejrh.2026.103110},
  url = {https://doi.org/10.1016/j.ejrh.2026.103110}
}