Quantifying the utility and uncertainty of multi-source fused precipitation products in hydrological simulations
(2025) In Journal of Hydrology: Regional Studies 61.- Abstract
Study Region: The Juzhang River Basin in Hubei, China. Study Focus: Precipitation is crucial information in hydrological modelling, and its uncertainty significantly affects the simulation performance. This study constructs a decision tree-based gridded fusion model to integrate multiple precipitation products, generating multi-source fused precipitation products (MSFPPs). The mesoscale hydrological model (mHM) is driven by these MSFPPs to simulate the hydrology of the Juzhang River Basin. Additionally, data from The Gravity Recovery and Climate Experiment (GRACE) and other sources are incorporated to validate the results and analyze simulation uncertainties arising from different precipitation products. New Hydrological Insights: The... (More)
Study Region: The Juzhang River Basin in Hubei, China. Study Focus: Precipitation is crucial information in hydrological modelling, and its uncertainty significantly affects the simulation performance. This study constructs a decision tree-based gridded fusion model to integrate multiple precipitation products, generating multi-source fused precipitation products (MSFPPs). The mesoscale hydrological model (mHM) is driven by these MSFPPs to simulate the hydrology of the Juzhang River Basin. Additionally, data from The Gravity Recovery and Climate Experiment (GRACE) and other sources are incorporated to validate the results and analyze simulation uncertainties arising from different precipitation products. New Hydrological Insights: The fusion algorithm notably enhances precipitation data accuracy, with precipitation products (PP) fused by XGBoost (XGBoost-PP) and random forest (RF-PP) showing superior performance, as indicated by improved Probability of Detection (POD), Critical Success Index (CSI), and reduced False Alarm Rate (FAR). MSFPPs effectively improve the accuracy of runoff modelling during the calibration period (NSE>0.70, KGE>0.78) and help reduce uncertainties in simulations of other variables. Precipitation contributes most to uncertainty in runoff simulation, with higher contribution rates in summer and autumn. In response to precipitation errors, terrestrial water storage (TWS) and soil moisture (SM) exhibit high sensitivity, while runoff and evapotranspiration (ET) demonstrate weaker responses. This study elucidates the response patterns of hydrological simulation uncertainties to precipitation errors, providing valuable insights for enhancing hydrological modeling in river basins worldwide.
(Less)
- author
- You, Zhiwen
; Wang, Kaixun
; Sun, Huaiwei
; Zhang, Hong
; Ye, Yuanyao
; Qu, Siyao
; Chen, Lin
; Li, Huixian
; Wang, Fengyu
; Qin, Hui
and Cai, Zhanzhang
LU
(Less) - organization
- publishing date
- 2025-10
- type
- Contribution to journal
- publication status
- published
- subject
- keywords
- Hydrological simulation, Machine learning application, MHM model, Precipitation products fusion, Uncertainty
- in
- Journal of Hydrology: Regional Studies
- volume
- 61
- article number
- 102620
- publisher
- Elsevier
- external identifiers
-
- scopus:105010877353
- ISSN
- 2214-5818
- DOI
- 10.1016/j.ejrh.2025.102620
- language
- English
- LU publication?
- yes
- id
- 2769ae55-c680-4c18-ad1b-a207a01a1c5a
- date added to LUP
- 2025-11-04 11:15:57
- date last changed
- 2025-11-04 11:16:28
@article{2769ae55-c680-4c18-ad1b-a207a01a1c5a,
abstract = {{<p>Study Region: The Juzhang River Basin in Hubei, China. Study Focus: Precipitation is crucial information in hydrological modelling, and its uncertainty significantly affects the simulation performance. This study constructs a decision tree-based gridded fusion model to integrate multiple precipitation products, generating multi-source fused precipitation products (MSFPPs). The mesoscale hydrological model (mHM) is driven by these MSFPPs to simulate the hydrology of the Juzhang River Basin. Additionally, data from The Gravity Recovery and Climate Experiment (GRACE) and other sources are incorporated to validate the results and analyze simulation uncertainties arising from different precipitation products. New Hydrological Insights: The fusion algorithm notably enhances precipitation data accuracy, with precipitation products (PP) fused by XGBoost (XGBoost-PP) and random forest (RF-PP) showing superior performance, as indicated by improved Probability of Detection (POD), Critical Success Index (CSI), and reduced False Alarm Rate (FAR). MSFPPs effectively improve the accuracy of runoff modelling during the calibration period (NSE>0.70, KGE>0.78) and help reduce uncertainties in simulations of other variables. Precipitation contributes most to uncertainty in runoff simulation, with higher contribution rates in summer and autumn. In response to precipitation errors, terrestrial water storage (TWS) and soil moisture (SM) exhibit high sensitivity, while runoff and evapotranspiration (ET) demonstrate weaker responses. This study elucidates the response patterns of hydrological simulation uncertainties to precipitation errors, providing valuable insights for enhancing hydrological modeling in river basins worldwide.</p>}},
author = {{You, Zhiwen and Wang, Kaixun and Sun, Huaiwei and Zhang, Hong and Ye, Yuanyao and Qu, Siyao and Chen, Lin and Li, Huixian and Wang, Fengyu and Qin, Hui and Cai, Zhanzhang}},
issn = {{2214-5818}},
keywords = {{Hydrological simulation; Machine learning application; MHM model; Precipitation products fusion; Uncertainty}},
language = {{eng}},
publisher = {{Elsevier}},
series = {{Journal of Hydrology: Regional Studies}},
title = {{Quantifying the utility and uncertainty of multi-source fused precipitation products in hydrological simulations}},
url = {{http://dx.doi.org/10.1016/j.ejrh.2025.102620}},
doi = {{10.1016/j.ejrh.2025.102620}},
volume = {{61}},
year = {{2025}},
}