Assessing the National Water Model’s Streamflow Estimates Using a Multi-Decade Retrospective Dataset across the Contiguous United States

Mohamed Abdelkader, Marouane Temimi, Taha B.M.J. Ouarda

Research output: Contribution to journalArticlepeer-review

15 Scopus citations

Abstract

The goal of this study is to evaluate the performance of the National Water Model (NWM) in time and space across the contiguous United States. Retrospective streamflow simulations were compared to records from 3260 USGS gauging stations, considering both regulated and natural flow conditions. Statistical metrics, including Kling–Gupta efficiency, Percent Bias, Pearson Correlation Coefficient, Root Mean Squared Error, and Normalized Root Mean Squared Error, were employed to assess the agreement between observed and simulated streamflow. A comparison of historical trends in daily flow data between the model and observed streamflow provided additional insight into the utility of retrospective NWM datasets. Our findings demonstrate a superior agreement between the simulated and observed streamflow for natural flow in comparison to regulated flow. The most favorable agreement between the NWM estimates and observed data was achieved in humid regions during the winter season, whereas a reduced degree of agreement was observed in the Great Plains region. Enhancements to model performance for regulated flow are necessary, and bias correction is crucial for utilizing the NWM retrospective streamflow dataset. The study concludes that the model-agnostic NextGen NWM framework, which accounts for regional performance of the utilized model, could be more suitable for continental-scale hydrologic prediction.

Original languageEnglish
Article number2319
JournalWater (Switzerland)
Volume15
Issue number13
DOIs
StatePublished - Jul 2023

Keywords

  • NWM
  • continental-scale hydrological model
  • model performance
  • natural flow
  • regulated flow
  • simulated streamflow

Fingerprint

Dive into the research topics of 'Assessing the National Water Model’s Streamflow Estimates Using a Multi-Decade Retrospective Dataset across the Contiguous United States'. Together they form a unique fingerprint.

Cite this