Chapagain et al. (2025) Evaluating the U.S. National Water Model Retrospective Evapotranspiration Simulation using Eddy-Covariance Flux Tower Measurements

Identification

• Journal: Journal of Hydrology Regional Studies
• Year: 2025
• Date: 2025-10-17
• Authors: Abin Raj Chapagain, Iman Maghami, Daniel P. Ames
• DOI: 10.1016/j.ejrh.2025.102826

Research Groups

• Department of Civil and Construction Engineering, Brigham Young University, Provo, UT, USA

Short Summary

This study provides the first national-scale evaluation of the U.S. National Water Model's (NWM) evapotranspiration (ET) simulations using eddy-covariance flux tower measurements across 72 sites in the contiguous United States. It found moderate overall performance, with better agreement in humid, forested regions and under water-limited conditions, but lower performance in arid, agricultural, and wetland areas, with temperature forcings not being a major source of error.

Objective

• To assess how effectively the NWM simulates evapotranspiration (ET) compared to eddy-covariance tower observations across the contiguous United States (CONUS), and how performance varies by land cover and hydroclimatic characteristics.
• To determine the extent to which biases in temperature forcings contribute to errors in NWM-simulated ET.
• To evaluate how NWM ET performance differs between energy-limited and water-limited regimes.

Study Configuration

• Spatial Scale: Contiguous United States (CONUS) on a 1 kilometer (km) grid; 72 eddy-covariance flux tower sites.
• Temporal Scale: Monthly comparisons; NWM V3.0 Retrospective simulation spanning 45 years (1979–2023); AmeriFlux tower data ranging from 1 to 29 years per site (median 5 years, average 7.9 years).

Methodology and Data

• Models used: U.S. National Water Model (NWM) based on the Weather Research and Forecasting Hydrological modeling system (WRF-Hydro) architecture, which integrates the community Noah land surface model with multi-parameterization options (Noah-MP).
• Data sources:
  • Eddy-covariance flux tower measurements (latent heat flux and air temperature) from the AmeriFlux network.
  • NWM Version 3.0 Retrospective simulation outputs for ET and meteorological forcings.
  • Analysis of Record for Calibration (AORC) V1.1 for NWM meteorological forcings (e.g., 2 meter (m) air temperature, precipitation rate).
  • National Land Cover Database (NLCD) 2019 for land cover classification.
  • K¨oppen-Geiger climate classification (Beck et al., 2018) for climate zones.
  • Performance metrics: Scaled modified Kling-Gupta Efficiency (KGEs), Percent Bias (PBIAS), and Coefficient of Determination (R²).

Main Results

• NWM-simulated ET showed a wide range of performance across sites, with Scaled Kling-Gupta Efficiency (KGEs) values from -0.14 to 0.86 (median: 0.48) and Percent Bias (PBIAS) from -75 % to +132 % (median: 11.4 %).
• Stronger agreement was generally observed in the Ohio and Northeast River Forecast Centers (RFCs), cold climates with warm summers, forested land covers (Mixed Forest, Deciduous Forest), and during colder months.
• Lower performance was more frequent in the California-Nevada RFC, temperate dry-summer climates, cultivated croplands, wetlands, and during warmer months. The California-Nevada RFC showed consistent ET underestimation and a phase mismatch, with the modeled ET peak occurring approximately four months earlier than observed.
• Most sites (67.2 %) performed better under water-limited (P/PET < 1) conditions, while 25.9 % performed better under energy-limited (P/PET > 1) conditions. Water-limited regimes dominated the dataset (68.4 % of site-months), especially in summer.
• The AORC V1.1 temperature forcings used by NWM V3.0 Retrospective showed strong agreement with tower-observed temperatures (R² values from 0.926 to 1.000, mean 0.991), indicating that temperature bias was not a major contributor to ET simulation errors.

Contributions

• This study provides the first national-scale evaluation of NWM-simulated ET using ground-based eddy-covariance observations, offering complementary insights to previous NWM assessments focused on streamflow, soil moisture, and snowpack.
• It identifies specific environmental and climatic conditions (RFCs, climate zones, land cover types, water/energy limitation, seasonality) under which the NWM performs well or exhibits systematic biases in ET simulation.
• The findings highlight opportunities for improving NWM ET simulation, particularly in underperforming conditions such as arid regions, cultivated croplands, and wetlands, by suggesting the need for better representation of anthropogenic water use (e.g., irrigation) and expanded ET calibration.
• It rules out temperature forcing errors as a major source of ET bias, redirecting focus to other drivers like precipitation forcing or internal Noah-MP land surface processes for future model improvements.
• The insights are critical for applications like water resource management, agricultural irrigation scheduling, and drought monitoring, and inform the development of a nationwide ET forecasting system within the Next-Generation National Water Model (NextGen NWM) framework.

Funding

• Cooperative Institute for Research to Operations in Hydrology (CIROH) under award NA22NWS4320003 from the NOAA Cooperative Institute Program.

Citation

@article{Chapagain2025Evaluating,
  author = {Chapagain, Abin Raj and Maghami, Iman and Ames, Daniel P.},
  title = {Evaluating the U.S. National Water Model Retrospective Evapotranspiration Simulation using Eddy-Covariance Flux Tower Measurements},
  journal = {Journal of Hydrology Regional Studies},
  year = {2025},
  doi = {10.1016/j.ejrh.2025.102826},
  url = {https://doi.org/10.1016/j.ejrh.2025.102826}
}