Performance of the National Water Model in Iowa Using Independent Observations

Abstract

This paper explores the performance of the analysis‐and‐assimilation configuration of the National Water Model (NWM) v1.0 in Iowa. The NWM assimilates streamflow observations from the United States Geological Survey (USGS), which increases the performance but also limits the available data for model evaluation. In this study, Iowa Flood Center Bridge Sensors (IFCBS) data provided an independent nonassimilated dataset for evaluation analyses. The authors compared NWM outputs for the period between May 2016 and April 2017, with two datasets: USGS streamflow and velocity observations; Stage and streamflow data from IFCBS. The distribution of Spearman rank correlation (rs), Nash–Sutcliffe efficiency (E), and Kling–Gupta efficiency (KGE) provided quantification of model performance. We found the performance was linked with the spatial scale of the basins. Analysis at USGS gauges showed the strongest performance in large (>10,000 km2) basins (rs = 0.9, E = 0.9, KGE = 0.8), with some decrease at small (<1,000 km2) basins (rs = 0.6, E = −0.25, KGE = −0.2). Analysis with independent IFCBS observations was used to report performance at large basins (rs = 0.6, KGE = 0.1) and small basins (rs = 0.2, KGE = −0.4). Data assimilation improves simulations at downstream basins. We found differences in the characterization of the model and observed data flow velocity distributions. The authors recommend checking the connection of USGS gauges and NHDPlus reaches for selected locations where performance is weak.

Continental Scale Assessment of Variation in Floodplain Roughness with Vegetation and Flow Characteristics

“…Many attempts are currently being made with the purpose of reducing the uncertainty of nationwide models (Johnson et al, 2019;Rojas et al, 2020), but have been focused on improving its performance by updating the geometry and roughness parameters of the main channel, without extending to the floodplain (Heldmyer et al, 2022). Integrating the results from our work on floodplain roughness at USGS gauge locations into the NWM could be a logical next step.…”

Comprehensive analysis of the NOAA National Water Model: A call for heterogeneous formulations and diagnostic model selection

“…For example, Salas et al evaluated an uncalibrated version of WRF-Hydro for the summer of 2015 at 5,700 gauges, providing a benchmark for the evolving hydrology program within the National Weather Service (Salas et al, 2017). Lin et al evaluated streamflow prediction in Texas, finding that dry regions are strongly affected by a positive bias (Lin et al, 2018) and Rojas et al evaluated NWM v1.0 in Iowa finding performance was linked to the size of the contributing basins with the best performance occurring in basins larger than 10,000 km 2 (Rojas et al, 2020).…”

Neglecting Model Parametric Uncertainty Can Drastically Underestimate Flood Risks

“…These previous studies break important new ground, but are mostly silent on the impacts of parametric uncertainties on hazards and dynamics. Neglecting parametric uncertainties can underestimate the tails of flood hazard probability distribution (Bates et al., 2021; Mendoza et al., 2015; Rojas et al., 2020; F. R. Salas et al., 2018), and can result in poor decisions and outcomes (Ruckert et al., 2019; Wong & Keller, 2017; Zarekarizi et al., 2020).…”