Evaluating Baseflow Simulation in the National Water Model: A Case Study in the Northern High Plains Region, USA

Abstract

The National Water Model (NWM) is a high‐resolution hydrological model capable of providing streamflow forecast at 2.7 million reaches across the conterminous United States (U.S.). It utilizes a conceptual (not physically explicit) module for estimating groundwater discharge (baseflow) to streams, and the baseflow module only allows for one‐way flux from the aquifer to the streams, with no interflow between surficial or groundwater catchments. This study evaluated the ability of the NWM to simulate baseflow in different geologic conditions with a case study in the Northern High Plains region. The study also evaluated two alternative formulations for the baseflow module in the NWM with the goal of improving simulated baseflow accuracy. The NWM simulated baseflow more accurately in a clayey catchment than in a sandy catchment. The model, however, over simulated baseflow during storm runoff events and under simulated during low flow events in both catchments. Analysis of groundwater recharge simulation showed that the NWM over simulated groundwater recharge for both catchment types, and reducing groundwater recharge using U.S. Geological Survey Baseflow Indexes improved the model performance for baseflow simulation, especially in the sandy catchment. Evaluation of the alteranative formulations showed that a modified Rorabaugh‐Rutledge formulation was able to improve the baseflow simulation with RMSE of 7.98 m3/s and R2 of 0.41 and also improved the timing of peak flows.

Modeling the Hydrologic Influence of Subsurface Tile Drainage Using the National Water Model

“…Even though CalibTD reduced low‐flow biases compared to Calib , it still overestimated low flows by 50%. Analyzing the distributions of NSE (Figure 5g), NSE WT (Figure 5i), and KGE (Figure 5k) indicated that the NWM, in general, failed to reproduce observed low flow accurately, consistent with previous studies (Hansen et al., 2013; Jachens et al., 2021; Karki et al., 2021). One of the reasons for the overestimation of low flows can be the high groundwater recharge (deep percolation loss) rate in the NWM (Karki et al., 2021).…”

“…Analyzing the distributions of NSE (Figure 5g), NSE WT (Figure 5i), and KGE (Figure 5k) indicated that the NWM, in general, failed to reproduce observed low flow accurately, consistent with previous studies (Hansen et al, 2013;Jachens et al, 2021;Karki et al, 2021). One of the reasons for the overestimation of low flows can be the high groundwater recharge (deep percolation loss) rate in the NWM (Karki et al, 2021). The existing groundwater scheme in the NWM represents surface water-groundwater connectivity using a one-way connection from the underlying aquifer to the stream channel and omitted the influences of the stream on groundwater, and ignoring the two-way stream-aquifer fluxes in the NWM lead to overprediction of low flows (Jachens et al, 2021).…”

“…To provide flash flood forecasts and other hydrologic guidance with longer lead time and less uncertainties, the National Weather Service (NWS) Office of Water Prediction (OWP) of the National Oceanic and Atmospheric Administration (NOAA) developed a hydrologic modeling framework, the National Water Model (NWM), to simulate observed and forecast streamflow for about 2.7 million stream reaches of the CONUS. However, the NWM has considerable uncertainties in the streamflow prediction over the Midwestern US (Dugger et al., 2017; Karki et al., 2021). One of the reasons for the underperformance of the NWM can be the lack of representation of subsurface TD hydrology in the NWM (Hansen et al., 2013).…”

Representing Bidirectional Hydraulic Continuum Between the Stream and Hillslope in the National Water Model for Improved Streamflow Prediction

“…This simplified baseflow estimation is also not a physically explicit representation of the aquifer system, so the fitting parameters must be introduced (i.e., C and Gm Equation 6) and empirically derived. More importantly, as reported in other studies (Karki et al, 2021), two things should be noted about the characteristics of the modeled baseflow in the NWM: (a) the lack of GW storage is found in most NLSR(s), meaning the total amount of NLSR inflow (i.e., soil drainage) is almost identical to that of NLSR outflow (i.e., baseflow), (b) almost no time lag (mostly less than 1 hr) between GW inflow and outflow is found. The (almost) linear relationship between the river discharge Q and GW storage S (i.e., Q 𝐴𝐴 ≅ S) thus could be concluded from the current NLSR configuration as the result of the above two factors (a) and (b).…”

Insights From Dayflow: A Historical Streamflow Reanalysis Dataset for the Conterminous United States

“…A comprehensive evaluation of large-scale high-resolution streamflow reanalyses has not yet been reported except for a few at regional scales (e.g., Hansen et al, 2019;Jachens et al, 2021;Karki et al, 2021;Rojas et al, 2020). Using a lumped hydrologic modeling framework, provided a well-calibrated Catchment Attributes and Meteorology for Large-sample Studies (CAMELS) streamflow reanalysis dataset for 671 small-to-medium headwater basins across the CONUS, but such an approach cannot provide flow estimates for all river segments in the basin.…”