Development of benchmarking datasets for testing and improving NexGen NWM predictions

Researchers have postulated that the future direction of hydrology will be determined by our ability to make high-quality predictions using global-scale earth system models. One of the challenges of earth system model development efforts is our inability to integrate computationally efficient algorithms to simulate and upscale subsurface processes for predicting unsaturated soil moisture transport and groundwater fluxes (e.g., recharge). The goal of this project is to develop data products and modeling methods to address this need. Our first objective is to develop recharge and groundwater datasets using a semi-distributed hydrological modeling tool SWAT (and its variant SWAT-MODFLOW) for a selected set of river basins in the southeastern region. The SWAT modeling efforts will help develop recharge and groundwater level datasets, which can be used as benchmark data in model intercomparison studies. The second objective is to build laboratory capabilities develop lab-scale benchmark datasets and use them to test and develop more efficient soil moisture transport solvers for supporting NexGen NWM developmental efforts.

As part of the first objective, we will develop a semi-distributed hydrological model for a major river basin in Alabama, the Black Warrior-Tombigbee Basin, using the SWAT model. We plan to compare the results of the model-based recharge data with other data-driven recharge products, such as the USGS product developed by Reitz et al. 2017. The original Reitz et al. dataset is a CONUS-scale dataset that employed a data-driven approach to estimate annual average values of recharge, runoff, and evapotranspiration (ET) at an 800 m resolution for the period 2000-2013. We will compare SWAT predictions with USGS data to better understand the quality of these groundwater recharge datasets. Depending on the outcomes of this study, we plan to do similar comparisons at other test catchments and utilize SWAT-MODFLOW to develop recharge and groundwater level datasets at different scales.

As part of the second objective, we propose to explore various laboratory and modeling methods for improving the methodologies used for simulating soil moisture transport processes. Large-scale earth system models, such as NextGen NWM, employ empirical formulations for describing soil water retention (SWR) and unsaturated hydraulic conductivity (UHC) functions, and these functions are rarely tested or benchmarked against real data. We propose to develop a novel, low-cost laboratory method, namely the Modified Evaporation (EV) method, for rapidly generating SWC and UHC functions. We are proposing to employ the modified EV method to characterize different types of surface soils collected from various field sites in the southeastern region. These data will be integrated with the sand/silt/clay data available in other soil databases to develop a more comprehensive map of unsaturated soil hydraulic properties. Our laboratory capabilities will also be used to develop a set of benchmark experiments for testing different types of soil moisture transport solver routines.

The data products and research outcomes of this effort will improve the methods used for simulating subsurface transport processes in NextGen NWM. Accurate simulations of these processes will be critical for obtaining robust operational drought forecasts at low flow conditions that will be predominantly dominated by unsaturated and saturated subsurface flow processes.