Benchmarking Model Formulations for Operations with the Next Generation Water Resources Modeling Framework

The Next Generation Water Resources Modeling Framework (NextGen) provides unparalleled flexibility in model coupling and execution, thanks to its use of the Basic Model Interface (BMI) and a novel architecture that supports various programming languages. Like many technological advancements, NextGen enables researchers and practitioners to perform experiments that may have previously been intractable while also creating a new set of challenges. For example, if BMI and NextGen let us couple and run a suite of independently developed hydrologic models, how do we determine which model formulations perform best and where? This presentation covers our first attempt at answering this question.

Our benchmarking exercise draws on lessons learned from the previous years of NextGen Framework development, using software tools and data products created by the NOAA National Weather Service (NWS) Office of Water Prediction (OWP). These include the NextGen Framework itself, along with the hydrofabric to define the simulation domains, the Analysis of Record for Calibration (AORC) meteorological forcing data, a set of hydrologic models that comprise the formulations we evaluate, the T-Route streamflow routing module, and model-agnostic ngen-cal calibration tools.

Our first step was to identify a set of larger gaged basins using the Harmonized Attributes of River Basins in One Repo (HARBOR) dataset and input from NWS River Forecast Centers. This then directed our efforts to subset the hydrofabric and AORC forcing data. We next identified calibration strategies and time periods for model calibration and validation. Once we had our strategy and supporting data, we created a cloud build of NextGen and the model formulations, which included Noah-OWP-Modular, Snow-17, potential evapotranspiration (PET) routines, version 2.0 of the Conceptual Functional Equivalent (CFE), TOPMODEL, Sac-SMA, and the Lumped Arid and Semi-Arid Model (LASAM), plus other peripheral software. In this presentation we will detail the outcomes from calibrating and evaluating these formulations and identify next steps for improving formulation performance.