Deploying Coupled Snow and Runoff Models in the Next Generation Water Resources Modeling Framework
Abstract
Accurate streamflow forecasts in cold and temperate regions require models that skillfully represent both snowmelt and runoff processes. However, in some cases, a model that performs well at doing the former does not at the latter. Over large spatial extents and a spread of hydroclimatology, the optimal forecast configuration typically varies because there is no one best model. In the United States, the National Water Model (NWM) provides hourly streamflow forecasts for several time windows at approximately 2.7 million locations, covering a vast range of hydrologic regimes. Like all models, it faces the "no one best model" challenge. To that end, a collaboration between NOAA's Office of Water Prediction, other government agencies, research institutes, universities, and the private sector is developing and implementing the Next Generation Water Resources Modeling Framework (NextGen) to support future NWM versions plus other research and operations applications at process appropriate scales.
A key feature of NextGen is that it enables plug-and-play interoperability for models that incorporate the Basic Model Interface (BMI) for easy coupling and runtime control by a centralized framework. In practice, this means we can implement model combinations and configurations that are optimized for a location and application. To better represent snow and runoff processes, we have implemented a preliminary model set that includes Noah-OWP-Modular, Snow-17, the Conceptual Functional Equivalent (CFE), TOPMODEL, and the Long Short Term Memory (LSTM) machine learning model. Through NextGen and BMI, we can, for example, couple a physics-based model like Noah-OWP-Modular to simulate snow accumulation and melt to the conceptual TOPMODEL, which simulates runoff. We can also evaluate the different configurations to identify which combination or single model best matches observed streamflow in a given location. This presentation will cover the improvements we have made to snow process representation along with an initial investigation of streamflow simulated by the different model sets.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.H16F..03J