A new framework could transform national flood prediction - Earth - Environment

Conceptual model showing the relative size of the parameter space required (brown band) and resulting predictability (green band) for modeling stormflow generation as a function of land surface-, soil-, and groundwater-dominated mechanisms. This conceptual model considers infiltration–excess and saturation–excess storm flow generation mechanisms as highly predictable end members. Credit: JAWRA Journal of the American Water Resources Association (2026). DOI: 10.1111/1752-1688.70089

When severe weather strikes, the National Weather Service's (NWS) Office of Water Prediction (OWP) makes critical flood forecasts with the National Water Model. Despite improvements over time, the model's performance has plateaued in recent years, leaving researchers from the federal government, academia, and private industry searching for a better solution.

Now a new set of software tools, the Next Generation Water Resources Modeling framework (NextGen), will help develop better predictions. As detailed in a new study led by OWP and the University of Vermont (UVM), with collaborators from nine other institutions, NextGen is a novel framework that allows OWP to make more accurate, representative flood forecasts; and allows researchers, students, and practitioners to experiment with and develop more advanced hydrologic models.

The paper is published in the JAWRA Journal of the American Water Resources Association.

"This paper represents years of close collaboration between scientists and software engineers at the National Weather Service, other federal agencies, the private sector, and universities across the country," said Dr. Keith Jennings, the Director of Research at UVM's Water Resources Institute. "The ultimate goal is to create more accurate, timely flood forecasts with the National Water Model to save lives during severe weather events. The NextGen framework is the software platform we created to do just that."

Complex challenges in water prediction

In addition to improving flood forecasts, NextGen work was also inspired by some of the biggest challenges facing hydrologists.

Water moves through the landscape in complex ways, traversing terrain at varying rates and across different routes. This unpredictability causes numerous challenges for researchers, developers, water managers, and operational forecasters when it comes to water prediction.

There are sizable gaps in understanding water processes from streamflow to rainfall and snowmelt. Hydrologists and water resources engineers complement field studies by employing or developing computational models to understand this hard-to-observe phenomenon. These models can overcome shortcomings in measurements, process understanding, and forecasting.

But computational models, with their variety of formulations, programming languages, compilers, data models, and required inputs, require considerable effort to understand and apply. So researchers and practitioners tend to choose models based on familiarity instead of the model that is most appropriate for their study and objectives.

There is no one perfect model.

How the NextGen framework works

To support water prediction needs and enable greater flexibility to use the right model in the right place at the right time, NextGen is "model-agnostic." It uses common standards, and is user-friendly and open-source.

The framework enables scientific evaluation of water prediction models that simulate diverse hydrologic and hydraulic processes. Its design supports models written in multiple programming languages, and runs on laptops, cloud, and supercomputers.

These capabilities will advance hydrologic science and help OWP make better forecasts. The National Water Model, which is slated to use the NextGen framework for its next operational version, needs to provide accurate water resources modeling for a variety of purposes, including flood forecasting, reservoir operations, and drought prediction.

However, the current operational National Water Model, like all computer models, suffers biases and errors that affect its predictive capabilities. Its performance also varies, working best in the Pacific Northwest and northern Rocky Mountains. These challenges, combined with the ever-present need for improving hydrologic forecast accuracy, created a need for a new modeling framework.

Key capabilities and future impact

NextGen has a wide range of capabilities that will enhance the next National Water Model. These include 1) the use of user-defined, not model-specific, geospatial data to define model domains, initialize parameter values and model states, and to build stream networks for flow routing, 2) the execution of various models and modules in the same framework using the same configuration system and execution commands, and 3) the production of model outputs in a standardized format over a consistent domain regardless of the chosen model.

NextGen has capabilities that promise to advance hydrologic and hydraulic modeling. By providing a common operating environment, collaboration will be increased between the research community and federal agencies. This will speed up operational implementation, initiate an increased pace of discovery, create more accurate and less redundant modeling, and ultimately, provide more effective water resource forecasts and predictions. 

Provided by University of Vermont 

by Hannah Fischer, University of Vermont

edited by Stephanie Baum, reviewed by Robert Egan 

Source: A new framework could transform national flood prediction