Comprehensive Data and Timely Analysis with a New National Water Model

By Diana S. Sinton

In August 2016, the National Oceanic and Atmospheric Administration released one of its most significant technological achievements ever: the first version of its National Water Model, built on the WRF-Hydro modeling system. With this in place, NOAA aims to produce geographically continuous estimates of hydrological states and conditions across the continental United States.

It’s difficult to appreciate how noteworthy this accomplishment is, given how much has already been implemented by NOAA’s National Weather Service and the U.S. Geological Survey with regards to stream monitoring and flood forecasting. Currently almost half of USGS’s 8000+ stream gauges that monitor water flow are contributing live data to thirteen different River Forecast Centers. These centers can be essential sources of cautionary information for the regional communities they serve, such as in the case of Louisiana recently.

So what does the National Water Model add to what is already in place? How will it improve forecasting and reduce the risks that floods represent? What does it mean to the geospatial community that the NWM is up and running?

One way to gain perspective on what the NWM will be capable of is to remind yourself about all of the distinct elements of the complete water cycle, including its numerous phases and fluxes of energy and water across the terrain, the ocean and the atmosphere. The comprehensive water cycle is clearly much more complex than a simple connection between rain falling locally and an adjacent river flooding. Being able to access live and continuous stream gauge data is tremendous, but it’s also point data taken from a sub-selection of line data, necessarily providing us with only a limited glimpse into the full hydrologic scene.

In contrast, the NWM takes an integrated earth-systems’ physics approach. It begins its analytical processes by merging incoming meteorological data, generated by the numerical models and observations of the National Weather Service, with a 1-kilometer gridded land surface model that has incorporated data on soil moisture, snowpack, land cover, evapotranspiration, and plant canopy details, among many other relevant variables. Then, the processes being simulated by the land surface model go through a back-and-forth coupling with a 250-meter gridded terrain routing module. This higher resolution grid enables finer-scale modeling of routing such as overland and sub-surface flow, and applies numerous physics operations to predicting where water will be and how it will get there.

Next, the NWM model connects with the National Hydrography Dataset, the mapped system of surface water that includes 2.7 million unique stream and river reaches, ponds, lakes and other features of surface water. Finally, this accumulated information is applied through channel and reservoir routing modules. Soil drainage and channel inflow are understood at the levels of both catchments and channel reaches, and a new suite of forecasted products is possible.

The advantages of a more holistic system such as the NWM over previous efforts are numerous. The NWM enables predictions along river and stream reaches where none had existed before, due to inherent limitations of earlier systems. This expands access to more accurate risk-reduction information for previously underserved communities. Sub-components of the model are running continuously, allowing the model to produce forecasts at an hourly basis as well as several weeks out. Flood forecasting is a quintessential space-time challenge and any improvements to provide accurate warnings earlier will reduce the tremendous financial and societal burden that flooding represents.

Processing these amounts of data at these temporal scales constantly – to enable rapid responses to quickly changing circumstances – is not a trivial matter. Only high power computing makes it possible at all, and the team behind the NWM is leveraging academic and government collaborations to make this sustainable. A new National Water Center, based at the University of Alabama in Tuscaloosa, is an example of where these joint undertakings are underway. WRF-Hydro itself is a product of the Research Applications Laboratory at the National Center for Atmospheric Research, an entity of the University Corporation for Atmospheric Research. As David Gochis, an NCAR scientist, recently said, the NWM exemplifies the R2O-O2R paradigm, enacting research-to-operations and operations-to-research activities.

A related collaborative endeavor that promotes the NWM is the Consortium of Universities for the Advancement of Hydrologic Science, Inc., or CUAHSI. CUAHSI has just concluded a webinar series on the NWM that provides the public with detailed information on the model’s operations. Furthermore, for the last several years CUAHSI has hosted a Summer Institute at the National Water Center that brings together graduate students and other scholars to collectively focus on water-related projects and problems. Summaries of the 2016 research accomplished during this 7-week-long Innovators’ Program, when the topic was flood forecasting and inundation mapping, can be reviewed in Technical Report 13. The emphasis for the upcoming Summer Institute of 2017 will be hyper-resolution flood forecasting in urban areas. Applications are now being accepted.

Reaching this point also reflects the persistence of big-picture thinkers that have had a vision of what a national-level, spatially continuous forecasting model could be. The development of the NWM has benefited greatly by contributions from experts such as David Maidment, from the University of Texas, whose knowledge gained by years of applying GIS to hydrological processes and producing Arc Hydro has helped shape the NWM and its applications. Dr. Maidment continues to share his expertise by being part of CUAHSI’s Summer Institutes.

As much as has been achieved, the NWM is admittedly still in its first version. Model calibration is an ongoing process. Fortunately for them, weather keeps happening and water will almost always flow downhill, so efforts to improve inundation forecasting has a few enduring givens. The overall model will necessarily become more accurate over time as improvements are made to its sub-components. For example, inputs to the land surface model, such as the soil classification and land cover data, will continue to be refined and updated. Legacy shortcomings of the National Hydrography Database itself, such as local inconsistencies in the density of stream coverage, is a current issue. How humans influence hydrologic patterns through their management of water via reservoirs and dams is another area of further interest. At a future point, the NWM will be able to track nutrients and sediment flow, to consider water quality as well as water quantity. Some of these issues are long-term matters, but some improvements will be integrated into Version 2 of the NWM that will be released in April 2017.

What has the NWM meant for the geospatial technologies industry? For one thing, it’s yet another example of how the industry is diving into complex, big data projects. Only now is a model of this scope and scale, geographically and temporally, feasible from a computational stand-point. As one example, research at the CyberGIS Center for Advanced Digital and Spatial Studies at the University of Illinois has been linked with hydrologic terrain analyses, such as TauDEM at Utah State University, to help improve inundation mapping, a key part of the NWM. Overall, the broader industry benefits by an emphasis on open data and open source approaches. CUAHSI supports HydroShare, and an R package exists for working with WRF-Hydro as well.

In this era of extreme weather events that influence billions of people globally, and when advances in the domain of geospatial sciences and technologies is occurring constantly, the NWM represents a forward-looking approach – and a suitable modeling architecture that is both flexible and extensible – to accommodate and be responsive to new data and new sources of knowledge. Traditional flood risk maps face numerous challenges based on their sources of data, not to mention economics and politics. The NWM is a next generation approach to the age old questions about where water is and when it will pose its risks.



Published Wednesday, November 2nd, 2016

Written by Diana S. Sinton

If you liked this article subscribe to our bimonthly newsletter...stay informed on the latest geospatial technology

Sign up

© 2017 Directions Media. All Rights Reserved.