HRRRx marks the spot!
GSD’s experimental model nails a difficult and unusual forecast for Houston flooding
Flooding in Houston, TX April 18, 2016
As we approach the peak of the severe weather season, the modeling community is scrutinizing the next upgrade to NOAA’s storm-scale prediction model. The new model is due to NWS forecast offices later this spring, but the recent flooding event in Houston, TX is a good example of how researchers are watching every forecast it produces.
GSD’s hourly updating High-Resolution Rapid Refresh - Experimental (HRRRx), a regional analysis and weather prediction system, nailed the heavy rainfall forecast for Houston on April 18, 2016. Specifically, HRRRx accurately predicted 15-20 inches of rain in a 12-hour period just northwest of the city. This amount of rainfall is rarely observed, and for a model to be able to forecast it? Unprecedented.
In an online post, Colorado State University meteorologist and professor Russ Schumacher called the Houston forecast "One of the most impressive convection-allowing model forecasts I remember seeing." ["Convection-allowing" models that are high-enough resolution to characterize individual thunderstorms cells, along with simulating their internal atmospheric processes.] "Aside from the outstanding rainfall forecast, the representation of the storm dynamics was perhaps even more impressive."
HRRRx is the experimental and advanced version of NOAA’s High-Resolution Rapid Refresh (HRRR), which has been operational in the National Weather Service (NWS) only since 2014. But GSD’s work doesn’t stop after a product transitions from research to operations. They continuously push the model to the edge of computing power and scientific understanding. The next version, HRRRv2 makes better use of radar, satellite, lightning and other observations and is due to NWS forecast offices in June.
Researches are developing the HRRRx for a variety of users including sub-hourly information for the aviation and renewable energy communities, and atmospheric chemistry and physics for prediction of wildfire impacts including smoke. They are also working on expanded forecast coverage over Alaska and likelihood estimation of various weather hazards including heavy precipitation and severe weather. Eventually, these will merge back together into a powerful HRRR system that will help improve short-term forecasts of high impact weather.
Right now, HRRR-Ensemble (HRRRE) is being demonstrated in the NOAA Hazardous Weather Testbed. HRRRE is the foundation of the future Warn-on-Forecast prototype system for high-impact weather. To prepare to start the HRRRE model with current weather conditions, the model uses special techniques to bring surface, aircraft, and other observations into the early stages of the model run. HRRRE then runs multiple forecast models at the same time (called an ensemble), all starting with slightly different atmospheric conditions, to produce hourly snapshots of possible weather. HRRRE starts over every hour with merged forecast model output and real-time data to bring small-scale storm structure into focus.
This community effort to develop a suite of high-resolution models supports NOAA’s Weather-Ready Nation initiative. Partners include the Cooperative Institute for Research in Environmental Sciences, and the Cooperative Institute for Research in Atmospheric Sciences.
For more information contact: Susan Cobb 303-497-5093
High Resolution Rapid Refresh Experimental Model 12-hour precip forecast initiated April 18, 2016 00Z
MRMS Model 12-hour precip forecast ending April 18, 2016 12Z
Screen shot of Harris County Flood Warning System interface showing rainfall totals
