Researchers from the National Oceanic and Atmospheric Administration (NOAA) have detailed how the Multi-Radar Multi-Sensor (MRMS) and Flooded Locations and Analysis in Short-term Hydrology (FLASH) systems enhance flash flood predictions in the United States using real-time data and physics-based modeling.

Flash floods are among the deadliest weather hazards in the United States, with a national average of 88 flood deaths per year. Lightning accounts for an average of 41 deaths annually, and tornadoes claim an average of 68 lives each year. Six inches of fast-moving flood water can knock a person off their feet, while twelve inches can carry away a car.

JJ Gourley, a researcher at the NOAA National Severe Storms Laboratory (NSSL), said, "The bigger killers are the ones that are difficult to predict." He added, "These are going to be driven by convective storms." Gourley noted, "They could be supercells or storms that train over the same area, over and over again." He explained, "These tend to produce really intense rainfall rates and you can see several inches of precipitation just a matter of minutes."

The MRMS system integrates raw data from weather radars, rain gauges, satellites, and numerical models. This data feeds into models to improve forecasts, providing forecasters with a complete image of atmospheric conditions in real-time. FLASH, a suite of products developed at the NOAA National Severe Storms Laboratory, translates MRMS rainfall data into runoff predictions using physics-based modeling.

FLASH predicts water flow for more than 10 million grid points across the United States every ten minutes. Race Clark, an NSSL research meteorologist and MRMS program lead, said, "Many of the worst flash flooding impacts aren't occurring on big rivers." Clark added, "They're occurring on small streams and other areas that aren't necessarily well-observed using traditional observations." He stated, "MRMS and FLASH fill in those gaps."

The Warn-on-Forecast System (WoFS) contributes to these efforts by generating multiple storm scenarios. WoFS uses advances in computing, artificial intelligence, and machine learning to produce rainfall rates and other severe weather outputs several hours into the future. These WoFS scenarios are then processed through hydrologic models to anticipate where flooding could develop. Gourley said, "We're going to help forecasters improve their ability to forecast impacts and also at much finer resolution—even down to intersections—where we're going to have flooded areas that can be anticipated hours in advance."