FLASH: Research
Four active research sub-topics related to the FLASH project are described below.
Flash Flood Observation Database
In order to evaluate forecasting tools, researchers need to use observations of flash flooding. When scientists at NOAA National Severe Storms Laboratory first began this research, they thought it would be possible to simply download flash flood observations from a unified database. However, they soon learned flash floods are not only difficult to predict, but they are also difficult to observe.
Since then, researchers assembled flash flood observations from United States Geological Survey (USGS) automated discharge measurements, National Weather Service trained spotter reports, and reports from NSSL’s Severe Hazards Analysis and Verification Experiment (SHAVE). This flash flood database is available for community research purposes. Currently, NSSL supports the database in Google Earth kmz format, GIS shapefile format, and comma-delimited text files that can be easily read into Excel.
map of flash flood observations
Evaluation of skill relative to benchmarks
In order to demonstrate improvement, NSSL researchers need to know how the current tools used for flash flooding perform. This is particularly important as the characteristics of the Multi-Radar Multi-Sensor forcings improve and change over time due to technological and algorithmic advances, like using the dual-polarization radar variables in estimating heavy rainfall rates. Analysis of these skill metrics as a function of basin-scale, geographic location, time of the year, and so forth, are very useful for guiding FLASH’s current use and focusing future development efforts. Researchers find these objective evaluations agree with subjective findings established from National Weather Service forecasters’ experience.
Probabilistic Products to Guide Impact-Based Flash Flood Warnings
The National Weather Service currently issues impact-based Flash Flood Warnings, that fall into base, considerable, and catastrophic impacts that are anticipated. These trigger different levels of actions by their end users. To help guide their warnings, probabilistic products are being developed through funding by the NOAA Joint Technology Transfer Initiative, managed by the NOAA Weather Program Office. These products will utilize the suite of product outputs, as well as information from prior cases that caused flash floods using artificial intelligence and machine learning (AI/ML). The post-processor will generate base-, considerable-, and catastrophic-level probabilities of flash flooding at each grid point.
Hydrologic Warn-on-Forecast
While researchers at NSSL presently focus on rainfall forcing from the Multi-Radar Multi-Sensor observations, lead time may be increased using forecasts of rainfall from storm-scale numerical weather prediction models. While these forecast products are quickly improving following enhanced grid cell resolution, explicit physical process representation, and radar data assimilation, errors with the specific locations of intense rainfall are still common. We are currently developing techniques to incorporate precipitation forecasts as forcing to the hydrologic models that also account for locational uncertainties while maintaining a high cadence of generating products.
Additional information about NSSL's research and development efforts for severe weather warning tools can be found here.