One of the most intractable problems facing those who issue flash flood warnings and watches is verification. The term “flash flooding” means different things to different people, so observing these events is distinct from, say, observing a tornado. Most people immediately know what a tornado is but do you immediately know what a flash flood is? It’s okay if you answered “no”! We recognize that different strategies exist for locating and classifying flash floods. In an effort to be as comprehensive as possible, the HWT-Hydro experiment is including observations from several distinct sources.
USGS Stream Gauges
Automated systems can provide useful information in these situations. USGS stream gauges automatically measure stream height (or “stage”) at over 16,000 sites across the United States every 15 minutes. At these sites, cross sectional information about the river channel is known and so these stage measurements can be converted into stream flow estimates. Additionally, the USGS and its partner agencies know enough about these locations to estimate the relative rarity (or “return period) of a particular flow, and at some locations, the National Weather Service has assigned flood stages corresponding to specific impacts at specific stream height levels. On flash.ou.edu, blue triangles mark stream gauge locations where the NWS “Action Stage” stage height has been exceeded by the gauge, the stream height has risen by a significant level in a short period of time, or the USGS estimated 2-year return period flow has been surpassed. These USGS observations are being fed to the FLASH system and the HWT-Hydro experiment by our partners at the NCEP Weather Prediction Center’s FFaIR experiment. The downside of USGS observations should already be clear: what happens if there’s a flood in an ungauged location? The answer to this question means we must seek out additional sources of information.
Local Storm Reports
The NWS operates 122 Weather Forecast Offices across the country, each of which is responsible for a set of counties, parishes, or boroughs. These offices issue flash flood watches and warnings as necessary. They also collect reports of flash flooding and other phenomena (called Local Storm Reports, or LSRs) because as forecasters, they have a vested interest in determining if a given watch or warning was successfully verified by on-the-ground conditions. So if they don’t receive reports volunteered by the public, broadcasters, social media, or local government officials, they will call known sources inside a watch or warning until they determine if flooding actually occurred. The NWS is responsible for the whole US so the LSRs should be nearly comprehensive but there are caveats to this: many areas with low or no population density will lack appropriate LSRs and in areas where no warning is issued there still could have been flash flooding. Outside of these warning polygons, if no one bothers to report impacts to their local NWS office no LSR will be issued. On the flash.ou.edu website, LSRs are represented by reddish-brown circles. They can be moused over to see the date and time at which the report was issued (in UTC) as well as the text describing the nature of the flash flooding impact.
Some folks may experience flash flooding but can’t, don’t know how, or don’t want to report this via phone to local officials or to their local NWS office. However, as smartphones proliferate, useful weather information can be garnered via crowd sourcing. One of the leading projects in this field was developed at the National Severe Storms Laboratory and the University of Oklahoma — mPING. mPING allows geolocated smartphone owners to report the type of precipitation or severe weather impact (including flash flooding) occurring at their location. This data is used by scientists at OU and NSSL, as well as NWS forecasters and others. All reports can be accessed via the mPING website and flash flooding reports in particular can be accessed via the FLASH website, where they appear as purple equilateral triangles. A report of flash flooding can fall into one of four possible impact classes in the mPING app, ranging from nuisance and minor flooding all the way up to deadly or damaging flooding. Remember, safety comes first when using mPING. Never drive into flooded roadways or walk into floodwaters! The free mPING app is available for iOS devices and Android devices from the Apple App Store or the Google Play Store, respectively.
One final source of observations is unique to the HWT-Hydro project. As part of the support provided to the program by the US Weather Research Program, we have hired two callers to work as part of NSSL’s 2014 SHAVE project. As our forecast participants and other guests issue experimental watches and warnings each afternoon and evening, SHAVE callers will use a national database of landline telephone numbers mapped to addresses to call people residing within and near these experimental products and give them telephone surveys about flash flooding impacts in and near their areas. These reports thus provide an extremely dense storm focused set of observations. On flash.ou.edu, SHAVE reports will appear as small squares. Those surveys where the respondent reported no flash flooding will appear in gray. Respondents reporting flash flooding will appear in blue.
Early each afternoon, HWT-Hydro participants will look back at the products they issued the afternoon before. They will evaluate the skill and usefulness of forecast tools and the utility of the various types of observations described above. Yesterday our forecasters split into two areas: one pair focused on Tennessee, Arkansas, and the surrounding area, while the other pair worked in the Desert Southwest with the ongoing monsoon.