The impact of tropical rainfall rates on flash flood detection

Heavy convective rainfall needed to produce flash flooding can be produced by two distinct physical rainfall production mechanisms: the ice crystal mechanism, and the coalescence mechanism. The precipitation algorithm of the Weather Surveillance Radar, 1988 Doppler (WSR-88D) accounts for these two physical mechanisms by providing a choice of a standard convective rainfall rate (ice crystal mechanism), or a tropical convective rainfall rate (the coalescence mechanism). The radar operator must manually select which of these two rainfall rates will be used by the WSR-88D to estimate rainfall.

While the occurrence of tropical rainfall rates in tropical cyclones is expected, tropical rainfall rates can and do occur with thunderstorms not associated with tropical cyclones. The detection of flash flood producing rainfall with the WSR-88D can be greatly impacted by the occurrence of unanticipated tropical rainfall rates. Many cases of major “non-tropical storm” related flash flood events in the continental United States have resulted from heavy rainfall produced by tropical rainfall rates. Some examples include the flash floods in Johnstown, PA (1977), Brady’s Bend, PA (1980), Hyndman, PA (1984), Shadyside, OH (1990), Brookville, PA (1996), Dallas, TX (1995), Fort Collins, CO (1997), and Kansas City, MO (1998). The detection of evolving flash floods with the Flash Flood Monitoring and Prediction System (FFMP) may depend on anticipating the occurrence of tropical rainfall rates.

The impact of tropical rainfall rates on flash flood detection is shown by comparing the radar rainfall estimates for the Shadyside, OH (1990) flash flood using both standard convective and tropical convective rainfall rates. The occurrence of tropical rainfall rates can be anticipated by the examination of sounding parameters linked to heavy rainfall production. The sounding parameters related to heavy rainfall production will be discussed in some detail. The near real-time detection of tropical rainfall rates requires a comparison of radar rainfall estimates with rain gage readings. FFMP provides these real-time comparisons of WSR-88D rainfall estimates with rain gage readings. The disaster survey team for the Shadyside, OH (1990) flash flood was not able to find any rain gage data in the flooded watersheds. To demonstrate the effective use of rain gages to determine the occurrence of tropical rainfall rates, radar estimates of storm total rainfall from the Johnstown, PA (1977) flash flood are compared with rain gage readings.