Integrating the Areal Mean Basin Estimated Rainfall (AMBER) and the Advanced Weather Interactive Processing System (AWIPS) into warning operations at NWSFO Pittsburgh during widespread flooding in southeast Ohio on 26-28 June 1998

Flash floods cause more fatalities in the United States than any other storm-related event, including tornadoes. Forecasting flash flood situations, however, represents one of the greatest challenges for National Weather Service (NWS) forecasters. This is because killer floods result from a complex arrangement of meteorological and hydrologic conditions, occurring over a variety of scales. Thus, forecasting the large scale meteorological conditions favoring flash floods represents only part of a more complex problem. Forecasters must anticipate both the likelihood of an event, and the amount of precipitation that will fall over a particular watershed. In order to do this effectively, forecasters must anticipate how precursor synoptic and mesoscale conditions will influence storm development, and how storms will act within a particular watershed to produce enhanced potential for flash floods. During the warning decision process, forecasters need technology and flash flood detection software to synthesize all relevant real-time information, from the synoptic scale down to the storm scale.

This paper will examine the forecast and warning process at NWSFO Pittsburgh during widespread flash flooding across five southeast Ohio counties during a 54 hour period between 6PM 6/26/98 and 12AM 6/28/98. Some locations received between 7 and 9 inches of rain during this period, resulting in 6 deaths and great property loss. Six Flash Flood Watches, four River Flood Warnings, and 88 Flash Flood Warnings were issued by NWSFO Pittsburgh during the event. All of the River Flood Warnings and 88 percent of the Flash Flood Warnings verified. Flooding was not reported outside of the warned areas.

The forecast process will focus on determining the physical mechanisms favoring the development and regeneration of flash flood-producing storms. The warning process will focus on how new technology (AWIPS) and flash flood detection software (AMBER) were used operationally to anticipate and detect flash flood-producing rainfall distributions during the event.

AWIPS technology is a significant part of the Modernization and Associated Restructuring of the National Weather Service. AWIPS workstations allow forecasters to integrate satellite, radar, lightning, numerical model, and other data on a single platform. During the flooding, AWIPS displays were configured with 12 screens, all looping independently. The screens provided critical real-time radar, satellite, and lightning data on a single platform, and allowed forecasters to view the development and evolution of flood-producing storms.

AMBER uses detailed radar rainfall estimates computed from the Digital Hybrid Scan Reflectivity (DHR) product of the Weather Surveillance Radar 1988 Doppler (WSR-88D) to compute Average Basin Rainfall (ABR) for 3,000 watersheds every 5-6 minutes, down to an area of 15 km2 (3 km2 in urban areas). AMBER compares ABR with the Flash Flood Guidance (FFG) produced by River Forecast Centers (RFC) to determine the probability and intensity of flash flooding. Rainfall rates of ABR were available every 5-6 minutes during the flood event, and alerted forecasters to the likelihood of flash flooding, especially across small basins. By combining rainfall data with stream basin analysis, AMBER provided critical real-time guidance during the warning decision process. In addition, AMBER verified flood-producing ABR amounts over very large areas (one exceeding 200 square miles) after warnings were issued.