HR: 09:45h
AN: A31A-06 [Abstracts]
TI: A Quantitative Approach to Flash Flood Prediction in Southern Utah
AU: * Hurwitz, M M
EM: mmh32@cam.ac.uk
AF: National Weather Service, Salt Lake City Weather Forecast Office, 2242 W. North Temple, Salt Lake City, UT 84116 United States
AU: * Hurwitz, M M
EM: mmh32@cam.ac.uk
AF: Centre for Atmospheric Science, Department of Chemistry, Cambridge University, Lensfield Rd.,
Cambridge, CB2 1EW United Kingdom
AU: Gibson, C V
EM: chris.gibson@noaa.gov
AF: National Weather Service, Salt Lake City Weather Forecast Office, 2242 W. North Temple, Salt Lake City, UT 84116 United States
AU: Jackson, M
EM: mark.jackson@noaa.gov
AF: National Weather Service, Salt Lake City Weather Forecast Office, 2242 W. North Temple, Salt Lake City, UT 84116 United States
AU: McInerney, B
EM: brian.mcinerney@noaa.gov
AF: National Weather Service, Salt Lake City Weather Forecast Office, 2242 W. North Temple, Salt Lake City, UT 84116 United States
AB:
Flash flood monitoring and prediction is considered to be a critical part of National Weather Service (NWS) severe weather
operations in the semi-arid western United States. The complex terrain and steep slopes in this area, combined with
impervious rock and soils, can induce flash flooding with relatively light rainfall. This reduces the value of using the
more common conceptual flash flood models developed for the central and eastern United States. Thus, forecasters at the NWS
Weather Forecast Office in Salt Lake City, Utah, have relied on a locally developed conceptual model to predict the
likelihood of flash flooding on a given day. Until this study, common practice was to assume that humid and unstable air
combined with low wind speeds in the lower troposphere would yield rainfall conductive to flash flooding. A new approach to
flash flood prediction, exploring the connection between atmospheric variables and flash flood reports, will increase
situational awareness and provide forecasters with quantitative flash flood guidance.
A record of historical flash floods in southern Utah was compiled to determine the frequency of events from 1959 to 2003. A
complete data set, consisting of both historical flash flooding days and non-event days, was assembled. A trial of the 2003
three-month flash flood season assessed which variables and which dataset to use in studying the eight flash flood seasons
from 1996 to 2003; the trial concluded that the best source of atmospheric data was a set of soundings from Flagstaff,
Arizona, a location close to and generally upstream of southern Utah. Neural networks were used to determine the
relationship between the atmospheric state and a particular day's flash flood severity. The final neural network used six
input variables and a discretized output variable. Precipitable water, low-level relative humidity, convective available
potential energy, the 500hPa height change between 12Z and 0Z the following day, and the previous day's flash flood severity
were found to be the important determinants of flash flooding in southern Utah. Data collected throughout the 2004 flash
flood season was used to verify the accuracy of the above-mentioned flash flood prediction algorithm.
DE: 1821 Floods
DE: 3322 Land/atmosphere interactions
DE: 9350 North America
SC: Atmospheric Sciences [A]
MN: 2005 Joint Assembly