WSP07528 - Laserfiche WebLink

Home Browse Search

OOJ.h2 Figure 4 shows the model predicted peak storm intensity plotted against the adjusted peak observed hourly rainfall. Storm intensities that lasted more or less than one hour were adjusted to correspond to the hour interval. Most model predictions (~65 percent) were within 1/2 inch of the observed intensity. This output was judged acceptable for use in hydrologic flash flood models. However, about 20 percent (6 events) were an inch or more heavier than forecast. In these cases the storm durations were less than 30 minutes and the prorated intensities were poorly correlated to the observed total rainfall (Figure 5). Actual predictions of storm total rainfall compared to observed totals are presented in Figure 5. For storms of less than 2 inch total rainfall, most predictions were within a half inch of the observed rainfall. For storms of 2 inches or more rainfall most model predictions were within .75 inches. In general storms less than 2 inches total rainfall were well forecast with a slight over prediction tendency. For storms over 2 inches of total rainfall a slight under-prediction tendency was noted. In general the model output was judged adequate for use in hydrologic models. assuming storm duration was specified, A continuing effort is in progress to develop predicted storm intensity and rainfall in 15-30 minute intervals with a high degree of accuracy so as to derive maximum benefit from the hydrologic basin models. The uniqueness of the Denver F2-P2 is that the primary input to decision-making by local officials comes in the form of a.meteoroloqical prediction not a remotely-sensed observation. While the use of remote sensing of rainfall to trigger decision making works well in slowly developing flooding situations, true flash-flood situations require the use of a prediction to assure adequate timely community response. The Denver F2-P2 experience emphasizes the importance of the man in the man-machine mix used to predict extreme convective events. This experience is contrary to the experience of other operational programs which emphasize a reliance on the remote sensing of rainfall and runoff coupled to computerized basin response models to trigger flash flood warnings. We believe this "measured-computed" warning type system is inadequate in flash flood situations compared to the F2-P2 approach. An Example - Operational Prediction of the Denver Severe Hail/Rainstorm of 13 June 1984 On June 13, 1984, two waves of severe thunderstorms merged over the northwestern suburbs of Denver, Colorado. into a massive rain/hailstorm that killed one person, injured 23 others and produced an estimated $350 million damage. If the damage estimates hold up this storm will be one of the single worst damage incidents in recent United States history. The development sequence, severe weather and rain production and location of the storm system were predicted 3-6 hours in advance of the storm's formation by Henz Kelly & Associates meteorologists in support of the UDFCD flash flood prediction service. The development of the June 13 Denver mega-rain and hail storm occurred in a classic heavy rain-large hail weather pattern. On the -9-