Laserfiche WebLink
<br />OJJ483 <br /> <br />3 <br /> <br />analyzed. Radar patterns for previous storms should be reviewed along with <br />point rainfall readings if they are available. The spacial variability of <br />storm precipitation over the watershed should be assessed with consideration <br />to potential orographic effects, <br /> <br />Simulation Models <br />Quantitative forecasts of a flood event are provided by using simulation <br />models. Two types of models are required to provide a timely flood warning: <br />1) Quantitative Precipitation Forecast (QPF) models, and 2) Quantitative <br />Runoff Forecast (QRF) models. Where the watershed response is slow the QPF <br />may be reduced to just interpreting measured rainfall, but where watershed <br />response is rapid a more sophisticated precipitation forecast is necessary. <br />This means that for all but very large rivers with response times in excess of <br />one day, that a QPF wi 11 be required in order to provide a timely flood <br />warning. <br />Selection of simulation models should be based largely on information <br />gathered on hydrologic and meteorologic processes. The models selected should <br />accurately reflect the dominant physical processes that create flood events. <br />Models should not be selected on the basis that they have performed well in a <br />flood-warning system in some other region of the country. For example, a <br />model that provides an accurate QPF from weather radar in the Mississippi <br />valley would not be adequate for Manitou Springs unless it was enhanced to <br />addresses all of the dominant meteorological processes for the Manitou Springs <br />region, Likewise for runoff models, there is by no means a general hydrologic <br />response model avai lable for all conditions and runoff routines in a model <br />should be reviewed carefully. <br />Performance of the simulation models should also be evaluated. Models <br />that require a significant amount of computer time to execute may not be able <br />to update a flood forecast rapidly enough. The model should use computer <br />resources efficiently and with a minimum of routines for processing infor- <br />mation or presenting information to an end user. These routines are common in <br />hydrologic models used for analysis but are not required in a real-time <br />system. Real-time simulation models communicate only with the computer and do <br />not need to be anything more than fast number crunchers. <br />The structure of the ideal simulation model for real-time flood fore- <br />casting would be a set of modular routines modeling each of the dominant phy- <br />