CHARACTERIZING AND ESTIMATING SPATI AL AND TEMPORAL VARIABILITY OF TIMES BETWEEN STORMS

Abstract

An exploratory investigation was conducted on methods to characterize times between storms and estimate aparameter used to identify storms in a precipitation record for the purpose of simulating the occurrence of storms. Theparameter is called the critical duration (CD) and is fundamental to the operation of a stormgenerator model. It is definedas the minimum dry period between storms that separates bursts of rainfall into statistically independent storms. Data usedcovered an area of about 225,000 km 2 in the plains area of Colorado, Nebraska, Kansas, and Wyoming for the months fromMay through September. The exponential method that yields an exponential frequency distribution of times between storms(TBS) appears to give reasonable fits to the data. However, some precipitation gauge data yield CD estimates that appearto be outliers for individual months.<br><br>CD varies noticeably with month and location. An equation developed earlier for estimating CD over a wide range ofclimates is inadequate for estimating monthly CD values. Computing CD by collapsing TBS and precipitation data into 2,3, 4, and 5month periods did not improve the estimating capability of the equation. Separate regressions for monthly CDdata against average monthly precipitation were statistically insignificant for all months, and led to computing averages ofthe data for each month. A multiplicative power equation that was successful in another study to estimate monthly CD didnot work as well in the present study. Mapping CD over the study area appears to be the best method to estimate CD onungauged areas for parameterization and storm modeling. The study results are useful for guiding stormgeneration modeldevelopment and parameterization, and drought studies.