|
Abstract:
|
The goal of this study is to describe the rainfall characteristics of the Texas HIPLEX region as completely as possible. In order to classify precipitation events, storms were stratified by the quartile which contained the largest change in cumulative rain volume. Since stratification by this method did not explain the rainfall variance satisfactorily, a second classification, based on radar reflectivity patterns, was employed and is referred to as the convective category classification.
Initial analyses produced curves of intrastorm temporal rainfall distributions for the various time-quartiles. Examination of radar films revealed the importance of advection of the rainfall pattern and storm size relative to raingage network size in determining the shape of these curves. The quartile classification was therefore abandoned.
Characteristics of the convective categories (cells, small clusters, large clusters and nested clusters) were found to be distinctly different in mean values for storm duration and rainfall volume. A lesser difference v/as noted between intrastorm temporal rainfall distributions for the convective categories. Empirical cumulative probability distributions for storm duration, rainfall volume and time between storms were derived for each of the convective categories and showed marked dissimilarities. These empirical cumulative probability distributions were then combined to illustrate distributions for "all storms". Several theoretical distributions were tested to obtain a "best fit" with the empirical curves. In all cases, the Johnson S distribution produced the best fit based on minimizing the residual sum of the squares.
An independent data set was used to check the convective category scheme as well as to determine the representativeness of the derived distributions. Since no significant differences could be found in any feature, it can be concluded that the convective category classification is a meaningful way to stratify rainfall events in the area. |