Cumulative storm rainfall distributions: comparison of Huff curves

Watershed models of hydrology and water quality often require a long record of precipitation measured in short time increments. Huff curves, a probabilistic representation of storm intensities using isopleths of probability, can provide this information stochastically or as design storms. However, m...

Full description

Saved in:
Bibliographic Details
Published in:Journal of hydrology, New Zealand New Zealand, 2003-01, Vol.42 (1), p.65-74
Main Authors: Bonta, J. V., Shahalam, A.
Format: Article
Language:English
Subjects:
Datasets
Earth sciences
Earth, ocean, space
Exact sciences and technology
Hydrology
Hydrology. Hydrogeology
New Zealand
Nomographs
Precipitation
Rain
Rain gauges
Sample size
Standard deviation
Storms
Watershed hydrology
Watersheds
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Watershed models of hydrology and water quality often require a long record of precipitation measured in short time increments. Huff curves, a probabilistic representation of storm intensities using isopleths of probability, can provide this information stochastically or as design storms. However, many factors affect the development of Huff curves, and an objective method for comparing them is needed. This study investigated two methods of comparing Huff curves, a "curve approach" in which the isopleths of probability of Huff curves were compared using a measure of disparity between isopleths, and a statistical test using the Kolmogorov-Smirnov test ("K-S approach"). Fifteen years of precipitation data for May and June from Invercargill, New Zealand were used. The minimum number of storms needed to develop Huff curves that were stable (i.e. a set of curves that does not change with added storms) was used to illustrate the two methods. Both methods used three regions within Huff curves to detect the location of differences. The curve approach showed that a sample size of 110-140 storms was sufficient to achieve stability. The K-S approach showed that a sample size of about 145 storms was adequate, but lowering the significance probability to 5% from 10% revealed that a sample size of 110 to 120 storms was sufficient. The data suggest that a minimum storm sample size of 120 storms was sufficient to develop a stable set of Huff curves based on the results of both methods. The results suggest that either method can be used to identify the minimum number of storms. The two methods can be used to evaluate other factors important in Huffcurve construction for quality control.
ISSN:0022-1708
2463-3933