Rainfall variability and design storms prediction using block peak method

Abstract

Design storm is essential for the water resources planning and management. It is derived using rainfall data that has great variability in nature. The structure of design storm basically shall contemplate the intensity-duration frequency (IDF) relation and the structure of temporal storm pattern. The primary research aim to explore and investigate the rainfall variability and the stucture of storm pattern. At first, inspection is conducted on the presentation issues of the data extraction, regionalization, spatial data generation and rainfall aggregation. The peak block method is used to improve and to confirm the structure of temporal pattern. The daily and 5-min increment rainfall data from Selangor including Federal Territory and Johor, Malaysia are utilized to perform the analysis. The spatial data for 13 duration patterns is derived by applying kriging method and used in frequency analysis. The L-moments parameter diagram is exercised to examine the best fit of data distribution and the probability weighted moments (PWM) is used in the method of analysis. The probable maximum precipitation (PMP) calculation for local used is improved and has remarkable stability to the presence of outliers. It has been identified that the short and long duration patterns are obviously a single storm and multiple storms respectivelly. For particular duration, the same storm paterns are indicated for different return frequencies and the IDF curve developed based on daily data is overvalue about 10% compared to those of increment data. The spatial data analysis is an alternative solution for the ungauged location within homogeneous region. It can also facilitate the IDF curve development in the location with lack of increment data by using daily data. The deviation of result values in storm intensity of the constructed IDF curve for West Coast of Peninsular Malaysia is caused of the differences in data presentation issues.