Establishment of rainfall depth-duration-frequency relationships at conventional raingauges & ungauged sites in Hong Kong

Abstract

Rainfall intensity (depth)-duration-frequency (IDF/DDF) relationships provide information essential for hydrosystem infrastructures designs. For small catchments, rainfall DDF relationships with short duration can be established based on rainfall records from automatic raingauges. Due to the progression of instrumentation technology development, wide spread installation of automatic raingauges does not happen until 2~3 decades ago. Therefore, record lengths at majority of automatic raingauges are relatively short and the derived rainfall DDF relationships on the basis of at-site frequency analysis are potentially subject to significant sampling error. On the other hand, many conventional raingauges exist long before automatic raingauges were deployed. However, daily rainfall data at conventional raingauges are of limited use to establish rainfall DDF relationships in areas with small catchments like Hong Kong where design storm duration significantly shorter than 24-hr are needed. In addition, the unavailability of rainfall data at ungauged sites also requires an approach to establish the rainfall DDF relationships at the location. This study presents a practical methodological framework to derive rainfall DDF relationships with short durations at conventional raingauge sites and ungauged locations. The core components of the framework for establishing rainfall DDF at conventional raingauges include a rainfall scaling model and a transformation relationship between annual maximum rolling-time 1440min rainfall and annual maximum daily rainfall. For ungauged sites, the framework for estimating rainfall DDF relationships at conventional stations is linked with Ordinary Kriging (OK) approach for interpolating the statistical moments of annual maximum daily rainfall to derive the design rainfalls. Two types of probabilistic point estimation methods are applied to assess the uncertainty of the estimated design rainfalls. Finally, various approaches establishing the Kriging model for interpolating design rainfalls are investigated and their relative performances are assessed through cross validation technique. The result shows that the proposed method is able to provide reasonable rainfall DDF relationships at conventional raingauge sites and ungauged locations in Hong Kong.

Date of Award	2013
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology