Spatial Modeling of Rainfall Intensity-Frequency-Duration Curves

• Activity Number: 164
• Type: Topic Contributed
• Date/Time: Monday, August 5, 2013 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Aloke Phatak*+ and Mark Palmer and Eric Lehmann and Kwok Wai Lau
• Companies: CSIRO Mathematics, Informatics and Statistics and CSIRO Mathematics, Informatics and Statistics and CSIRO Mathematics, Informatics and Statistics and CSIRO Mathematics, Informatics and Statistics
• Keywords: rainfall extremes ; IFD curves ; climate change ; regional climate model

Abstract

Rainfall intensity-frequency-duration (IFD) curves describe how rainfall intensity varies with duration over a range of return periods. They are an important element in the design of hydraulic structures such as dams, weirs, culverts, and other drainage systems. In this work, we use a Bayesian hierarchical model to model the spatial variability of IFD curves based on the generalized extreme value (GEV) distribution. The model combines both daily data and data from pluviographs, and to obtain more precise estimates of the GEV parameters, data from different durations is also combined. We construct a spatial model that can be extended to incorporate projections from regional climate models (RCM), and in so doing, we can estimate IFD curves under different climate change scenarios. Because of the Bayesian framework, we can also derive estimates of uncertainty in the IFD curves for current and future climates. We illustrate the methods in the region around Sydney, Australia.