| |
CRC PUBLICATIONS |
Application of Hidden State Markov Model to Australian Annual Rainfall Data
Ratnasingham Srikanthan
Mark Thyer
George Kuczera
Tom McMahon
Publication Type:
Working Document
This is a publication of the current CRC for Catchment Hydrology
CRC Program:
Climate Variability
|
Publication Keywords:
|
Abstract / Summary:
In the past, the stochastic generation of annual data was performed generally with a first order autogressive model which does not explicitly model the observed long periods of wet and dry periods in the annual data. Though geographers and geomorphologists have observed long cycles or changes in the mean level of rainfall and streamflow, it was not explicitly included in annual stochastic data models until the recent work of Thyer and Kuczera (1999, 2000). Using 180 years of flood stage records at Windsor and 90 years of discharge data at Penrith, Warner (1987) defined alternating flood dominated and drought dominated regimes for the Hawkesbury-Nepean system. The drought dominated regime periods are 1821 - 1863 and 1901 - 1948 and those of the flood dominated regimes are 1799 - 1820, 1864 - 1900 and 1948 onwards. In order to model the long periods of wet and dry spells explicitly in the annual rainfall data, Thyer and Kuzcera used a first order Markov chain. The model is referred to as the hidden state
Markov (HSM) model. They showed that Sydney and Brisbane data exhibited two-state persistence structure. However, data from Melbourne, Adelaide and Perth did not show the existence of this persistence. The purpose of this study is to apply the HSM model to annual rainfall data from a number of rainfall sites across Australia and identify the the sites where a two-state persistence structure was likely to exist.
This working document is available for downloading and is available for purchase from the Centre Office.
 |
1 Download(s) |
|
|
|
|
|
|
|
|
Centre Office:
CRC for Catchment Hydrology
Dept of Civil Engineering
Building 60
Monash University Vic 3800
Tel: +61 3 9905 2704
Fax: +61 3 9905 5033
|
|
|
