Statistical Analysis and Research Conducted by the Centre for Applied Statistics (CAS), UWA 2019

This project represents collaborations between CAS and other parties, generally involving the development of improved statistical techniques for the analysis of large and complex data. The primary project that utilised the allocation in 2019 was: Fully Bayesian Spatio-Temporal Modelling of Daily Precipitation.
Person

Principal investigator

Matthew Tuson matthew.tuson@uwa.edu.au
Magnifying glass

Area of science

Mathematics, Statistics
CPU

Systems used

Magnus and Zeus
Computer

Applications used

R, C++
Partner Institution: The University of Western Australia| Project Code: pawsey0133

The Challenge

Model site-specific daily rainfall measurements at many locations across Australia.

The Solution

Site-specific daily rainfall measurements at many locations across Australia were modelled, with the models focusing on capturing all features of daily rainfall measurements: temporal variation, missing rainfall measurements, zero-rainfall days, and extreme tail behaviour. The project code was mainly written in C++, with some pre-processing in R.

The Outcome

The project involved computing several variants of a model across the different locations. Use of the Pawsey facilities was required in order to estimate these models, due to a high computational burden arising from their complexity.

List of Publications

Climate inference on daily rainfall across the Australian continent, 1876–2015
M Bertolacci, E Cripps, O Rosen, JW Lau, S Cripps (2019)
The Annals of Applied Statistics 13 (2), 683-712