To mitigate the risk posed by extreme rainfall events, we require statistical models that allow us to extrapolate outside the range of our observed data into the tail of our distribution and to estimate the probability of record events we are yet to observe. For this type of statistical analysis, we typically use methods from extreme value theory. The challenge for modelling rainfall extremes using many of these extreme value methods is that there exists a gap between statistical theory and climate reality.
In this project, you will be applying methods from extreme value theory to model Australian rainfall extremes and adapting the existing methods where necessary so that they suitably reflect the underlying physical processes.This project will be particularly interested in estimating the probability of back-to-back events, such as in the Hawksbury River’s region where there have been 4 floods in less than 18 months. In the longer-term, methodology and estimates developed from this project can be used to better inform community recovery and to reduce insurance losses in the face of back-to-back events.
Eventually, depending on outcomes, the research could be published in a scientific journal. The project also would be an ideal lead-in to a Doctor of Philosophy, where the research problem could be more thoroughly explored with more sophisticated probabilistic and statistical tools.
This project is most suitable for students who have an interest in probability and statistics, with applications in climate science. The project can be adapted to a student’s probabilistic and statistical interests.
Desirable skills include:
- stochastic modelling
- statistical inference
- R programming
- data visualisation
Supervisor: Kate Saunders (kate.saunders@monash.edu)
Location: Monash