 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
General
Fundamentally, I am interested in the application of statistic methods to environmental, scientific and industrial problems. I also have a keen interest in various aspects of computing.
My statistical interests thus far include:
|
|
Spectral Analysis |
|
|
Spatial Statistics |
|
|
Extreme Value Modelling |
|
|
Non-parametric Methods |
|
|
Random Effects Modelling |
|
|
Spatio-temporal Modelling |
If you wish to discuss any of my interests then please feel free to contact me at carl.scarrott@canterbury.ac.nz.
I studied for my PhD at the Mathematics and Statistics Department at Lancaster University, on an EPSRC studentship with some additional support from Magnox Electric Plc part of BNFL. The focus of my Ph.D. research is the application of novel statistical methods to nuclear reactor temperature modelling and risk assessment. The fundamental issue is that of inference from a sample of measurements which are used for reactor control, and are therefore limited in the upper tail. Therefore, the statistics of the sample are not representative of the population, particularly in the upper tail of most concern for reactor safety.
I developed spatial multi-taper spectral analysis methods to identify the regular effects of reactor geometry and fuel age on temperature. These effects were built into a spatial regression model, incorporating random effects to account for the non-deterministic spatial covariance. This model takes account of control action on the spatial covariance, so can be used to predict the unmeasured temperatures from the biased subset of measurements. Extreme value methods were used to quantify the residual uncertainty for the risk assessment. Bootstrap techniques were used to quantify uncertainty in the risk predictions.
See my Publications page for papers/presentations on this work.
I have experience of using the following programming languages:
C
SQL*PLUS
running on UNIX, LINUX and Windows operating systems. I wrote this guide to creating movies in MATLAB. I also have extensive experience of using LaTeX, and regularly use various office, web, and graphics packages.