The Goal

Global climate models are useful tools to predict what the climate might look like decades into the future. However, as the figure above shows, projections of global land-surface temperatures vary between 21 current generation climate models. So which model's projections are the most "correct"? Of course we cannot …

The current drought around the Cape Town area got me to do a basic analysis of long-term (1977-2017) rainfall and temperature changes observed from the weather station at Cape Town International Airport. This is the analysis that I have been looking for but have not found. I mainly wanted to …

Prelude

A key concept in supervised statistical learning / machine learning is the bias-variance tradeoff (also known as the accuracy-precision tradeoff). This is an optimisation problem where these two properties compete against each other. Let us begin with the ordinary least squares regression approach, which only tries to reduce bias:

This …

2010-2014 seasonal temperature anomalies. At every time step, the temperature anomaly shown is relative to the normal temperature at a given grid-cell, for that particular time of year. More specifically, the anomaly is relative to the climatology for the month-long period 15 days ahead and 15 days behind the current …

Motivation

Three things prompted me to write this blog post. 1) A blog post by Damien Irving - a tech savvy PhD student from Melbourne, in which he lists his software "toolbox" for his climate research, 2) a recent task Nadja has been busy with consisting of restructuring the CMIP archives …

Considering I'll be working in the field of "extremes attribution" for at least the next few years, I thought it may be worth my while communicating some of what I do on a public platform. I'll attempt to communicate science using non-academic jargon and by keeping my posts entertaining (by …