I attained a Masters in Physics from Warwick University and from there went to do a postgraduate research degree. I did an undergraduate research project during my the summer of my second year on Super Time Stepping as an alternative for implicit schemes.
In addition to my research, I am interested in astrophysics and solar physics. I also have an interest in scientific programming and highly parallel computing, including GPUs.
I am studying the use of Multi Model Ensembles (MMEs) for Upper Atmospheric Modelling. It is important to accurately characterise the upper atmosphere in general for communications and protection from the sun. I focus on the total neutral density in the thermosphere, which is the atmosphere between 90 km and 600 km. Many satellites orbit around this altitude, including the International Space Station and experience a drag force proportional to the neutral density. The density can vary wildly depending on things like solar activity and time of day, so an inaccurate prediction could leave satellites hundreds of kilometres off course. Therefore having an accurate neutral density makes Space Situational Awareness (SSA) more accurate, which is knowing where all satellites are.
SSA is important because if a satellites position is inaccurate, collisions can occur. A satellite collision can have a closing speed of several thousand kilometres per hour, which destroys the satellites and leaves a cloud of debris in it’s wake. This debris can go on to cause other collisions creating more debris. Such a snowballing event is called the Kessler syndrome and in the worst case could leave Low Earth Orbit unusable. There have been many collisions but the first accidental hyper-velocity collision was in February 2009, when Cosmos 2251 and Iridium 33 collided.
There are many models for total neutral density in the thermosphere, but they all have an error. They assume different things, have different datasets behind them and have been made by different people. By combining the output of these different models together the result can be improved. This is similar to many weather forecasters combining their knowledge to achieve a more likely result. A simple way to achieve this is to average the results, and this can improve the result. Other methods can use training data to assess the models performance by removing bias and weighting in favour of better models. Much MME research has been carried out for climatology.