Dr Rachel Oughton

Associate Professor in the Department of Mathematical Sciences

What do you do?

I'm an Associate Professor in Statistics, and have been in Durham since 2004 when I started as an undergraduate. I have two daughters and have worked part time since coming back after the first was born. I love being a statistician because you have to learn about all sorts of other fields in order to be able to build models and make inferences.

How are you involved in this area of science? 

My PhD was focussed on using emulators to compare multiple simulators of the same system. This is an important problem because there are usually several different simulators of something like the weather, or in my case ocean ecosystem models. They might have different input parameters, different underlying equations and different outputs, but in some sense they should be related. There is not an obvious way to learn about their differences and similarities, but in my PhD I developed several ways to use emulators to do that.

Since then I have developed the work from my PhD, and have applied emulation to rainfall runoff and flood inundation models with a particular focus on flooding and flood warning systems in Indonesia.

What do you love about this topic?

The way the emulator is an attempt to honestly capture our understanding. It has always bothered me that models in science are often presented as absolutely true, when in fact they are usually approximations (albeit sometimes very good ones). An emulator can capture all this scientific knowledge and also the uncertainty that comes from various sources, ultimately resulting in sounder inference and better decision making. It is a very general methodology that has proved useful in numerous fields.

How does this work deliver real-world impact?

The work has delivered real world impact in a number of areas, both scientific and industrial. Simulators are ubiquitous, but come with many obstacles, such as the high dimensionality of the input space, slow running speed and the inherent uncertainty in the output that comes from the model being necessarily an approximation to the truth. Emulators have been used to answer many real world questions that the simulator alone can't.