Angesh Anupam at COP22

Benefits of data driven modelling for research by Angesh Anupam

Grantham Scholar Angesh Anupam reflects on how data driven modelling can be used in all sort of research projects. 

Our first Journal Club

Journal Club is a wonderful way for Grantham Scholars to get an appreciation of interdisciplinary research. On 14 November 2014 the first Journal Club took place, hosted by the Department of Automatic Control and Systems Engineering. All the Grantham Scholars attended with enthusiasm.

The paper I selected was from the journal ‘Proceedings of the Royal Society A’ – A century of variation in the dependence of Greenland iceberg calving on ice sheet surface mass balance and regional climate change.

This paper is authored by (among others) my supervisor Professor Visakan Kadirkamanathan.

Greenland icebergs

As the title suggests, the paper deals with Greenland icebergs.

A long record of Greenland iceberg data became available through the International Ice Patrol’s record of icebergs (I48N). This record shows how much I48N varies from year-to-year. And the level of variation has increased significantly over recent decades.

This paper uses a technique to identify the mathematical model which should be applied to studies of I48N. Specifically, nonlinear system identification technique. And it combines models that describe both oceans and icebergs through coupled ocean-iceberg modelling.

What causes variation in Greenland iceberg melt?

Before this work, there was debate about what the causes of the variation in I48N were.

On many occasions, open ocean icebergs melting due to temperature change were perceived to be the biggest cause. This study shows that I48N variability is predominantly caused by fluctuation in discharge from the Greenland ice sheet (GrIS) breaking apart at the edges (or ‘calving’) rather than open ocean icebergs melting.

But that their contribution to I48N variability has still varied over the years.

Benefits of data driven modelling

As stated above, this work uses the concept of data driven modelling in the form of nonlinear system identification. And this approach can be applied to many areas sustainability research.

In fact, my own project, ‘Computational simulation of plant-soil-water processes for design of agricultural production’, is a good example. I will use this methodology to model the different processes.

My motivation in choosing this paper was to demonstrate this particular modelling approach can be useful in many research projects. It is especially useful where it is difficult to capture the internal dynamics of the system being studied merely through the involved science within the system.

Sometimes, there are many available models which work with a fine precision in the particular areas they are used to study. The modelling approach discussed in this paper can be used as a glue to link some of the these available models that fall within the remit of the research question.

A group of Grantham Scholars posing outside a university building
The Grantham Scholars of 2014, who took part in the first Journal Club. Angesh is third from the right.

Modelling is useful for all sorts of research

Before the discussion, most of the Grantham Scholars were perplexed about the choice of paper. But at the same time they were eager to grasp the bigger picture it conveys.

Monica Ortiz will be using modelling to understand how climate will affect crops. She realised that the paper was useful to give perspective on the powerful benefits of modelling, even on such large scales.

Hannah Sewell made a great contribution to the discussion. Hannah’s work will identify the genes and gene variants which are a part of a much larger network of genes that all interact. These interactions can be modelled mathematically giving greater insight into how these interactions produce the end result. In this case, how the many genes interact to give the stomatal patterning seen in crops in response to carbon dioxide levels. Modelling how much each gene and gene variation affects the density of this stomatal patterning could give researchers more control over it.

All other participants also agreed with the potential usefulness of modelling and especially the approach described in the paper. The overall discussion was fruitful and all the participants appreciated the importance of data driven modelling and the system identification technique in general.

We look forward to the next Journal Club meeting to explore another topic in the same fashion.

Journal Club is meet up of Grantham Scholars to discuss publications from a multidisciplinary perspective. It is part of the Grantham Scholar training programme.

The main image shows Angesh Anupam at COP22.

Edited by Claire Moran.