Dr James Gibbons is lecturer in ecological modelling in the school of the environment, natural resources and geography at Bangor University.
As the statistician George Box said: “All models are wrong, some are useful.”
All of us rely on models but understand their lack of perfection – we know weather forecasts will not be always right but use them as a guide.
The Bank of England forecasts inflation using economic models – the forecasts are nearly always wrong but government finds them useful.
The design of medicines, cars, aircraft and bridges is helped by computermodels, which makes products cheaper to and safer.
The main reason models are not perfect is that they are simplifications of the system being modelled.
This simplification is necessary because scientists rarely have a complete understanding of the world. More importantly, a useful model keeps the essential while discarding the inessential.
If the model is too complicated it is difficult to use, if it is too simple it may mislead. Or as Einstein put it: a model should be “as simple as possible but no simpler”.
This is hard. In many ways a good model is like a good portrait. It is not an exact copy of the subject but is recognisable and contains the essential features.
A poor model, like a poor picture, distorts and hides features and so misleads.
My research uses modelling to explain and predict how change affects our natural environment. This is important because nature feeds us but we also gain other benefits – clean air and water, amenity value and even health benefits.
It has been estimated the environment contributes £8.8bn (9% of GDP) to the Welsh economy every year.
How should we manage our environment to maintain food production while keeping or enhancing the other benefits? And what will happen to all these benefits if the climate changes or the European Union changes agriculturalpolicy?
To answer these questions requires connecting policy, climate, crops, livestock and the decisions made by farmers.
What happens is also determined by and affects the numbers of plant, animal and microbe species.
Surveys have found that the number of farmland birds and some plant species have declined over the last few decades. Can we reverse these trends while maintaining food production?
Modelling allows us to investigate these complex questions about the environment. This cannot happen in isolation – I work with scientists collecting data about all these aspects. Their knowledge and insight is essential to ensure the models are sensible and discard the inessential.
There is still a lot we don’t know and modelling cannot substitute for this knowledge. By including lack of knowledge and uncertainty as part of my models I am able to show the range of what might happen without misleading.
The results are a rough sketch of the truth rather than an exquisite fantasy. This process is useful because it also helps find parts of the environment when more knowledge would substantially improve our ability to make predictions.
The goal of this research is to be able design the countryside so it enhances all the benefits we receive from the environment.
To achieve this goal I’m happy for my models to be wrong but hope they are useful.
To contact James please email email@example.com.
This article first appeared in the Western Mail‘s Health Wales supplement on the 19th September 2011, as part of the Welsh Crucible series of research profiles.