Understanding the Complexities of Morecambe Bay
As with many people in the area I am both excited and inspired by the prospect of an Eden Project North in Morecambe. I think back to the trip of the Ensemble team to the Eden Project in Cornwall in late November to discuss what it would mean to have a Morecambe Bay Virtual Observatory and how the work of Ensemble can support this.
I remember in particular a bold phrase that was used by the Eden team – “to make Morecambe Bay one of the most understood areas in the world“. But what does this mean, and what tools do we have at our disposal, particularly in terms of the role of digital technology supporting such a laudable aim? A lot of it comes down to monitoring and modelling and the good news is that there are real innovations occurring in each of these areas.
In terms of monitoring, we are now in a position to collect unprecedented amounts of data at a variety of scales and about many of the facets that make up Morecambe Bay. We have massive growth in the amount of Earth observation data from satellites thanks to initiatives like Sentinel-2. In addition, breakthroughs in sensor technology, we can now instrument the natural environment to gain fine resolution data about the Bay (an Internet of Things for Morecambe Bay). To add to this, we also have significant growth in Citizen Science and also the ability to scrape data from the web using AI (Artificial Intelligence) techniques.
Monitoring provides us with lots of data, but what can we do with this data to build a genuine understanding of the Bay? This is where modelling comes in. Environmental Scientists have developed a wide range of environmental models to both capture their understanding of the physical processes involved and then use this to make projections or predictions of the future. This is particularly important in the context of a changing climate. These models are typically referred to as ‘process models’ because of their emphasis on representing the physical, chemical and/or biological processes involved – that is they represent the best understanding of the science. Thanks to developments in high performance computing infrastructure, these (often very complex) models can be run efficiently and often many times to understand the sensitivities and uncertainties of these models.
But, now there is a new kid on the block – data science and associated data models. Data Scientists advocate the use of techniques such as machine learning or deep learning to extract meaning from often complex data sets; they may also use sophisticated statistical models to understand the data associated with extreme events or study fundamental change in the underlying data. They do not attempt to represent the underlying science or scientific processes but rather extract patterns from the data. There have been some successes in this area, for example related to image processing, but their use in the environmental sciences is at a relatively early stage.
The nature of Nature
Where does this leave us in terms of understanding? Before answering this question, it is worth dwelling on, effectively, the ‘nature of Nature’. In our work in Ensemble, we have looked at the interactions between shifting sands, lunar cycles and tides, cockles and mussels, the rich bird life around the bay, and the actions taken by society in terms of, for example, visiting or fishing in the Bay. This has given us a deep understanding of the sheer complexity of natural systems, the complex inter-dependencies, the feedback loops, the chaotic behaviour and the unintended consequences. And this is before you even broaden out to consider the areas that feed into the Bay… the many river systems, the meeting of fresh water and salt water, the pollutants that run off the land… Nature is truly awesome and delicate at the same time. Can this complexity be captured by data? Can Bayes Theorem (one of the fundamental building blocks of data science) underpin an understanding of the complexity of our natural environment?
So is it possible therefore for us to truly understand such complex systems, using our tools of monitoring and modelling? At one level, the answer is clearly “no”, but this is not very helpful and science has a proud history of pushing on with what may appear like impossible dreams. One thing we can say is that our toolbox is expanding greatly and there has never been a better time to study an area like Morecambe Bay and strive to make Morecambe Bay one the most analysed areas on the planet. Through this, we can also inform decision makers to preserve the wonders of our natural environment, enhance its biodiversity and seek a better balance between our actions and intricate balances of nature. Crucial to this, is a need to work together in cross-disciplinary teams, to not see environmental models and data models as being in competition but to seek new ways in which they can work together. Can data models be used to select and adapt process models to better represent observations? Can process models support the understanding of patterns or correlations identified in the data?
Author: Gordon Blair
Photo credit: Morecambe Bay from Grange-over-Sands by Philip Platt