Microsoft Research Tackles Ecosystem Modeling

| Josh Henretig

clip_image002What if there was a giant computer model that could dramatically enhance our understanding of the environment and lead to policy decisions that better support conservation and biodiversity? A team of researchers at Microsoft Research are building just such a model that one day may eventually do just that, and have published an article today in Nature (paid access) arguing for other scientists to get on board and try doing the same.

When Drew Purves, head of Microsoft’s Computational Ecology and Environmental Science Group (CEES) and his colleagues at Microsoft Research in Cambridge, United Kingdom, began working with the United Nations Environment Programme World Conservation Monitoring Center (UNEP-WCMC), they didn’t know they would end up modeling life at global scales. “UNEP-WCMC is an international hub of important conservation activity, and we were pretty open-minded about exactly what we might do together,” says Purves. But they quickly realized that what was really needed was a general ecosystem model (GEM) – something that hasn’t been possible to date because of the vast scale involved. In turn, findings from a GEM could contribute to better informed policy decisions about biodiversity.

But first, a primer on terminology. A GCM (general circulation model) is a mathematical model that mimics the physics and chemistry of the planet’s land, ocean and atmosphere. While scientists use these models to better understand how the earth’s climate systems work, they are also used to make predictions about climate change and inform public policy. Because these models have been so successful, members of the conservation community are looking for a model that could improve their understanding of biodiversity.

Building a GEM is challenging—but not impossible. Microsoft Research and the UNEP-WCMC have spent the past two years developing a prototype GEM for terrestrial and marine ecosystems. The prototype is dubbed the Madingley Model, and is built on top of another hugely ambitious project that the group just finished, modeling the global carbon cycle. With this as starting point, they set out to model all animal life too: herbivores, omnivores, and carnivores, of all sizes, on land in the sea. The Computational Ecology group were in a unique position to do this, because the group includes actual ecologists (like Purves), doing novel research within Microsoft Research itself. In addition, they’re developing novel software tools for doing this kind of science. That has helped the team as it’s come up against all kinds of computational and technical challenges. Nonetheless, the model’s outputs have been widely consistent with current understandings of ecosystems.

One challenge is that while some of the data needed to create an effective GEM has already been collected and is stored away in research institutions, more data is needed. A new major data-gathering program would be expensive, so supporters of GEMs are calling on governments around the world to support programs that manage large-scale collection of ecological and climate data.

But if you build it, will they come? Drew Purves knows building a realistic GEM is possible, but he believes the real challenge is constructing a model that will enable policy makers to manage our natural resources better – and that means making sure the predictions are accurate. If such an accurate, trustworthy model can be achieved, one day conservationists will be able to couple data from GEMs and models from other fields to provide a more comprehensive guide to global conservation policy.

Finding solutions to climate change and ecosystem preservation is too big of a challenge for any one entity to tackle in isolation. And that’s exactly why we think that computer modeling has potential. It’s another great example of the continually evolving role that technology will play in addressing the environmental challenges facing the planet—and we’re honored to be working hand in hand with the United Nations Environment Programme to begin solving those challenges.

Tags: , , , , , , , ,