Predictive Models of Ecological Systems

The Computational Ecology and Environmental Science Group (CEES) develop, for ecological systems, the kind of useful predictive models we take for granted in other areas, such as physical engineering. To develop such models, we find that we need to invent a variety of new concepts, algorithms, and methods – which we then package into re-useable software tools. In this way, we can help Microsoft to substantially improve environmental management. [read more here]

Why Microsoft?

Individuals, governments and businesses are showing an unprecedented level of concern for environmental issues, and for good reason – humanity is facing some serious global environmental challenges, including climate change, food security, and the possibility of pandemics. Solving these challenges requires new kinds of science enabled by new kinds of scientific software. And the world’s largest software company is Microsoft. How could it not be involved in this area? [Microsoft and the environment]

Real ecologists, real ecology

Although we are part of Microsoft, all CEES members are active research ecologists with PhDs in Ecology, and a proven track record of publishing in top ecology journals. Our specialisms include vegetation, forests and climate change; conservation; biogeography; and disease dynamics. In addition to developing new scientific software, we write scientific papers, co-supervise PhD students, and lecture at universities. [list of projects] [list of scientific publications]

The science: predictive modelling of the biosphere

We lead, collaborate on or are involved with a wide variety of scientific projects involving different scientific questions, taxa, spatiotemporal scales and methodology. But we are currently working especially hard to assemble a set of predictive models, predicting the response of key aspects of the biosphere to human pressures. These models include CCF1.0, a model of the terrestrial carbon cycle; and the Madingley model, a new model of global biodiversity and ecosystem function. We hope that CCF1.0 will enable more accurate, transparent and defensible predictions of terrestrial carbon down to regional scales; whereas the Madingley model allows us to begin to understand how different biotic processes fit together to make the biosphere. [CCF1.0] [The Madingley Model] [list of CEES projects]

The software: new tools for new science

Since our inception we’ve experimented with many prototype software tools covering all parts of the scientific process from generating new environmental data; storing, finding and making sense of the data that do exist; defining new models; parameterizing and refining models into predictive tools; running model simulations at scale; visualizing and packaging science for stakeholders. We’ve now packaged up alpha version of some of these prototypes and made them available on our new tools site.