Few research issues are as significant as trying to work out how climate change might affect the environment. Unlike many scientific experiments, this isn't something that we can try in a laboratory several times to see how things pan out under different conditions. Nor do we have the luxury of being able to wait to see which of the many theories on offer is the most scientifically complete. This is why computer modeling is so important to understanding climate change.
Computer models offer the only way of anticipating how mankind's massive experiment with the planet's climate might play out
Computer models of the climate have a major bearing on public policy debates and directions. For example, much of the discussion within the Intergovernmental Panel on Climate Change (IPCC), which produced its Fourth Assessment Report late last year, hinges on analysis and interpretation of models.
Naturally enough, given the importance of climate change, in particular, and environmental issues in general, the world's research labs are awash with data. And that data comes from many places. For example, environmental data is collected from space, using satellites, as well as from terrestrial sources including sensors measuring weather, the oceans and terrain.
These data fuel the growing discipline of environmental informatics which combines research in fields of artificial intelligence, geographical information systems (GIS) modeling and simulation and user interfaces. The challenge is to find the best ways to combine data and to feed it into the various models of the climate. This is where the Environmental Scenario Search Engine (ESSE) comes into play.
ESSE is a test bed in more ways than one. Not only does it provide a tool for environmental research, it is also a real world use of ‘fuzzy logic' in databases. In this way, it gives researchers at Microsoft Research's laboratories in Cambridge, UK, an eye into the potential use of fuzzy logic in searching conventional databases, for example.
Fuzzy logic offers a way of translating search questions from human language, with all its uncertainties, into something that computer systems can handle. Dr Vassily Lyutsarev, Manager of Scientific Computing at Microsoft Research Cambridge, UK, explains that environmental data is held in different forms in databases all over the place. “It is hard to analyse this data using conventional database techniques,”says Dr Lyutsarev. Fuzzy logic, he adds, is a good way to work with large databases.
“What makes it so different from conventional text-based search engines is that it actually searches inside the numeric datasets. With ESSE, scientists will be able to find specific parameter values, conditions and scenarios among the huge amount of available environmental data. ESSE will help you find useful data even if you don't know exactly what you are looking for.”
ESSE is part of Microsoft's European Science Initiative. “The idea of the initiative,” says Dr Lyutsarev, “is to bring together computer scientists and researchers from the natural sciences, in particular, biology and environmental sciences.” The objective, he adds, “is to build new kinds of tools that can revolutionise how science is being done.”
As well as Microsoft Research Cambridge, UK, the ESSE team also includes scientists at the Geophysical Center of the Russian Academy of Sciences in Moscow, Moscow State University, the National Geophysical Data Center in Boulder, Colorado – part of NOAA, the US National Oceanic & Atmospheric Administration and the world's largest provider of publicly available geophysical data. The researchers describe ESSE as “a set of algorithms and software tools for distributed querying and mining large environmental data archives.” As they put it, “ESSE acts as a bridge between questions a user needs to ask of the environment and the data which describes it.”
Vassily Lyutsarev, Manager of Scientific Computing at Microsoft Research Cambridge, UK
Dr Lyutsarev explains that the team in Moscow, led by Dr Mikhail Zhizhin, in the Geophysical Center of the Russian Academy of Sciences, is developing software and building databases. The group at NOAA, led by Dr Eric Kihn in the National Geophysical Data Center, provides expertise on environmental data.
Dr Lyutsarev describes ESSE as ‘work in progress'. It has, though, already started to deliver results. “We are moving from developing computer tools to doing some real environmental research,” he explains.
A new Microsoft backed project that is about to get under way brings together the ESSE team, the Geophysical Center and the Space Research Institute of the Russian Academy of Sciences in a new project called Climate Induced Vegetation Change Analysis Tool. “The idea is to use tools developed in ESSE to analyse climate data, together with satellite images to find dependencies between climate and vegetation, particularly in Russia,” says Dr Lyutsarev.
ESSE's tools also underpin a new NOAA project as a part of the Comprehensive Large Array-data Stewardship System (CLASS) system. NOAA describes CLASS as a “web-based data archive and distribution system for NOAA's environmental data. The system is NOAA's main way of distributing environmental satellite data.” As Dr Lyutsarev puts it, CLASS is about providing remote access to vast amounts of data, which is where ESSE comes in.
Another project with backing from Microsoft addresses the issue of what you do with environmental data and how to run computer models to see how the climate might change under different conditions. The complexity of climate modeling requires considerable computer power. ClimatePrediction.net, which describes itself as “the largest experiment to try and produce a forecast of the climate in the 21st century,” makes it possible for anyone with a computer to contribute to the number-crunching.
Based in the Atmospheric, Oceanic and Planetary Physics group at Oxford University, climateprediction.net, called CPDN for short, addresses one of the challenges identified by the IPCC as a high priority: the need to “improve methods to quantify uncertainties of climate projections and scenarios, including long-term ensemble simulations using complex models.” As well as support from Microsoft, the main funding for CPDN comes from the Natural Environment Research Council and the UK Government's ‘e-science' programme.
Modelling needs plenty of computer power so that researchers can run their models many times, making small changes to see how they affect the outcome. Dr Suzanne Rosier, coordinator of CPDN, explains that thanks to distributed computing, climate modellers have access to “more computer power than they could get from sending their work to the world's biggest computers.”
The Oxford group works with models from the Met Office in the UK, one of the world's leading climate monitoring and modelling organisations. The data crunching then goes out to a growing network of nearly 50,000 ‘hosts': people all over the world who give CPDN some of their computer time to run climate models. Experiments run in the background on the host PCs.
“It all relies on volunteers,” says Dr Rosier. “We package the models up so that people can download them over the Web,” she explains. “All the data comes back to Oxford and gets handled here.”
The project, now in its fifth year, recently achieved a landmark of 30 million ‘model years'. Like ESSE, CPDN is moving on to a new level. After running global models of the climate, the group is about to start regional modelling. For example, it plans to release a model of one region in the very near future.
Another objective for CPDN is to feed into the scientific work of the IPCC. Like ESSE, now that the Oxford project has proved that its approach works, it hopes to become a part of the wider global modelling community and to contribute to the IPPC's next Assessment Report.
With so many years under their belts, and several published papers, both projects are now in a position to help to achieve the IPCC's goal of better models of the climate. By signing up for climateprediction.net, you can also do your bit for research into climate change.