Environmental problems are increasingly growing and we need to take into account human activity and preservation of the environment (see for instance the European Common Agricultural Policy in 2005, the French Grenelle of Environment in 2007) leading to the concept of sustainable development (conference of Rio, 1992).
International policies delimitated new goals for these ecosystems due to ecological services in particular, for the two major changes for the following decades which are the availability of fresh unpolluted water and the regulation of carbon emission (through carbon storage, finding energy leading to low carbon emission). In order to achieve these goals, we tend now to create new natural systems as surrogates of the degraded natural systems to provide these ecological services (conf. Rio, 1992). Recent works have, for example, demonstrated the use of natural prairies to provide alternative biofuels (carbon positive biofuels) (Tilman et al., 2006), or their role in carbon storage (Ni, 2002; Purakayastha et al., 2008).
Ecological problems have the specificity of being dependent on biological elements with complex interactions and which response may be delayed at a year or pluri-year scales due to biological cycles. New tools are therefore needed to go past biological constraints and take into account the complexity of living ecosystems. The ViP Project aims at using modeling for
- Providing extensive virtual experiments for testing solutions in a shorter time that would have been possible through real experimentation,
- Optimizing experimental designs to access to the most efficient results.
Our ViP Modeling uses a combination of Agent Base Model and Partial Differential Equations to simulate the ecosystem of a prairie in constant interaction with its environment. ViP is a highly inter-disciplinary project that requests numerous competences in Ecology indeed, but also Applied Mathematics and Computer Science (of course!). ViP starts from the study of the ecosystem of clonal plants led by the team of Pr. Cendrine Mony and her collaborators from UMR Ecobio 6553 at University of Rennes-1 and involves extensive semi controlled experiments and a posteriori statistical analysis. Data acquisition is facilitated by image analysis and requires new functional analysis techniques (Pr. El Hamidi and Pr. Michel Menard from University of La Rochelle), and deep mathematical understanding of the mathematic of stochastic agent based models (Dr. Fabien Campillo and his collaborators from Inria).
The large scale computational side of ViP is led by our CS@UH team and relies heavily on the development of new algorithms and the use of volunteer computing environment (BOINC) developed by Pr. David Andersen from U.C. Berkeley Space Sciences Laboratory, who is also affiliated to our department.
A prairie is a fairly complex system with emerging properties that need to be explained, and one needs the development of a new generation of multi-objective genetic algorithm that can run on the volunteer computing environment (PhD of Malek Smaoui) or parallel data mining methods that can run on the fly (PhD Thesis of Waree Rinsurongkawong) while BOINC generates millions of data set.
This project has been well funded by grants by the three following major national research agency from France: CNRS, Cemagref and ANR, that supports our international collaborations. ViP will lead hopefully to amazing discoveries and improved engineering of prairies. If you are willing to donate some of your desktop time you may want to learn more by connecting to the website of the project http://vcsc.cs.uh.edu/virtual-prairie/
Although, our journal log, that is posted on the project’s website, gives regular update to the general public on our progress...