Collaborative study of GENIEfy Earth System Models using scripted database workflows in a Grid-enabled PSE
Andrew Price, Z. Jiao, Ivan I. Voutchkov, Tim Lenton, Gethin Williams, Daniel John Lunt, Robert Marsh, Paul J. Valdes, Simón Cox · ePrints Soton (University of Southampton) · 2006
The integration of compute and data grids into a common problem solving environment enables collaboration between members of the GENIEfy project. In addition, state-of-the-art optimisation algorithms complement the component framework to provide a comprehensive toolset for Earth system modelling. In this paper, we present for the first time, the application of the non-dominated sorting genetic algorithm (NSGA-II) to perform a multi-objective tuning of a 3D atmosphere model. We then demonstrate how scripted database workflows enable the collective pooling of available resource at distributed client sites to collaboratively progress ensembles of simulations through to completion on the computational grid. A systematic study of the oceanic thermohaline circulation in a hierarchy of 3D atmosphere-ocean-sea-ice models is presented providing evidence for bi-stability in the Atlantic Meridional Overturning Circulation.