Ocean microbes in the petascale age: towards the billion-particle model

点击翻译按钮获取中文摘要

"The world's oceans are responsible for about 50% of the global primary productivity and therefore play a key role in the global carbon balance and bio-geochemical cycles. The oceans are also a major source of food for human populations worldwide. Microbes form the basis for both these ecosystem functions, and predicting the complex interactions between marine microbes as well as the outcomes of these interactions remains one of the grand challenges in climate and environmental research [1, 2]. Because the ocean is in constant motion, microbial life and the marine ecosystem involve many inherently Lagrangian processes that can be modelled using individual-based models (IBMs). These models, in contrast to traditional population-based methods used in ecological modelling, can resolve fine-scale processes and complex nonlinearities underlying Lagrangian movement-driven contact between microbes using large numbers of virtual particles. The key biological challenge targeted by this project will be to incorporate intra-population and life-cycle variability, rapid acclimation/adaptation and microbe-microbe competition/invasion dynamics, into the Lagrangian movement IBM model to predict how climate-forced changes in climatic temperature and oceanic currents may alter ecosystem dynamics. One promising additional avenue would be to model interactions of marine microbial species with species at higher trophic levels (e.g., zooplankton, fish) to predict effects of incomplete mixing conditions and strong spatial variability/patchiness on population dynamics in food chains.The key target for quantitative/computational innovation in this project will be to tackle the problem that current Lagrangian models are severely limited by the high computational demand arising from the large numbers of particles required, which in turn limits predictive capabilities. The student will tackle this problem by building upon the OceanPARCELS (Parcels) software designed at Imperial College London combined with real data on microbial physiology from the BioTraits database also developed at ICL. The OceanPARCELS framework allows rapid creation of complex movement-driven ecosystem models by programmatically defining the behaviour of individual agents in a simple high-level language, while utilising the power of high performance computing (HPC) resources through modern computational techniques, such as the Just-in-Time (JIT) compilation. The result will be development of a model with the vast numbers of particles required for a realistic oceanic microbial IBM --- billions of particles, relevant to modelling real microbe-microbe interactions in the open ocean. The parameterisation of realistic ecological behaviour and variation of the model microbes will be achieved through the BioTraits data, which covers hundreds of species and thousands of experimental measurements of microbial physiology and other traits relevant to microbe-microbe life history and interaction outcomes. References: [1] Allen et al. Marine ecosystem models for earth systems applications:The MarQUEST experience. Journal of Marine Systems, 2009[2] Doney. Major challenges confronting marine biogeochemical modeling. Global Biogeochemical Cycles, 1999"