A New Top-Down Approach based on Engineering Protocol to Modeling Global Climate Change

Yuanshun Dai

Abstract

We are developing an innovative modeling technology to systematically model and efficiently analyze CO2 fluxes between terrestrial ecosystems and the atmosphere across different sites, regions, and continents in order to understand global climate change and to develop effective and realistic strategies for global management ofCO2 sources and sinks. This new modeling technology can be easily parallelized to well fit to utilizing Supercomputer, Grid systems or other HPC systems.

By viewing the climate system as a “big machine”, we are developing an engineering-based modeling technology with some adaptations to fit the special characteristics of the natural “climate machine” in modeling the CO2 fluxes of terrestrial ecosystems. This engineering-based modeling approach starts with experimental design followed by performance testing and observation. From the testing results, we developed a new modeling tool based on feature selection and REDA techniques to model and analyze the dominant factors or features and their interactions that have significant influence on a specific function (such as CO2 flux) of the “climate machine”. This new modeling technology will enable the analysis of spatial patterns and global correlations of CO2 fluxes based on the data from approximately 200 CO2 flux sites at AmeriFlux, CarboEurope, Canada Fluxnet, and AoFlux (we are collaborating with these countries now). The model and analysis will be scaled up to separate regions as per different types of transects (e.g. grass or forest), which further integrates the MODIS satellite data and weather reanalysis data. The research results can further help explore realistic and effective strategies for function optimization and manipulation by humans (e.g. control of atmospheric CO2 concentrations through regulating its sources/sinks). This new modeling technology has been applied in practice, which shows great effectiveness and efficiency. Some modeling results for global climate changes will be shown in this presentation.

back to agenda