HR: 0800h
AN: A51A-0078
TI: Atmospheric CO2 simulation inside GEOS-5: Data mining, evaluation and treaty verification
AU: Erickson, D J
EM: ericksondj@ornl.gov
AF: Computational Earth Sciences, Oak Ridge National Lab, Oak Ridge, TN, United States
AU: Pawson, S
EM: steven.pawson-1@nasa.gov
AF: Goddard Space Flight Center, NASA, Greenbelt, MD, United States
AU: Daniel, J
EM: d65@ornl.gov
AF: Computational Earth Sciences, Oak Ridge National Lab, Oak Ridge, TN, United States
AU: Allen, M
EM: allenmr@ornl.gov
AF: Computational Earth Sciences, Oak Ridge National Lab, Oak Ridge, TN, United States
AU: Ott, L E
EM: lesley.e.ott@nasa.gov
AF: Goddard Space Flight Center, NASA, Greenbelt, MD, United States
AU: Ganguly, A
EM: gz4@ornl.gov
AF: Computational Earth Sciences, Oak Ridge National Lab, Oak Ridge, TN, United States
AU: Nielsen, E
EM: nielsen@gmao.gsfc.nasa.gov
AF: Goddard Space Flight Center, NASA, Greenbelt, MD, United States
AU: Hoffman, F
EM: hof@ornl.gov
AF: Computational Earth Sciences, Oak Ridge National Lab, Oak Ridge, TN, United States
AB: We present a multi-year simulation of atmospheric CO2 inside the NASA GEOS-5 modeling and assimilation system. These calculations represent an example of a simulated global climate model rendition of the NASA data streams that will flow into the geophysical/climate community science and assessment community over the next 5-10 years. The 3-D atmospheric wave structures and transport physics interact with spatially and time varying surface sources and sinks of CO2. This results in an exceedingly complicated evolution of atmospheric CO2 in time and space. Approaches such as this are applicable for initiating the assimilation and data mining of remotely sensed atmospheric CO2 concentrations such as those that will be available from the ASCENDS mission. The 4-D model simulations and satellite generated observations presented here comprise massive (100’s of Terabytes) records of hundreds of climate and geophysical variables. These represent new opportunities for developing knowledge discovery and data mining applications, including but not limited to descriptive analysis, predictive modeling and anomaly detection. Our analysis with similar data sets have revealed that while na´ve applications of existing data mining tools may offer some predictive insights, new scientific understanding and significant advances in predictive modeling or anomaly analysis require data mining algorithms tailored to the specific attributes of climate data. These findings have implications for inverse models that attempt to estimate surface source/sink regions especially when the surface sinks are co-located with regions of strong anthropogenic CO2 emissions. Intensive data mining and knowledge discovery will contribute to the accurate determination of the sources and sinks of atmospheric CO2 , thus allowing quantitative input for treaty verification.
DE: [0322] ATMOSPHERIC COMPOSITION AND STRUCTURE / Constituent sources and sinks
DE: [0368] ATMOSPHERIC COMPOSITION AND STRUCTURE / Troposphere: constituent transport and chemistry
DE: [0428] BIOGEOSCIENCES / Carbon cycling
DE: [3315] ATMOSPHERIC PROCESSES / Data assimilation
SC: Atmospheric Sciences (A)
MN: 2009 Fall Meeting

Acknowledgements
Research partially sponsored by the Climate and Environmental Sciences Division (CESD) of the Office of Biological and Environmental Research (OBER), U.S. Department of Energy Office of Science (SC). This research used resources of the National Center for Computational Science (NCCS) at Oak Ridge National Laboratory (ORNL) which is managed by UT-Battelle, LLC, for the U.S. Department of Energy under Contract No. DE-AC05-00OR22725.