Implementation of biogeochemical cycles in CCSM has been a high-priority item for DOE Office of Biological and Environmental Research. The Climate Change Prediction Program (CCPP) of the Climate Change Research Division has sponsored this work through the DOE Office of Science SciDAC Program.
The software integration effort has been led by Argonne and work has progressed steadily with strong collaboration of the NCAR CCSM Software Engineering Group (CSEG). A document outlining the necessary software engineering and design was written in 4Q04, and the effort resulted in 3Q05 deliverable at which time a complete prototype was run for a five-day smoke test.
Progress and the milestones are documented on the project wiki .
The FY05 Milestone for this project was accomplished with a 5 year integration of the complete model.
The complete prototype consists of the biogeochemistry-enabled POP2.0 (replacing POP1.4.3 in the CCSM3), a biogeochemistry-enabled version of CLM3 using the CASA' land model, and a full tropospheric chemistry model (CAM3 finite volume dynamics plus MOZART chemistry). The active ocean ecosystem model based on DML, employs the CO2_DMSA coupling option for fully interactive CO2 and DMS fluxes between atmosphere and ocean. This is the alpha version of an earth system model incorporating dynamic ecosystems and biogeochemical coupling.
The milestone for FY05 was to demonstrate the software engineering and feasibility of model integrations. This milestone has been met ahead of schedule thanks to the considerable efforts of our SciDAC Consortium and the members of the CCSM Biogeochemistry Working Group. More extensive testing of the fully coupled system and scientific exploration of feedbacks within this system will now begin. Once fully developed, this model will be a significant scientific tool in exploring climate dynamics.
The evidence of active linkage of the atmospheric chemistry to the ocean biogeochemistry (through the generalized coupler) is seen from model DMS and DMS flux output of the atmosphere, ocean, and coupler components.
The carbon cycle linkages are demonstrated from the transported atmospheric CO2, the ocean partial pressure of CO2, and the coupler carbon flux.
Chemically active components are demonstrated by the atmophseric OH and the ocean concentrations of iron (FE) and NO3.
A more comprehensive set of plots and animations is given at the project wiki .
The first accomplishment to note is that the model appeared to be stable over the five year integration. This is partially attributed to the initial conditions. Land carbon pools were initialized with year 69 of f12.001 run of CAM3 with CLM3 and CASA'. After a one month initial run, a continuation run was started using an 8-year spinup ocean restart which was equilibrated with atmospheric CO2 levels of 360ppm. Terrestrial wood-carbon pools show a slight declining trend but no conclusion can be drawn from these preliminary results. Before continuing the simulation, a careful examination of the present results will be undertaken and a control run generated with the coupled system using the FV Dycore.
The spatial structure of DMS in the atmosphere reflects the input of DMS, passed through the coupler, emitted by the ocean ecosystem model. The seasonal cycle of the DMS system is reasonable. The distribution of C02 and DMS in the atmosphere model reflects surface sources and sinks due to anthropogentic and ecological processes in the ocean and the terrestrial biosphere.
Software changes to MCT and CPL6 were straightforward. MCT has proven robust and adaptable, requiring only one minor modification to allow CPL6 to handle exceptions from MCTs AttrVect query routines (which was released as part of MCT 2.1.0). CPL6 and interfaces to it from the other components in CCSM were modified to make it more flexible and to take fuller advantage of MCTs novel features. We removed the hard-coded integer field indices in cpl_fields_mod.F90, and added two new functions: cpl_interface_contractIndex(), which uses string tokens to index fields, and cpl_interface_contractNumatt(), which returns the number of fields stored. These changes have expanded the coupled systems capabilities in the following ways: 1) component models now set the indices used for moving data between the coupler buffers and internal datatypes (automatic and thus less error-prone); 2) the number of and specifically which chemical species are exchanged can now be set at run-time; and 3) interfaces between the physical components and the coupler (e.g., ccsm_msg.F90) can now include code for multiple tracer combinations without CPP directives.
The computer resources for this project have been provided by the ORNL Center for Computational Sciences, the LBNL National Energy Research Supercomputer Center, and by the NCAR Climate System Laboratory.