Alumni Project
Improving the Processes of Land-Atmosphere
Interactions in CCSM 2.0 at High Resolution
Robert E. Dickinson, Gatech
William Chamedies, Gatech
Yongjiu Dai, Gatech
Andrea N. Hahmann, UA
Yongqiang Liu. Gatech
Muhammad Shaikh, Gatech
Wanru Wu, Gatech,
Hongin Yu, Gatech
Summary
This project addresses the issue of integrating the land component of the NCAR climate model at much higher resolutions than is current practice. It does so in several directions: a) improvement of the processes parameterizations that become increasingly important at high resolution; b) provision of the data sets needed at the high resolutions to be used; c) development of methodologies to integrate land at higher resolution than the atmospheric model; d) simulations with latest NCAR models at the highest resolution feasible- and assessment of the impacts of the high resolution and new data sets; e) develop treatments of chemical exchanges with the land surface that require a high resolution land model.
Our main focus on land process parameterizations has initially to get the new NCAR land model working properly. This model was developed in two stages with the end results of both still being actively research (Zeng et al, 2002, Bonan et al, 2003). Initially, the new land model was developed by a many year effort lead by Yongjiu Dai (Dai et. al, 2003) . The code delivered to NCAR was extensively further developed at NCAR to better fit the NCAR computational environment and research needs of NCAR land scientists and then publicly released in 2002 as CLM 2.0. We have addressed several performance problems of the NCAR version through collaborative work with colleagues. The 3 most serious issues on which we made major progress are: a) excessively high soil surface temperatures under sparse vegetation because of an inappropriate parameterization for transfer processes between soil and canopy air space; b) too much reevaporation from canopy – several possible subgrid parameterizations were identified and tested to show they fixed the problem; c) too much soil evaporation in tropical forests – shown to result from inappropriate data for the canopy properties and to go away with use of better data.
Our primary effort to generate model data makes use of collaborative work with NASA scientists in a related NASA sponsored project to use the data from the MODIS instrument on the terra platform that was launched in 2000 and now has about 3 years of data available. We are starting with 1 km data for LAI, and albedo. composited every 16 days and generating annual cycle descriptions of these fields at the resolutions needed by our DOE project. In addition as part of the present project, Hongbin Yu has worked with NASA scientists to provide a global data set for aerosol properties that can be used to improve our simulation and understanding of aerosol impacts on land surface climate.
Methodologies to integrate land at higher resolution than the atmospheric model were developed in our earlier DOE projects. Their further extension and testing is being led by Hahmann at the University of Arizona. The objective is to improve the representation of the heterogeneities of the land surface through a fine-mesh representation. Considerable attention was paid to conservation issues in the disaggregation from the atmospheric model to the land submesh and the aggregation of fluxes back to the atmosphere. Substantial progress has been made on the issues of disaggregation of topographic aspects and of precipitation. The effect of topographic differences are taken into account by including the slope and the aspect of each point in the land submesh. Precipitation is allowed to occur over a small fraction of the atmospheric gridsquare through use of the land submesh. Where that occurs is determined randomly and is uncorrelated in time.
An additional effort by Dickinson and Shaikh has developed and tested a methodology that combines the fine mesh approach with the mosaic approach of the current NCAR land model. Its objective is to use the benefits of the fine mesh approach but improve the accuracy of the land cover representation. This framework has also been used to examine details of the diurnal cycle of surface temperatures and their dependence on land cover.
Efforts to integrate latest NCAR models at high resolution are being led by Mohammad Shaikh. An earlier effort along these lines in our previous DOE project carried out multiple resolution integrations. We have been in communication with a similar effort at LLNL and have discussed this objective with ORNL scientists. Our integrations will build on the latest land model version, CLM2.1 released in early 2003 and include the currently best available fixes to the know defects of the released model. It will also include the latest satellite land data that has been developed through our NASA project. After testing at standard resolution, a series of higher resolution simulations are to be executed.
Our subproject to incorporate land surface chemical fluxes is currently developing the use of our new satellite data sets to improve estimates of fluxes of isoprene into the atmosphere.
Contact: Robert Dickinson
robted@eas.gatech.edu
back to project page
