CCSM Land Model Working Group Report
Wednesday, 7 July 2004
Eldorado Hotel, Santa Fe, New Mexico

Meeting Summary

The Land Model Working Group (LMWG) met on Wednesday, 7 July 2004. The format consisted of several 15-minute talks that provided updates to ongoing projects and to identify important new areas of research for the working group.

Gordon Bonan (NCAR) began the meeting with an overview of activities of the CCSM support staff over the past year. The highlights of the past 12 months pertained mostly to software engineering activities, especially the vectorization of the Community Land Model and the release of CLM3 with CCSM3. Technical descriptions, user's guides, and reference guides were written for CLM3 and its dynamic global vegetation model. These documents, as well as model source code and data sets, are available from the CLM release site (http://www.cgd.ucar.edu/tss/clm/distribution/clm3.0/index.html). Many thanks are owed to Mariana Vertenstein (NCAR), Forrest Hoffman (ORNL), Keith Oleson (NCAR), and Sam Levis (NCAR) for coordinating the release. Bonan concluded his presentation by noting that the past year was extremely productive for software engineering and release of CLM. Now the working group needs to focus its collective resources on science. Four primary areas of scientific development that build ongoing or new research into coordinated themes are: land cover change; biogeochemistry; hydrology; and a high-resolution CLM.

Keith Oleson (NCAR) gave an overview of CLM3/CCSM3 control simulations. The Northern Hemisphere winter warm temperature bias has been reduced compared to CCSM2, but the bias still exists and can be large (several degrees) in some regions. The Northern Hemisphere summer warm temperature bias has also been reduced from CCSM2. Precipitation biases are similar to CCSM2. The model performs reasonably well compared to monthly observations of temperature, precipitation, runoff, riverflow, and snow depth for selected regions. The simulation of surface climate in the Amazon, particularly low evapotranspiration, remains a problem. Oleson and Robert Dickinson (Ga. Tech.) are the lead authors for the paper documenting the land surface climate of CCSM3.

The next four talks concerned land cover change and its effects on climate. Johan Feddema (Univ. of Kansas) discussed the transient land cover change data sets he is developing and the planned experiments with CLM3/CCSM3. Peter Lawrence (Univ. of Colorado) described fine-scale (0.05 degree) data sets of land surface parameters (plant functional type, leaf area index, albedo) he is developing from Moderate Resolution Imaging Spectroradiometer (MODIS) data. Sam Levis (NCAR) described his implementation of a crop growth model for CLM, in collaboration with Jon Foley, Chris Kucharik, and Navin Ramankutty (Univ. of Wisconsin). Menglin Jin (Univ. of Maryland) described an urban land cover parameterization for CLM and the use of satellite products (skin temperature, albedo, emissivity) to study urban climates.

The next two talks focused on biogeochemistry. Peter Thornton (NCAR) gave an update on his implementation of the terrestrial carbon and nitrogen cycles in CLM. He showed how his new parameterization of sunlit and shaded portions of the canopy improves the simulated global photosynthesis. He also presented results of his simulations of the global carbon cycle. David Noone (Univ. of Colorado) discussed the implementation of water isotopes in CLM. He pointed out that to efficiently implement water traces and isotopes, CLM needs to compute tendencies in all water pools and then update all states at once. CLM3 currently does not do this.

The next four talks examined aspects of the hydrologic cycle. Aiguo Dai (NCAR) presented results from retrospective simulations with CLM3 for the period 1948-2000. He compared historical runoff, soil moisture, and other land surface conditions to the model simulation. Zong-Liang Yang (Univ. of Texas) presented highlights of his new developments in snow and runoff research. He showed that a vegetation radiative transfer model that accounts for 3-D canopy structure (e.g., clumping) and snow interception by foliage improves the simulated surface albedo. He also presented results of his research to derive a new parameterization of fractional snow cover for CLM from satellite data. He showed quite promising improvements in the Amazon related to his runoff research. Use of a topographically-derived runoff model (TOPMODEL) improves runoff in the Amazon by allowing more water to infiltrate. The wetter soil allows more evapotranspiration. Joe Alfieri (Univ. of Colorado) presented results of his comparison of CLM3 with tower flux and soil moisture data from the International H2O Project (IHOP) field experiment. He found the model was generally drier than the observations and underestimated latent heat flux. Ed Beighley (San Diego State Univ.) discussed a new continental scale river routing approach that uses the Pfafstetter decomposition and allows for parallel computer architecture. The initial implementation is for the Amazon basin, but the approach could be applied globally.

The final four talks of the meeting concerned high-resolution land modeling. Steve Ghan (PNL) discussed his parameterization of subgrid orography in CLM/CAM. His approach divides the model grid cell into several elevation classes. This subgrid tiling approach does a better job at resolving temperature, precipitation, and snow in western U.S. compared to the standard T42 implementation of CAM/CLM. Ruby Leung (PNL) discussed plans to develop a regional climate model using the Weather Research and Forecast (WRF) model. This involves implementing the CAM3 radiation and the CLM3 in WRF. She also discussed how she plans to implement river routing and surface/ground water interactions in the high-resolution model. Muhammad Shaikh (Ga. Tech) presented results of work he and Robert Dickinson are doing to downscale precipitation in CLM. Their approach distributes precipitation into multiple tiles in a grid cell. The downscaling greatly improves the hydrologic cycle in the Amazon by reducing canopy evaporation, increasing soil wetness, and increasing transpiration. Andrew Gettelman (NCAR) gave the final talk of the session. He presented preliminary plans to downscale atmospheric output from CAM to a high-resolution grid (1-20 km). His approach recognizes the need to distribute atmospheric data while conserving mass and energy and focuses on radiative fluxes, temperature, and precipitation.

Action Items

Based on these talks, the working group recognized the need to coordinate various research tasks. In particular, several research tasks on the biases in the Amazon need to be systematically evaluated. Promising new research activities include water isotopes and high-resolution land modeling.

Subsequent to the CCSM meeting, Mariana Vertenstein (NCAR), who provides software engineering support for CLM, was assigned to head the CCSM Software Engineering Group. Her new duties will restrict the amount of time she can support the scientific development of CLM. Ongoing scientific activities that will continue to be supported include land cover change (crop model, urban model, transient land cover change) and the terrestrial C/N development, in addition to ongoing software engineering needs. In addition, NCAR staff will coordinate study of the Amazon biases and assist with papers documenting CLM3 for the Journal of Climate special issue.

Scientific developments that cannot be supported by NCAR staff include: isotopes, continental river routing, and high-resolution land modeling. It is hope that this work will be continued by the non-NCAR members of the working group.

Isotopes - To efficiently implement water traces and isotopes, CLM needs to compute tendencies in all water pools and then update all states at once. CLM3 currently does not do this. At the June 2003 CCSM meeting, the LMWG endorsed re-writing substantial parts of CLM to do this. Some progress was made in the spring of 2004, but substantial more work remains to be done.

Continental-river routing - The Pfafstetter approach is very promising because it allows for a high-resolution river routing network and also allows for parallel computer architecture. The initial implementation for the Amazon basin requires coupling CLM3 to the river routing model. In addition, the current 0.5-degree linear reservoir river routing model could be replaced by a high-resolution (5-minute) model developed by Mike Coe (Univ. of Wisconsin) that includes lakes and reservoirs. However, this effort requires substantial software engineering support to run on parallel computer architectures.

High-resolution CLM - There is substantial interest and multiple efforts to implement a high-resolution CLM, either through sub-grid tiling or by running CLM directly on a high-resolution grid. However, this work needs strong software engineering and scientific coordination. The resources are not available to coordinate this work.

Attendees:

Joseph Alfieri University of Colorado
Ian Baker Colorado State University
Michael Barlage University of Arizona
Edward Beighley University of California Santa Barbara
Gordon Bonan NCAR
James Bossert Los Alamos National Laboratory
Marcia Branstetter Oak Ridge National Laboratory
Scott Denning Colorado State University
Kenneth Eggert Los Alamos National Laboratory
Johannnes Feddema University of Kansas
Arthur Few Rice University
Mark Flanner University of California Irvine
Robert Gallimore University of Wisconsin Center for Climatic Research
Paul Higgins University of California Berkeley
Forrest Hoffman Oak Ridge National Laboratory
Menglin Jin Univeristy of Maryland
Nancy Kiang Columbia University
Yeonjoo Kim University of Connecticut
Randal Koster NASA Goddard Space Flight Center
Lara Kueppers University of California Santa Cruz
Peter Lawrence University of Colorado
Samuel Levis NCAR
Cynthia Nevison NCAR
Keiichi Nishizawa Central Research Institute of Electric Power Industry (CRIEPI)
Michael Notaro University of Wisconsin Center for Climatic Research
Keith Oleson NCAR
Taotao Qian NCAR
William Sacks NCAR
Muhammad Shaikh Georgia Institute of Technology
Willis Shem Georgia Institute of Technology
Allison Steiner University of California Berkeley
Peter Thornton NCAR
Wanru Wu Georgia Institute of Technology
Zong-Liang Yang University of Texas at Austin