CCSM Land Model Working Group Meeting Report
25 and 26 February 2003
NCAR, Boulder, Colorado
by Gordon Bonan, Co-Chair
The Land Model Working Group held a meeting in Boulder on February 25 and 26, 2003. The meeting had three goals: (1) At the June 2002 Breckenridge meeting, the working group agreed to finalize and release CLM2.1. One goal of the current meeting was to inform the working group about the status and release of CLM2.1. (2) The working group also agreed at the June 2002 meeting on several short-term changes to CLM to fix known biases in the model. A second goal of the meeting was to review these changes and to discuss whether they were ready to be included in the next version of the model (CLM2.2). (3) Finally, the working group discussed long-term model development.
CLM2.1 - Mariana Vertenstein (NCAR) gave an introduction to the software engineering of CLM2.1. This model, agreed upon during the June 2002 working group meeting, was released prior to the working group meeting. Software engineering differences from CLM2.0 were described, as were new features of CLM2.1.
CLM2.2 - The working group also agreed at the June 2002 meeting on several short-term changes to CLM to fix known biases in the model. The CCSM Scientific Steering Committee (SSC) has expressed a strong desire that component models fix prominent biases before CCSM is used in IPCC simulations. For the land model, these changes were: reformulating within-canopy ground heat transfer resistance (Xubin Zeng, Guiling Wang, Robert Dickinson, lead scientists); reformulating the fractional cover of snow on the ground (Zong-Liang Yang, Guo-Yue Niu, lead scientists); adding a correction for atmospheric stability to the 2-m temperature (Robert Dickinson, lead scientist); and adding a 2-m diagnostic air temperature over grass (Robert Dickinson, lead scientist). Other changes that were not thought to be ready for implementation were: efficient leaf temperature algorithm (Guiling Wang, lead scientist) and improved interception and sub-surface runoff (Guiling Wang, lead scientist). The SSC requested these changes be finalized by April 8.
Mike Barlage (University of Arizona) described the status of the within-canopy ground heat transfer resistance. This is a constant in CLM, leading to high resistance and high ground temperature over sparsely vegetated areas. The proposed formulation is an exponential function of leaf and stem area index. It cools ground temperature in sparsely vegetated areas by several degrees. This formulation was accepted by the working group and is ready for implementation in CLM2.2. Barlage also presented work on an improved dataset for fractional vegetation cover. The new dataset increases the bare soil fraction. Preliminary simulations show that it significantly improves the simulated climate. Longer simulations with CAM/CLM (e.g., over the period 1979-2000) are needed before the full impacts of the new dataset can be assessed.
Guo-Yue Niu (University of Texas) gave an update on improvements to snow hydrology in CLM. The snow cover fraction currently used in CLM results in a low fraction of the ground covered by snow even with deep snow. Low snow cover contributes to low surface albedo and is thought to contribute to the warm winter temperature bias in the Northern Hemisphere. A proposed algorithm differentiates snow cover fraction during accumulation and melt periods. This new algorithm improves simulated snow depth, surface albedo, and temperature and helps reduce the winter warm bias in the Northern Hemisphere. One problem with the new algorithm is with the transition from accumulation to melt seasons. It allows changes from accumulation phase to melt phase and back to accumulation from day-to-day. Working group members felt that this "flip-flopping" behavior is not desirable. One proposed alternative would be to fix the switch from accumulation to melt phase based on calendar day, which would vary with latitude. Peter Thornton proposed another approach that would use the proposed accumulation and melt phase fractional snow curves but provide a smooth rather than abrupt transition between these phases. Preliminary tests are very encouraging.
Keith Oleson (NCAR) implemented the proposed algorithms in the CLM2.1 framework and presented results of CAM/CLM simulations for the period 1979-2000. Simulations with the new within-canopy transfer resistance and fractional snow cover replicated previous work and was accepted by the working group (with the caveat of improving the snow fraction discontinuity). Robert Dickinson had previously noted that the 2-m temperature equation does not use corrections for atmospheric stability and that this can be seen in the model's simulated diurnal cycle. Adding this correction improves the diurnal cycle. The working group accepted this change. Oleson also noted that there was a discontinuity in the CLM equation relating the bulk density of fresh snow to air temperature. This was fixed upon the recommendation of Yongjiu Dai, principal developer of CLM. A final change requested by the Atmosphere Model Working Group was that the atmosphere model should determine the snow and rain fractions of precipitation. Currently, this is being done in the land model. The working group endorsed this change.
Oleson presented results on the implementation of a 2-m temperature diagnostic over grass. The goal of this, as originally proposed by Robert Dickinson, was to better match the observed 2-m temperature, which is typically measured over grass. Oleson implemented this diagnostic as a new grassland tile in every grid cell (except for those fully covered by lakes or glaciers). This is in contrast to Dickinson's approach, which took advantage of existing grassland tiles in grid cells. The problem with Dickinson's approach is that not every grid cell has grassland and, as dynamic vegetation is added to the model, these grassland tiles change over time. However, Oleson's implementation significantly increases the computational burden of the model. Moreover, it is not clear how to define surface properties (leaf area index, stem area index, height, soil color, soil texture) for the grassland tiles. Members of the working group felt that the 2-m grass temperature diagnostic was still in a preliminary stage of development. They endorsed a recommendation to not include this in CLM2.2, but to provide Oleson's implementation to Dickinson and his group for further development.
Forrest Hoffman (Oak Ridge National Laboratory) discussed software engineering changes to the CAM/CLM interface. These changes improve the communication between CAM and CLM and the scalability of CLM. They have already been endorsed by the Atmosphere Model Working Group and were endorsed by the working group.
Long-term model development - The remainder of the meeting was devoted to presentations related to long-term model development. One important direction is to facilitate biogeosciences research in the areas of biogeochemistry, carbon cycle, and dynamic vegetation.
Randy Koster (NASA/GSFC) gave an update on data assimilation activities and land model development at Goddard. Model development included Guiling Wang's work on within-canopy heat exchange in collaboration with Robert Dickinson and Xubin Zeng. It also included her work on leaf temperature, interception, and runoff. As discussed at the June 2002 meeting, this work has been endorsed for future model versions. One issue raised during the current meeting is that the CSU/SiB land modeling group is very interested in adding carbon isotopes to CLM. This requires within-canopy storage of carbon, which could also be added to heat and moisture. They proposed using the SiB surface flux parameterization. An evaluation of how to add canopy storage and the merits of the two proposed changes (Wang's, SiB) will be discussed during the June 2003 working group meeting.
Dave Williamson (NCAR) discussed research using CAM/CLM in numerical weather prediction mode for parameterization development. There is interest within the Atmosphere Model Working Group in this approach, which requires adding extra variables to the CLM initial file to completely replicate the state of the land.
Menglin Jin (University of Maryland) presented work using MODIS surface emissivity in CLM. She demonstrated the sensitivity to CLM to specified emissivity. Increased emissivity cools the ground, especially in desert and sparse vegetation. She suggested improvements to emissivity based on MODIS.
Mike Barlage (University of Arizona) presented work on integrating MODIS BRDF albedo with CLM. This work is leading to a new BRDF-based treatment of radiative transfer through the canopy for CLM.
Beth Holland (NCAR) gave an overview of the NCAR Biogeosciences Initiative. Development of the land model is very relevant to this initiative in terms of biogeochemical cycles, the coupling of energy, water, carbon, and nitrogen, and perturbations to these cycles through land use and land cover change.
Peter Thornton (NCAR) gave an update on his development of CLM to include the carbon and nitrogen cycles.
Sam Levis (NCAR) gave an update on his work to include dynamic vegetation in CLM.
Based on these presentations and recommendations from the June 2002 meeting, the following areas of research were endorsed as active research efforts for future versions of CLM: fractional vegetation cover (bare soil fraction); emissivity; leaf temperature and canopy storage; infiltration and runoff; 2-m temperature over grass; carbon and nitrogen cycles; dynamic vegetation; mineral aerosols; and biogenic volatile organic compounds. It is hoped that rapid progress can be made in each of these research areas for inclusion in future versions of CLM.
Gordon Bonan, NCAR
Ken Eggert, LANL
Randy Koster, NASA GSFC
Marcia Branstetter, ORNL
Guo-Yue Niu, Univ of Texas at Austin
Edward Davin, NCAR
Forrest Hoffman, ORNL
Beth Holland, NCAR
Michael Barlage, Univ of Arizona
Keith Oleson, NCAR
Neil Suits, CSU
Ina Baker, CSU
Jeff Anderson, NCAR/GFDL
Bette Otto-Bliesner, NCAR
Menglin Jin, Univ of Maryland
Sam Levis, NCAR
Peter Thornton, NCAR
Michael Notaro, Univ of Wisconsin-Madison
Dave Williamson, NCAR
Mariana Vertenstein, NCAR
David Baker, NCAR
Kevin Raeder, NCAR