[an error occurred while processing this directive] [an error occurred while processing this directive]

Report on CCSM Land Model Working Group Meeting

NCAR, Boulder, CO
Gordon Bonan and Paul Houser, co-chairs
27 and 28 March 2002

The Land Model Working Group held a joint, one-day meeting with the Biogeochemistry Working Group (March 27) and a separate one-day working group meeting (March 28).

Wednesday March 27

Wednesday's meeting was joint with the CCSM Biogeochemistry Working Group. The day was spent discussing a research and model development agenda that meets the requirements of both groups.

Gordon Bonan (NCAR) began the meeting with an overview of the current status of the Community Land Model (CLM2). Since the June 2001 Breckenridge meeting, several climate model simulations have been conducted with CLM2. A 17-year AMIP style simulation was performed with CCM3. This allowed comparison of CLM2 with the NCAR LSM - the land model for CCSM1. This study has been written-up and accepted for publication in Journal of Climate (Bonan, Oleson, Vertenstein, Levis, Zeng, Dai, Dickinson & Yang (2002) J. Climate, in press) and is available on the web (http://www.cgd.ucar.edu/tss/clm/). During the fall of 2001 and winter of 2001/2002 a 350-year spin-up of the fully coupled CCSM2 was performed. Bonan noted that soil water in the 10th soil layer decreased during the course of this simulation. During this time the ocean freshened. The spin-up simulation provided initial conditions for the CCSM2 control run. The simulated surface climate looks very good compared to observations with the following exceptions (results posted at the CLM web site (http://www.cgd.ucar.edu/tss/clm/)). The high latitudes of the Northern Hemisphere have a large (several degrees) warm bias in winter. Tropical America has too little rainfall. Jeff Kiehl has noted these biases and requested the LMWG address them. Bonan also discussed prototype coupling of a dynamic global vegetation model with CLM2. This coupling was initiated several years ago to outline and formalize a strategy for including dynamic vegetation in CLM2. This coupling has been successful, with a 350-year simulation with CLM2. However, it revealed a deficiency in CLM2. CLM2 intercepts a large amount of water on the plant canopy, resulting in dry soils.

Samuel Levis (NCAR) showed results of CLM2 with dynamic vegetation. The model simulates lower than expected net primary productivity, less global extent of trees, and greater geographic coverage of grasses. This indicates a dry soil, which was attributed to excessive interception. This problem is especially noticeable in the tropics, where the model simulates a high abundance of raingreen trees and grasses rather than evergreen trees.

Peter Thornton (NCAR) reported on model developments necessary to allow carbon and nitrogen cycling in CLM2. This involves changing data structures from a single vector of spatially independent sub-grid patches to one that recognizes three scales: land unit, snow/soil column, and plant functional type. Furthermore, plants can co-occur on a single soil column and compete for water. A similar report was given on the previous day during the Biogeochemistry working group meeting.

Much of the remainder of the day was spent discussing Thornton's proposed model changes. The members of the two working groups were cautiously optimistic that the changes would meet the needs of both the LMWG and BGCWG. A sub-committee of interested working group members was formed to further study this proposal. This committee consists of: Thornton, Hoffman, Vertenstein, Houser, Zeng, Fung, King, Holland, Yang, Noone, and Ojima. The proposed code will be distributed to these members as soon as possible for their comment.

Jay Famiglietti (UC-Irvine) discussed future developments related to river routing. There was a general agreement among working group members of the need to transport nutrients and materials in rivers and allow for dynamic wetland and river areas. He also discussed model development related to catchment-based river routing rather than the current cell-to-cell scheme. This allows for river flow within a watershed based on topography and transport between watersheds using the cell-to-cell routing scheme.

Steve Frolking (UNH) discussed issues related to adding the methane cycle to CCSM. This includes both wetland hydrology and wetland ecology, neither of which is represented in the current CLM2.


Thursday March 28

Dai (Georgia Tech.) discussed a new scheme for integrating leaf stomatal conductance and photosynthesis over the plant canopy. He also showed preliminary coupling of CLM with the WRF mesoscale model.

Dickinson (Georgia Tech.) discussed a sub-grid surface heterogeneity scheme for CLM2. He also reported on developments in a new 2 m temperature parameterization. This parameterization is based on a grass tile that is included in each grid cell. It is also based on either maximum/minimum daily temperatures or on a 24-hour average. Both methods give similar results.

Yang (Univ. Texas) discussed developments in the snowpack thermal model, fractional snow cover, and snow albedo. He showed that the CLM multi-layer snow model improves the simulation compared to the old NCAR LSM scheme. However, the new scheme substantially reduces the snow-covered fraction of the grid cell simply from the relationship of snow depth to snow cover. He also discussed issues related to the freezing point of soil water. Changing freezing point from 0C to -2C causes the surface to be more permeable and streamflow is better simulated. He also discussed improved runoff parameterizations for CLM. This is related to the assumed decrease in hydraulic conductivity with depth.

Zeng (Univ. Arizona) discussed developments in fractional vegetation cover and surface albedo based on satellite measurements.

Hoffman (ORNL) described the requirements document for CLM. He hopes to have a draft document prior to the June meeting for comments during the meeting.

Wang (GSFC) talked about new energy and water balance equations for CLM. Her new scheme results in faster solution convergence compared to the existing scheme.
It is numerically more efficient and does not adversely affect model results.

Koster (GSFC) talked about new watershed-based runoff scheme. This scheme discretizes the land surface not into grid cells but into watersheds.

Peters-Lidard (GSFC) talked about sub-grid heterogeneity schemes for mesoscale models.

Jin (Univ. Maryland) talked about using satellite-derived skin temperature and emissivity to evaluate land models.

Bonan (NCAR) discussed the upcoming CSL proposal and requested input from the LMWG about proposed experiments. Anyone asking for CSL time is required to describe the proposed experiment (e.g., what component models are to be used, length of simulation) and scientifically justify the simulation.