Notes from the QUEST Workshop on DGVMs and PFTs
October 2005, Exeter, UK

Tom Hilinski
Natural Resource Ecology Laboratory, Colorado State University

Summary

The focus of the workshop was on current modelling efforts using dynamic global vegetation models (DGVMs) in the context of issues relating to plant functional types (PFTs). LPJ and its variations were the object of most of the presentations. Almost all of the simulations described were global in scope, with regional results discussed as variations within the global context.

The presentations and discussions fall into three categories. Several DGVMs were presented with examples of their use. Analyses of vegetation responses to DGVM scenarios covered a range of "what if" and sensitivity issues. Discussions of PFTs in the contents of DGVM development considered the issues involved in defined PFTs.

As stated on the QUEST web site, the "workshop, dealing with Dynamic Global Vegetation Modelling and Plant Functional Types, led to the planning of a multi-author synthesis article to be headed by Dr Sharon Cowling (University of Toronto) provisionally entitled "Theory of plant function and its representation in dynamic global vegetation models". This will put flesh on the workshop's main conclusion: that there now exists sufficient information about the interrelations of plant traits and their functional significance to build a new generation of models in which the representation of plants is both simpler (requiring far fewer independent parameters) and more realistic than in current models. The discussions at the workshop will feed into the modelling and data analysis activities of QUEST Theme 1 (the QUERCC project) as well as helping to drive a research agenda at the international level through IGBP and DIVERSITAS."

Subjects I think are particularly applicable to our work include status of the Lenihan fire model (see Additional Discussions), and sensitivity analyses done by the Hadley Center (see Chris Jones' presentation).

The following notes reflect points I found particularly interesting or applicable to our work, and is not intended to be a comprehensive record of the workshop.

Models


LPJ

Global 1/2 degree resolution (~62K cells). Input includes climate, soil quality. Has 13 crop functional types. Calculates water stress (not just soil water content) as a function of canopy water.

Has been linked with a agricultural economic model "MAgPIE" (publication in preparation). Does cost minimization, 30 economic activities.

Link: www.pik-potsdam.de/lpj/

LPJ-GUESS

Scope is global as it uses LPJ. Uses dynamic age classes for gap analysis. Has same physiology as LPJ, with stochastic mortality added. Age classes have associated shade tolerance classes. Simulates succession (Smith, Sykes, et al, 2004) using a general vegetation algorithm.

Applications include succession; water stress in tree parts; island succession and colonization; EUROFLUX model-data intercomparison; FACE.

Link: www.natgeo.lu.se/embers/research.htm#lpj

ORCHIDEE

Comprised of linked biosphere-crop models. Calculated biosphere energy and hydrology balances. Crop model feedback includes N stress and irrigation. Uses a layered soil model. Has been coupled to a GCM via a flux coupler.

Link: www.ipsl.jussieu.fr/~ssipsl/

CLIMBER

This is a coupled atmosphere-ocean-TCM model with inland ice. The TCM is a simple C cycle with annual time step. Tree+grass controlled by climate. Ocean has a carbon cycle and sea ice.

Link: www.pik-potsdam.de/climber-3/

MC1 vs. SDGVM1

MC1 uses Century4 (?) for plant/soil. SDGVM uses Farquar photosynthesis. Comparison using elevated CO2 shows differing response. MC1 underestimates NBP and CO2. SDGMV1 overestimates them. Both have problems simulating boreal fire (due to the Lenihan fire model).

Link: www.cgd.ucar.edu/vemap/abstracts/MC.html

Analyses


Sharon Cowling

Used LPJ to examine "what if" effects of C4 plant dominance/losses. Experiments: (1) C4 plants removed globally. (2) C4 plants dominate globally. (3) Remove Amazon forest. Developed an index of soil:veg C and compared to changes in temperature and precipitation. Found distinct thresholds - pristine index < 1-2, disturbed 3-4+. Matching thresholds in Amazon field data and sims using the Hadley model.

Doris Barboni

Examined European plant phenology and morphology vs. climate using pollen. (Barboni, et al. Relationships between plant traits and climate in the Mediterranean region: A pollen data analysis. J. Veg. Sci. 15, 635–646) Established a plant trait -- climate relationship. Not enough info to extend this relationship to nutrient and C storage.

Rosa Maria Roman-Cuesta

Fire vs. PFT, emphasis on S. Amer. rain forest. Fire types = surface, crown, ground. Ground is common in tropical cloud-montane forests and in peat fires. Fuel availalability depends upon biomass, plant types, size, moisture.

Flamability factors: chemical components in leaves/wood; deciduous leaf morphology; fuel arrangement (compactness and connectivity); litter + dead wood accumulation amount; self-pruning character; fuel depth.

To find patterns, need to consider regional scale + biome regime. There are temporal scale patterns in the burns, and in the factors influencing the burns.

Thomas Hinkler

Examined plant hydraulic architecture in relationship to PFT. Classified averaged hydraulic values for particular species. Leaf wilting point should be PFT-specific rather than biome-specific. Issue found when average value for a PFT deviates significantly from a species' actual value. Ron N. and Prentice had much discussion on what this means in terms of defining a PFT.

Chris Jones

Presented results from extensive sensitivity analyses done at the Hadley Center, using several different DGVMs. Methodology was to perturb one type of input stochastically. The models agree "OK" up to the year 2000 or so. For the year 2100, the responses (e.g., CO2 fluxes) varied by ~10X.

Considering CO2 emissions average for the ensemble of models runs: The C cycle parameters were less influential than the climate parameters by at least 1/2 or less. For the C cycle parameters, increasing in order of influence are CO2 fertilization, heterotrophic respiration, GPP. For the climate parameters, the influence of temperature change is proportional to the range of temperature change (i.e., 1.5 to 4.5 and 1.5 to 10 degrees C).

Xu-Ri Xing

Added global dynamic N to LPJ, and examined effects with increasing global CO2. Results were:

Additional Discussions


Working Groups on PFTs

Divided into groups, each group considered what PFT information is needed for the next generation of DGVMs, and obtainable within 2 years. The results of these discussions will be posted on the QUEST web site.

Availability of Models

Dave Conklin, Sharon Cowling, and Christele Helly-Alleaume discussed experiences and issues related to obtaining the models, discovering which version or variant is best for you to use, and finding documentation. The ORNL biogeochemical model repository was discussed - this repository stores a "snapshot" of a model + data related to a publication. The need for this type of archive is discussed by Peter Thorton in an EOS essay (see the Links below). The decision was made to (1) start a Wiki on the QUEST web site (Rita Wania at U. of Bristol) , and (2) examine how to maintain a list of links to model pages. Issues were staffing, time, and money.

Lenihan Fire Model

In private conversations with the Oregon crew (Dave, Dominque, Ron), I received the following news regarding Jim Lenihan's fire model:

Links

QUEST: http://quest.bris.ac.uk/
ORNL Biogeochemical Model Archive: www-eosdis.ornl.gov/model_intro.shtml
Thorton, P., Archiving numerical models of biogeochemical dynamics: www-eosdis.ornl.gov/MODELS/EOS_Model_Archiving_Thornton.pdf

Images of Exeter

Crossmead Conference Centre
View of Exeter