Capturing Vegetation Diversity in the Ent Terrestrial Biosphere Model

We present preliminary results from data mining to develop parameter sets and global vegetation structure datasets to set boundary conditions for the Ent Terrestrial Biosphere Model (Ent TBM) for improved representation of diversity and to propagate uncertainty in simulations of land carbon dynamics in the 20th century and under future climate change. The Ent TBM is the only dynamic global vegetation model (DGVM) developed for coupling with general circulation models (GCMs) to account for the height structure of mixed canopies, including a canopy radiative transfer scheme that accounts for foliage clumping in dynamically changing canopies. It is flexibly programmed to incorporate any number of "plant functional types" (PFTs). It is now a coupled component of the ModelE2 version of the NASA Goddard Institute for Space Studies (GISS) general circulation model (GCM). We demonstrate a data mining method, linear manifold clustering, to be used with several very recently compiled large databases of plant traits and phenology combined with climate and satellite data, to identify new PFT groupings, and also conduct customized parameter fits of PFT traits already defined in Ent. These parameter sets are used together with satellite-derived global forest height structure and land cover derived from a combination of satellite and inventory sources and bioclimatic relations to provide a new estimate and uncertainty bounds on vegetation biomass carbon stocks. These parameter sets will also be used to reproduce atmospheric CO2 time series over the flask observational period, to evaluate the impact of improved representation of vegetation dynamics on soil carbon stocks, and finally to produce a projection of the land carbon sink under future climate change. This research is timely in taking advantage of new, globally ranging vegetation databases, satellite-derived forest heights, and the advanced framework of the Ent TBM. It will advance understanding of and reduce uncertainty in land carbon dynamics and stocks through development, implementation, and evaluation of new parameterizations for a DGVM, parameterizations which will expand and or revise representation of vegetation diversity in the Earth system or replace modules or static parameters with more continuous, functional schemes.