Improving and extending CMS land surface carbon flux products including estimates of uncertainties in fluxes and biomass.

George James Collatz, NASA GSFC, jim.collatz@nasa.gov (Presenter)
Stephan Randolph Kawa, NASA GSFC, stephan.r.kawa@nasa.gov
Lesley Ott, NASA GSFC, lesley.e.ott@nasa.gov
Alvaro Ivanoff, NASA GSFC/ADNET, alvaro.ivanoff-1@nasa.gov
Fanwei Zeng, NASA GSFC/SSAI, fanwei.zeng@nasa.gov
Yuping Liu, NASA GSFC/SSAI, yuping.liu-1@nasa.gov

The focus of this NASA CMS project is to provide land surface biological and fire fluxes as boundary conditions for atmospheric transport models. We use the CASA-GFED3 model driven by MERRA meteorology and GIMMS3g NDVI. Physiological (NPP, RH) and fire fluxes are simulated at monthly, 0.5o resolutions. Physiological fluxes are further decomposed in time to 3 hourly by scaling with 3 hourly solar radiation and temperature. Fire fluxes are scaled to quasi-daily using a MODIS active fire product. Our flux products span 2003-2011 with recent updates for 2012 and 2013. The monthly (0.5o) and 3 hourly (1ox1.25o) fluxes are available at the CMS and NACP websites (http://nacp-files.nacarbon.org/nacp-kawa-01/?C=M;O=D). Model output also includes biomass and detrital carbon pools. The impacts of uncertainties in model parameters on fluxes and carbon pools are estimated by selecting values from probability distribution functions for key parameter and propagating these in Monte Carlo ensemble simulations. Our CMS products are used by a number of other CMS projects (e.g. Bowman, Ott, French).

Model products are evaluated against atmospheric CO2 observations, atmospheric inversion fluxes, eddy covariance fluxes, and independent estimates of biomass. Comparisons with atmospheric CO2 measurements (flask, continuous, TCCON, GOSAT) show the model is good at matching the phase and amplitude of the measured CO2 seasonal cycle but is less successful at capturing interannual variability. The baseline model overestimates the atmospheric CO2 growth rate because it lacks representation of long term land sink mechanisms (e.g. CO2 fertilization). We have also developed model setups that are forced to reproduce the magnitude of the observed global land sink. Baseline modeled above ground biomass is within the uncertainties of independent estimates though biased low in northern latitudes. These and future evaluations provide bases for improving the model.

Associated Project(s): 

• Collatz (CMS 2013)