NCEAS Project 4400

Intercomparison of global scale ecological models and field data: EMDI II (Hosted by NCEAS)

  • Kathy A. Hibbard

ActivityDatesFurther Information
No activities scheduled at this time.

Understanding global-scale ecosystem responses to changing environmental conditions is important both as a scientific question and as the basis for making policy decisions. The confidence in regional models depends on how well the field data used to develop the model represent the region of interest, how well the environmental driving variables represent the region of interest, and how well regional model predictions agree with observed data for the region. To assess the accuracy of global model forecasts of terrestrial carbon cycling, the first Ecosystem Model-Data Intercomparison (EMDI) workshop was held in December 1999. We propose that the National Center for Ecological Analysis and Synthesis (NCEAS) co-sponsor with the Global Analysis, Interpretation and Modeling Task Force (GAIM) of the International Geosphere Biosphere Program the EMDI II Working Group to be held at Santa Barbara in mid-April 2001. This request describes the EMDI I progress, past NCEAS support, and outlines the EMDI II activities. The EMDI I workshop included 12 biogeochemical, satellite-driven, detailed process, and dynamic vegetation global model types. Extensive worldwide net primary productivity (NPP) data were assembled; model driver data, including vegetation, climate, and soils, were associated with each site; model simulations were performed; and the model-data differences analyzed. NPP and model driver data were compiled for 1200 study sites and for over 2100 0.5-degree grid cells. Initial results showed general agreement between model predictions of NPP and field measurements of NPP but with obvious differences that indicated areas for potential data and model improvement. Comparing the input data with an average NPP from an ensemble of model outputs provided a unique tool to improve NPP data, model driver data, and model processes. The workshop demonstrated that model-data intercomparison is an important new direction in model evaluation; but one that is an extraordinarily complex task. NCEAS has provided significant support for EMDI and NPP data compilation. The workshop to analyze the first EMDI was hosted by the GAIM office (see ), however, NCEAS provided support for a student intern to organize the data and an FTP site for distribution (see Much of the NPP data for EMDI came directly from the "Development of a Consistent Worldwide Net Primary Production (NPP) Database" funded by NCEAS (see Ten papers and four global data sets were generated from the series of three Working Groups (see This and the preliminary Web page describing EMDI ( - soon to become part of the ORNL DAAC publicly browsable Web pages) demonstrate how NCEAS-supported data synthesis/analysis activities have resulted in new ecological data that are available for the wider scientific community and that will be used for the proposed EMDI II Working Group. The EMDI II Working Group will include all initial models except one (model is no longer supported) and has expanded to include 5 additional global groups. The group now includes 16 models, primarily from Europe and North America, but one in China and one in Australia. The participating models will utilize data sets that have been improved through outlier analysis and perform new model runs for a set of approximately 4000 0.5� grid cells. Whereas the analysis in EMDI I was based on comparing ensemble values averaged from all 12 models strictly for total (above and belowground) NPP, EMDI II will compare model results with NPP data for individual models and also require models to supply above and belowground NPP estimates to assess how models allocate carbon in various biomes relative to the data. In addition, we plan to assemble multi-year NPP estimates for a variety of sites to compare to multi-year model estimates. In addition to the analysis of differences between models and data, EMDI will produce an enhanced multi-layered data collection, again that will be made available for additional use by the wider scientific community. The enhancements include associating climate, soils, NDVI, and vegetation characteristics with each of the NPP measurements, and performing an outlier analysis to address data quality concerns. We propose a co-funded workshop by IGBP/GAIM and NCEAS. Total estimated costs for 20 participants for 5 days is $25k, of which, GAIM will supply $15k. We are requesting additional support from NCEAS for $10k to cover additional costs. The proposed timetable is as follows: (1) Driver data is posted for a 50% 'blind' comparison of grid cells (as of 3 January) (2) NPP values for the 50% 'blind' cells will be provided to models that submit their results by 7 February (3) Driver data for additional 50% data cells posted 7 February (4) Model results are requested by 28 March (5) Post model results to UNH website for models to view by 4 April. For more information, contact: Kathy A. Hibbard phone: (603) 862-4255 IGBP International Carbon Cycle Project fax: (603) 862-2124 Climate Change Research Center email: University of New Hampshire Durham, NH 03824

TypeProducts of NCEAS Research
Journal Article Del Grosso, Stephen; Parton, William J.; Stohlgren, Thomas; Zheng, Daolan; Bachelet, Dominique; Prince, Stephen D.; Hibbard, Kathy A.; Olson, Richard J. 2008. Global potential net primary production predicted from vegetation class, precipitation, and temperature. Ecology. Vol: 89(8). Pages 2117-2126.
Report or White Paper Olson, Robert J.; Johnson, Kirk R.; Zheng, Daolan; Scurlock, Jonathan. 2001. Global and regional ecosystem modeling: Databases of model drivers and validation measurements. Environmental Sciences Division, Oak Ridge National Laboratory. Pages 84.