89th American Meteorological Society Annual Meeting

Sunday, 11 January 2009
Investigating the ability of CLM-CN 3.5 to accurately simulate vegetation density
Phoenix Convention Center
Eowyn C. Connolly-Brown, Cornell University, Ithaca, NY
Terrestrial vegetation engages in important feedbacks with the carbon cycle and the climate system. The National Center for Atmospheric Research's (NCAR) Community Land Model version 3.5 coupled to carbon/nitrogen cycling (CLM-CN 3.5) was compared to satellite observations of Normalized Difference Vegetation Index (NDVI) to investigate its ability to simulate temporal and spatial variability in vegetation density. High-resolution AVHRR satellite-derived NDVI data obtained from the University of Maryland Global Inventory Modeling and Mapping Studies (GIMMS) project was regridded and temporally averaged to match the monthly, 2.5 by 1.875 resolution of CLM-CN 3.5 NDVI values. In CLM-CN3.5 the NDVI was calculated from near-infrared and visible spectrum reflectances. Gridded monthly and annual climatological time series were produced to determine model accuracy at different spatial and temporal scales. Interannual variability simulation accuracy was examined using NDVI-anomaly plots for historically-documented droughts. Multidecadal trends in vegetation were investigated using time series of CLM-CN 3.5 NDVI data for 1870-2004 to see if the model captured the recent greening trend documented by previous research.

Our findings indicate CLM-CN 3.5 tended to overestimate NDVI when values were less than 0.15, and underestimated otherwise. Modeled NDVI peaked a month sooner than observed values. Regionally, CLM-CN 3.5 did not capture the monsoon-induced vegetation response in the Sahel. Drought-vegetation response simulation varied in accuracy. CLM-CN 3.5 captured recent (1960-2004) multidecadal greening trends with a high degree of certainty (R=0.96). These results will form a foundation for future research to determine the causes of error and to address biases in future model development.

Supplementary URL: