HR: 17:50h
AN: B44B-08    [Abstracts]
TI: An Inter-comparison of Vegetation Greenness From Satellite Observations and a Terrestrial Ecosystem Model
AU: * Twine, T E
EM: twine@atmos.uiuc.edu
AF: University of Wisconsin-Madison, Center for Sustainability and the Global Environment 1710 University Ave , Madison, WI 53726
AU: Kucharik, C J
EM: kucharik@wisc.edu
AF: University of Wisconsin-Madison, Center for Sustainability and the Global Environment 1710 University Ave , Madison, WI 53726
AB: Terrestrial ecosystem models simulate the structure and functioning of vegetation as well as the exchanges of energy, water, and nutrients between components of the land surface and the atmosphere. While these models use numerical methods that are based on a wealth of observations, the accuracy of a model in simulating ecosystem processes at the regional scale is difficult to test because evaluation has traditionally relied on {\it in situ} measurements made at point locations (on the order of several m$^{2}$ in area). Daily satellite observations may provide a means for better model evaluation through the sensing of ecosystems at regional to global scales; however, there are several challenges to this method of evaluation. Satellite measurements may suffer from signal corruption from the earth's atmosphere, sensor and solar geometry issues, and sensor calibration problems. In addition, most of the quantities of interest in model evaluation must be derived from the reflectances detected by the sensor, which increases the uncertainty in these variables, and are usually given to the community after downgrading the daily values to monthly average values. In this study, we compare twenty years of Pathfinder Advanced Very High Resolution Radiometer (AVHRR) monthly-averaged measurements of the Normalized Difference Vegetation Index (NDVI), the fraction of photosynthetically active radiation absorbed by the vegetation canopy (FPAR), and the leaf area index (LAI) with output from the Integrated Biosphere Simulator (IBIS) over grasslands, croplands, and forests within the United States. Because two variables, FPAR and LAI (secondary, or derived quantities), have different relationships with NDVI (primary quantity), this three-variable evaluation may provide a method of assessing uncertainty in both simulated and observed (derived) quantities. Results show that IBIS captures the observed seasonality and magnitude of NDVI over all biomes, although FPAR and LAI are underestimated in croplands and forests compared with the AVHRR observations. Recent satellite observations have shown an apparent `greening' of certain ecosystems through an increasing trend in the derived quantity of net primary productivity (NPP). Results from the 20-year IBIS simulation appear to capture the satellite-observed increasing trend in NPP over some North American biomes. Because IBIS is forced with observed climate, these results may act to validate the satellite observations.
DE: 1615 Biogeochemical processes (4805)
DE: 1640 Remote sensing
DE: 0315 Biosphere/atmosphere interactions
DE: 0400 Biogeosciences
SC: Biogeosciences [B]
MN: 2004 AGU Fall Meeting