Variations in Continental Terrestrial Primary Production Evapotranspiration and

Variations in Continental Terrestrial Primary Production, Evapotranspiration and Disturbance Faith Ann Heinsch, Maosheng Zhao, Qiaozhen Mu, David Mildrexler, and Steven W. Running Numerical Terradynamic Simulation Group, The University of Montana, Missoula, MT 59812 Email contact: faithann@ntsg. umt. edu Web site: www. ntsg. umt. edu I. MODIS Gross (GPP) and Net (NPP) Primary Production U. S. NACP III. High-Resolution MODIS GPP/NPP Ø One of the largest errors associated with the MODIS GPP/NPP algorithm derives from the use of coarse resolution (1. 00 1. 25 ) meteorology data, which accounts for 28 ± 48% of the error at 15 Ameri. Flux tower sites (Heinsch et al. , 2006). Tower Meteorology GMAO Meteorology Annual Correlation Coefficient 0. 792 ± 0. 206 0. 855 ± 0. 175 -2% 19% Annual Relative Error Ø There are sufficient surface weather stations in North America to allow 1 -km 2 resolution meteorology to be used with the MODIS algorithm. Ø Surface Gridded Observation System (SOGS) calculates daily meteorology for the U. S. at any resolution (Jolly et al. , 2005) Ø Uses the NOAA National Climatic Data Center’s “Global Surface Summary of the Day” Ø Total average annual NPP for North America is 394 ± 271 g. C/m 2/y Ø 8 -day and annual MODIS GPP results compare favorably with flux tower estimates Ø Interannual variability is captured as years with substantial droughts (e. g. , 2002) show decreasing NPP in response to the dry conditions Ø There is a general increase in NPP across much of the U. S. , with decreasing NPP in southern Mexico and across the interior of Canada II. MODIS-Based Evapotranspiration (ET) Ø With access to similar data from Canada and Mexico, it is possible to provide high resolution estimates of GPP, NPP, and ET for all of North America. IV. MODIS-Based Continental Disturbance Index (DI) r = 0. 873 r = 0. 861 RMSE (W m-2): Tower Meteorology GMAO Meteorology Daily 36. 1 38. 5 8 -Day 26. 5 28. 8 Ø The timing, location and magnitude of major disturbance events are major uncertainties in carbon cycle science. Ø We developed a simple, fast, automated disturbance detection algorithm (Mildrexler et al, 2007). Ø Annual maximum compositing of Land Surface Temperature (LST) and the Enhanced Vegetation Index (EVI), tracking positive and negative changes in land surface energy partitioning Ø Total average annual ET for North America is 239 ± 178 mm/y (max = 1092 mm/y) Ø 8 -day and annual MODIS-based ET compare favorably with flux tower data Ø Algorithm captures inter-annual variability (e. g. , droughts) and seasonality across the continent Ø Areas with high ET correspond to areas of high precipitation and high GPP Ø ET peaks in the summer, with higher ET east of the Rocky Mountains and along the Pacific Coast. Ø The boreal forest shows up clearly as an actively transpiriing system. We thank the Ameri. Flux investigators of these participating sites for the generous use of their data in these validation exercises. Ø Validation was performed, resulting in obvious correspondence between 2003 DI results and both MODIS active fire detection data and fire perimeter maps for the wildfires near Missoula, Montana (a, b) and southern California (c, d). Ø We present the continuous DI results for the western U. S. (2003 -2004) using 1 standard deviation (0. 32) from the mean (1. 0) to define the range of natural variability.