Monitoring Effects of Interannual Variation in Climate and Fire Regime on Regional Net Ecosystem Production with Remote Sensing and Modeling D.P. Turner 1, W.D. Ritts 1, J. Styles 1, Z. Yang 1 W.B. Cohen 2, B.E. Law 1, P. Thornton 3, M. Falk 4 1 Oregon State University 2 USDA Forest Service 3 National Center for Atmospheric Research 4 University of California, Berkeley
Western Oregon Study Area
Chronosequences: NEP by Age Class NEP Uncertainty ~25% NEP = (NPP A + NPP B ) (R hSoil +R hCWD +R hFWD ) NEP = NPP A + TBCA – R s (R hCWD + R hFWD ) wet dry Campbell et al. 2004
Biome-BGC Simulation Conifer forest in West Cascades Turner et al. 2003
Prognostic Modeling Approach to Bottom-up Scaling
Land base carbon budgets for two representative (100 km 2 ) study areas. (Units are gC/m 2 /yr) __________________________________________________________ Study Area A B C Net Ecosystem Harvest Land Production Removals C Pool (A+B) Coast Range West Cascades ___________________________________________________________ Turner et al. 2004
Land Base Carbon Budget Western Oregon Forests ( ) _________________5 yr mean_______________ Total NEP 13.8 TgC/yr Harvest Removals -5.5 Products (net) 1.4 Fire -0.1 Net 9.6 TgC/yr ________________________________________ Law et al. 2005
Application to larger domain Investigation of interannual variability in NEP Daily FPAR from satellite data Simpler process model (base rates for light use efficiency, Ra, Rh) No spin-ups Same distributed climate data Same land cover from satellite data (aggregated) Same stand age from satellite data (aggregated) Diagnostic Modeling Approach
NPP derived from USFS Inventory plot data Van Tuyl et al. 2005
How to include information on the disturbance regime? Metamodeling Approach GPP = ↓ PAR * FPAR * e g * S sa GPP = gross primary production ↓ PAR = incoming PAR FPAR = fraction PAR absorbed e g = light use efficiency S sa = stand age factor (0-1), output from Biome-BGC model Stand Age Factor for GPP Biome-BGC Model Run
How to include information on the disturbance regime? Metamodeling Approach R h = f (R h-base, FPAR, Tsoil, SW, SA) R h = heterotrophic respiration R h-base = base rate of heterotrophic respiration FPAR = Fraction PAR absorbed Tsoil = soil temperature SW = soil water content SA = stand age factor, output from Biome-BGC model Stand Age Factor for R h Biome-BGC Model Run
MODIS FPAR Spatial Resolution = 250m - 1 km Temporal Resolution = 8 day
Diagnostic Model (“Fusion”) Parameter Optimization Daily time step 1.Daily GPP parameters optimized with tower GPP (or Biome-BGC GPP) 2.Daily R a parameters optimized by reference to measured NPP (or Biome-BGC NPP) 3.Daily R h parameters optimzed with tower NEE (or Biome-BGC NEE)
Fusion daily NEP (line) compared to reference NEP (circles) At the Klamath Mountains ecozone conifer optimization site.
Bars are mean NPP for FIA plots from Van Tuyl et al. 2005
Bars are mean ecoregion NEP from Law et al. (2004)
Validation Boundary Layer BudgetDiagnostic NPP/NEP Model Top-down Flux Bottom-up Flux
Boundary layer budget footprint Monthly mean from the STILT model Weighting (Courtesy of M. Goeckede, OSU)
Conclusions Land-based carbon sinks significantly offset fossil carbon emissions in Oregon Post-fire increases in heterotrophic respiration reduce the regional carbon sink Interannual variation in climate can substantially modify regional NEP