The C budget of Japan: Ecosystem Model (TsuBiMo) Y. YAMAGATA and G. ALEXANDROV Climate Change Research Project, National Institute for Environmental Studies, JAPAN Regional Carbon Budgets: from Methodologies to Quantification Beijing, China, November 2004
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Structure of our forest ecosystem carbon balance model (TsuBiMo) Gross Primary Production Autotrophic respiration Heterotrophic respiration Litterfall
TsuBiMo: Litter-fall compartment very resistant fraction non-resistant fraction resistant fraction Herbaceous fraction of litter- fall Woody fraction of litter-fall The carbon pools of litter
Regional Carbon Budget Assessment using TsuBiMo CO2 fluxes (JapanFlux) Yield tables (MAFF) GPP (P g ) NPP (P n ) NEP (P E ) NBP Changes in total carbon stock of the managed forests in Japan Carbon stock changes in the pool of non- living organic matter Satellite images (JAXA) Carbon stock changes in the pool of living organic matter Accumulation in biomass Age distribution (MAFF)
Input data: grids of 1km resolution Monthly temperature Monthly precipitation Monthly solar radiation Forest age structure Managed forest 15 age classes of 5-years Natural forests 4 age classes of 20-years
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
2003/10/ /11/6 Pictures From Dr. Nishida Takayama Flux Cite
Scheme of using CO2 flux data for calibrating productivity model TsuBiMo function for GPP : Observation P max light-saturated photosynthesis K light attenuation coefficient ß light-use efficiency f PAR fraction of absorbed PAR Re (night time flux) NEE (day time flux) - GPP
Calibrated Productivity Model (Takayama Flux Data )
CO2 fluxes v.s. Productivity Model BLUE – observed fluxes YELLOW- 28-days moving average RED – model estimates The model could successfully replicate the forest ecosystem CO 2 flux responses to climate variations
Agreement with World Biometric Data of Calibrated Productivity Model Blue – Local Calibration Red – Global Calibration
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Yield Table data for Growth Function Age dependence of wood stock in m 3 /ha (yield tables) The estimates of conversion coefficient specified by age TsuBiMo function for biomass growth : Yield table data The model parameters depends on species and site fertility, however the data from ecological studies are not sufficient. -> We use yield table data for filling the gap.
Yield Tables v.s. Model Estimates Sugi, North Kinki (Fukuda et al., 2004) Line shows the values produced by the function above for b 2 = ; P n b 1 = assumed Dots show the values derived from the yield table above by using the assumed conversion coefficient warm-temperate cool-temperate subtropical
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Carbon Stock in Japanese Forest Very Low Low Average High Stock, tC/ha Stock classes GtC Carbon Stock is estimated using TsuBiMo
NPP of Japanese Forests NPP is estimated using TsuBiMo from vegetation period, light intensity, temperature and precipitation Legend: Subalpine conifer forest zone Cool-temperate broadleaf forest zone Warm-temperate broadleaf forest zone Subtropical forest zone NPP, g C/m 2 /yr Interpretation in terms of vegetation zones
Forest Age and Carbon Stock in Japan Area of the k-th age class (s k ) Forest age data Carbon stock estimate Carbon stock changes with age
NEP estimates in Japanese forest Alexandrov, G.A., Yamagata, Y., Net Biome Production of managed forests in Japan. Science in China, 45 (Supp): Scenario: harvest age lifted up to 70 years
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Model v.s. Carbon Stock Data Black: total carbon stock after breakup Grey: carbon stock changes in the tree biomass Lines: Model estimates (Kawaguchi and Yoda,1986) Beech forest in Japan
Comparison of NEP estimate (tentative) Ecological model Forestry inventory ( tC/ha ・ yr ) 17.9 MtC/yr21.0 MtC/yr Artificial forest only
Outline Structure of model Calibration of model using flux data Calibration using inventory data Estimation of carbon stock, NPP and NEP Comparison with ecological measurement and inventory based estimate Accuracy assessment of R/S (FPAR) data use Conclusion and future direction
Accuracy assessment of R/S data use Yellow – MODIS Blue – Ground Productivity can be estimated by using FPAR as input to Model
Conclusion and future direction We have try to estimate the national level carbon budget using a process based ecosystem model (TsuBiMo) Calibration with flux and inventory data showed that the global model underestimate the productivity at managed forest test sites comparisons of national level estimate with inventory approach showed regional discrepancies but rather good total coincidence Model-data integration using different data sources including R/S need to be developed