Inter-annual variability of carbon budget components in a cool-temperate deciduous forest in Japan (Takayama, AsiaFlux)

Net ecosystem production NEP, gross primary production, and total ecosystem respiration are estimated by a micrometeorological method for ten years at a cool-temperate deciduous forest in central Japan, one of the AsiaFlux forest sites. The CO2 flux has been estimated by an aerodynamic method since 1993 using the vertical gradient of CO2 concentration and the diffusion coefficient, and by the eddy covariance method since 1998. A ten-year average of annual NEP from 1994 to 2003 is estimated to be 240 +/- 90 gC m(-2) year(-1) (mean +/- SD). The year-to-year change in the annual NEP is primarily affected by the air temperature before and at beginning of the growing period (the mean temperature from April to June, T-AMJ), and by the solar radiation in summer (the mean solar radiation from June to July, S-JJ), and is functionally expressed as NEP = 51.4 T-AMJ + 1.8 S-JJ - 612.0 (r = 0.76, p < 0.05). Inter-annual variations of T-AMJ and NEP show a clear periodicity of about 4-5 years, and is coincident with the occurrences of El Nino events during the last decade. The annual NEP is compared with that obtained from the biometric method, in order to highlight the source and magnitude of uncertainties in both methods. The NEP estimated by the micrometeorological method is 200 gC m(-2) year(-1) higher than that by the biometric method (40 gC m(-2) year(-1)). A large uncertainty in the biometric method is related to the estimation of carbon flow in the soil. The uncertainty in the micrometeorological method is about the same as the difference in the NEP values obtained by the two methods. Further studies are clearly necessary to improve the estimation of carbon budget components and to reduce the uncertainties in both methods.