Examination of model-estimated ecosystem respiration using flux measurements from a cool-temperate deciduous broad-leaved forest in central Japan

Reducing uncertainty in the emission of carbon dioxide (CO₂) from plants and microbes is critically important in determining carbon budgets. We examined properties of net ecosystem CO₂ exchange (NEE) derived from a processbased model that simulates an ecosystem carbon cycle, focusing on nighttime flux determined from ecosystem respiration and soil efflux. The model simulated autotrophic and heterotrophic respiration using semi-empirical ecophysiological parameterizations. In a cool-temperate deciduous broad-leaved forest in central Japan, simulation results from 1998 to 2005 were compared with measurement of the forest made using eddy-covariance and chamber methods. The model estimated annual ecosystem respiration as 1397 g Cm^-2 yr^-1, of which 67% was from the soil surface, with a clear seasonal cycle. Compared to flux observations, the model appropriately captured daytime NEE, but produced substantial differences from the observed nighttime NEE. The differences were evident under stable atmospheric conditions (at low friction velocity), implying a problem with the observations. With regard to soil-surface CO₂ efflux (soil respiration), the model estimation was consistent with chamber observations, except in winter periods with thick snow cover. We discuss whether the model is applicable for estimating ecosystem respiration rates, and what is required to improve the predictability of the model.