Inversely Estimating the Vertical Profile of the Soil CO.sub.2 Production Rate in a Deciduous Broadleaf Forest Using a Particle Filtering Method

Carbon dioxide (CO.sub.2) efflux from the soil surface, which is a major source of CO.sub.2 from terrestrial ecosystems, represents the total CO.sub.2 production at all soil depths. Although many studies have estimated the vertical profile of the CO.sub.2 production rate, one of the difficulties in estimating the vertical profile is measuring diffusion coefficients of CO.sub.2 at all soil depths in a nondestructive manner. In this study, we estimated the temporal variation in the vertical profile of the CO.sub.2 production rate using a data assimilation method, the particle filtering method, in which the diffusion coefficients of CO.sub.2 were simultaneously estimated. The CO.sub.2 concentrations at several soil depths and CO.sub.2 efflux from the soil surface (only during the snow-free period) were measured at two points in a broadleaf forest in Japan, and the data were assimilated into a simple model including a diffusion equation. We found that there were large variations in the pattern of the vertical profile of the CO.sub.2 production rate between experiment sites: the peak CO.sub.2 production rate was at soil depths around 10 cm during the snow-free period at one site, but the peak was at the soil surface at the other site. Using this method to estimate the CO.sub.2 production rate during snow-cover periods allowed us to estimate CO.sub.2 efflux during that period as well. We estimated that the CO.sub.2 efflux during the snow-cover period (about half the year) accounted for around 13% of the annual CO.sub.2 efflux at this site. Although the method proposed in this study does not ensure the validity of the estimated diffusion coefficients and CO.sub.2 production rates, the method enables us to more closely approach the "actual" values by decreasing the variance of the posterior distribution of the values.