MODELING FRAMEWORK TO EVALUATE SAMPLING STRATEGIES AND ESTIMATE SURFACE EMISSIONS OF TRACE GASES IN MESOSCALE

摘要

The Bayesian inversion method is commonly used to estimate surface emissions of CO_2 and other trace gases in a global scale (Kasibhatla et al., 2000). The application of this approach to limited domains used in mesoscale or regional scale modeling is more challenging. In addition to estimation of the surface tracer flux it is necessary to evaluate unknown fluxes through model lateral boundaries. The inflow flux of CO_2 may be several orders of magnitude larger than the CO_2 flux from the surface of regional modeling domain. In addition, the CO_2 flux from the land surface shows a strong diurnal cycle related to the uptake of CO_2 by photosynthesizing plants and the release of CO_2 by microbial decomposition in the soil. The proposed modeling framework is based on a Lagrangian Particle Dispersion Model (LPDM) linked to CSU RAMS (Regional Atmospheric Modeling System) (Uliasz, 2000). The LPDM is used in a receptor-oriented mode (tracing particles backward in time) to derive influence functions for each concentration sample. The influence function provides information on potential contributions from surface sources and inflow fluxes through the modeling domain boundaries into tracer concentration sampled at the receptor. Then the Bayesian inversion technique is applied in an attempt to estimate unknown surface emissions.