OPTIMIZING AN INVERSE METHOD TO DEDUCE TIME-VARYING EMISSIONS OF TRACE GASES

In previous work, an inverse method based on the Kalman filter was use d to deduce regional emissions for chlorofluorocarbons (CFC-11) in a g lobal chemical transport model [Hartley, 1992; Hartley and Prinn, 1993 ]. CFC-11 has reasonably constant emissions over the years we addresse d; however, most trace gases with poorly constrained global budgets (i .e. CO2, CH4, and N2O) have seasonally varying sources and sinks. The goal of this work is to explore various adaptations that exist in Kalm an filter theory and to identify the optimum method for deducing time- varying sources/sinks. To make this study feasible in both computer ti me and cost, we utilize a simplified atmospheric chemical transport mo del to investigate the methodology. We test many available adaptations in Kalman filter theory and conclude that the most accurate method is to use an adaptive-iterative approach [Young, 1984; Sastri, 1985; Bel lgardt et al., 1986].