HR: 09:00h
AN: A31E-04    [Abstracts]
TI: Errors in Tracer-Transport Modeling: Can CTMs Find the Right Answer?
AU: * Prather, M J
EM: mprather@uci.edu
AF: Earth System Science Dept, 3329 Croul Hall UC Irvine, Irvine, CA 92697-3100 United States
AU: Zhu, X
EM: xzhu@uci.edu
AF: Earth System Science Dept, 3329 Croul Hall UC Irvine, Irvine, CA 92697-3100 United States
AB: In working with the NASA Global Modeling Initiative's framework for chemistry-transport models (CTMs), it became clear that different algorithms for tracer transport produce surprisingly different results. This difference can only be interpreted as tracer-transport error; it contributes significantly to the overall CTM error in simulating atmospheric chemistry; and it is comparable to the likely errors in meteorological fields, emissions, photochemistry, and sub-grid processes. In simulating the atmospheric distribution of fossil-fuel CO2, it was found that the UCI CTM (Prather second-order moments) and the GMI core CTM (Lin and Rood algorithm) showed differences in (i) stratospheric age-of-air of more than one year and (ii) tropospheric gradients that would alter the inferred CO2 sources/sinks. A new tactic in establishing the correct answer for the tracer-transport problem, and thus in defining CTM error, is taken here: doubling-to-convergence. The premise of this work is that the numerical error in tracer transport (as opposed to the error due to incorrect wind fields, for example) is proportional in some manner to the grid size, i.e., there are no systematic errors in the UCI and GMI CTMs in either coding or basic algorithms. By successive halving of the grid size (1, 1/2, 1/4, 1/8), we show that the CTMs converge. Note that the computational burden for the 1/8-grid case increases by a factor of 4096. In this simple case, we can quantify the transport error in both CTMs, and further we can derive an Aitken acceleration factor that allows a single doubling (1 and 1/2) to be used to produce a best answer with error estimates.
DE: 3337 Numerical modeling and data assimilation
DE: 3230 Numerical solutions
DE: 0341 Middle atmosphere--constituent transport and chemistry (3334)
DE: 0368 Troposphere--constituent transport and chemistry
SC: Atmospheric Sciences [A]
MN: 2004 AGU Fall Meeting