During the cloud dynamics and chemistry field experiment CLEOPATRA in the summer of 1992 in southern Germany, the Deutsche Forschungsanstalt f¿r Luft- und Raumfahrt (DLR) (German Aerospace Research Establishment) research aircraft Falcon traversed four times the anvil of a severe, isolated thunderstorm. The first two traverses were at 8 km altitude and close to the anvil cloud base, while the second two traverses were at 10 km. During the 8-km traverse, measured ozone mixing ratios dropped by 13 parts per billion by volume (ppbv) from the ambient cloud free environment to the anvil cloud, while water vapor increased by 0.3 g kg-1. At the 10-km traverses, ozone dropped by 25 ppbv, while water vapor increased by 0.18 g kg-1. Three-dimensional numerical thunderstorm simulations were performed to understand the cause of these changes. The simulations included the transport of two chemical inert tracers. Ozone was assumed to be one of them. The initial ozone profile was composed from an ozone routine sounding and the in situ Falcon measurements prior to the thunderstorm development. The second tracer is typical for a surface released pollutant with a nonzero, constant value in the boundary layer but zero above it.

The redistribution of both tracers by the storm is calculated and compared with the observations. For the anvil penetration at 10 km, the calculated difference in ozone mixing ratios is 21 ppbv, while for water vapor an increase of 0.25 g kg-1 was found, in good agreement with the observations. To validate the model results, the radar reflectivity was calculated from simulated fields of cloud water, rain, graupel, hail, and snow and ice crystals and compared with observed values. With respect to maximum reflectivity values and spatial scales, again, excellent agreement was achieved. It is concluded that the rapid transport from the boundary layer directly into the anvil level is the most likely cause of the observed ozone decrease and water vapor increase. Entrainment of ozone-rich environmental air into the anvil cloud occurred but left a protected core with undiluted boundary layer air in the anvil cloud even at a distance of 120 km from the main updraft. Processes such as production of O3 by electrical discharges, chemical reactions of ozone with boundary layer-released or lightning-produced nitrogen compounds, scavenging by hydrometeors, and heterogeneous reactions at the surface of ice crystals may occur, but on the timescale of 0.5--1 hour seem to have a negligible influence on the observed ozone drop. ¿ American Geophysical Union 1995