By John Austin
I worked on stratospheric modeling as part of the UK Met Office theory contribution to AAOE. In the mid 1980s, computers were much slower than today – about a factor of 10,000 – and in those days the Met Office stratospheric chemical model was a simple 1-D model. In other words, you can now run the chemistry in a whole climate model for about the same computer time as our original model. To maximise performance it was even written in IBM Assembler! I had no personal expertise in that low level language, but fortunately one of our group members (Robin Pallister) did. He proceeded to convert it laboriously into Fortran to increase our flexibility. This was a fortunate convergence of skills, because a few years before AAOE, the group leader Adrian Tuck came up with the idea of running a chemical model along an air parcel trajectory, thereby making the computations both realistic and manageable. I was tasked with setting up this infrastructure. With this tool then in place, we had a "solution looking for a problem" and the AAOE data provided an ideal such problem. During the AAOE itself, I had the task of coordinating the Met Office theory effort from the relative comforts of the Met Office headquarters building. This was almost as hazardous as going to Antarctica itself, as the building was later declared unfit due to concrete cancer!
One of our major studies (Jones et al., 1989) published in the JGR special issue made a careful comparison between AAOE observations and our model simulations. To reproduce observations we needed to include heterogeneous reactions occurring on polar stratospheric clouds. AAOE observations as well as satellite data demonstrated the existence of these clouds but chemical reaction rates on their surfaces were quite unknown. Some of the reactions were even a little controversial. More controversial still was the idea that you could repeat a single trajectory over and over to obtain a seasonal integration of our chemical model. Although this was also published in the JGR special issue (Austin et al., 1989), the long trajectory duration was a sticking point with the reviewers. Perhaps a little unwittingly, we had already been thinking of the nature of the Antarctic vortex. In other words, our calculations made the conceptual assumption that the vortex was a "reaction vessel" rather than a "flowing processor", although our arguments were certainly not expressed in those terms. Rather, if the small changes observed (and simulated) along a short atmospheric trajectory had occurred, can we be sure that such processes continuing for the whole season would be sufficient to destroy enough ozone? Conceptually this was indeed demonstrated to be the case.
The tools used for AAOE developed apace as computer power improved. In time, I also changed our chemical model into a family scheme to provide improved computational performance. This model provided the basis for the Met Office coupled chemistry climate model and the GFDL model at a later time (subsequently superceded by an all-atmosphere chemical model: such has been the pace of the increase in computational speed). Simulations of Antarctic ozone continue to be made to identify for example the role of different dynamics in different years so data from AAOE has provided a lasting legacy for theoretical work.