A road cut in the Bottaccione Gorge in Gubbio, Italy, exposes limestones deposited in late Cretaceous and early Teriary separated by an 1 cm layer, which shows a boundary in the rocks, which proves something serious in the geophysical history of the planet happened at that time, as Dr. Gisler started out with his presentation. The crater at Chicxulub in Mexico leads to further evidence for a tsunami focused search in the Gulf of Mexico.

According to Dr. Gisler, the Yucatan is a particularly bad spot for an impact. The Yucatan was submerged on the continental shelf in the late Cretaceous era. Layers of shallow water calcite and anhydrite from the fossilized coral reef appear above the granite curst and mantle. The upper target materials are volatile at impact pressures and

they are excavated readily, the speaker explained. They easily dissociate into calcium, sulfur, dioxide carbon, dioxide water, etc. Many of these chemicals persist a long time in the atmosphere.

Dr. Gisler's team performed a whole series of simulations using the SAGE code in order to visualize the effects of the asteroid impact at Chicxulub. The code was a joint development between SAIC and Los Alamos. It was originally used for waves produced by large underwater explosions. Currently, it is also applied for a wide range of projects in astrophysics and geophysics. SAGE functions using Eulerian with cell-by-cell adaptive mesh refinement. The mesh is re-allocated to processors on each adaption for lead leveling. The MPI offers good scaling up to 8192 processors, as the speaker stated.

The 3D calculations of Chicxulub asteroid impact start when the asteroid is located 40 km above the Earth surface. The shock wave preheats a part of the atmosphere before the actual impact. The water and carbonates are vaporized explosively at the moment of impact. The shock descends through the crust into the mantle, and is partially reflected. The explosive vaporization excavates the crater in a nearly symmetrical way. The excavation efficiency and distribution of ejecta depend on the impact angle, as the speaker showed in his simulation movies.

The role of the simulations in studying this event is multiple. If the simulation is correct, it can show what might happen when a certain projectile hits a certain target. It also can help relate certain observables. The parameter studies with a well validated code can be used to set limits on the parameters of a real event. At best, local simulations can estimate global effects, according to the speaker.

The improvements in SAGE include a better hydro and a better strength modelling, Dr. Gisler assured. The new runs also have a mixed water/calcium layer. The old runs were performed on LANLASC Q and the new runs on Pink. There are hundreds of millions of cells involved in each calculation.

Why has it taken so long to perform this kind of simulations? Dr. Gisler told the conference participants that the

computational resources for this sort of complex problems are still scarce and fiercely competed for. The clusters unfortunately are still less reliable than desired. There are node dropouts and problems with the interconnect.

There is rather low bandwidth to parallel disks and sometimes segment dropouts next to bit errors. Not to mention the inconsistencies between architectures.

In addition the data archiving and retrieval are painful at best whereas the visualization en masse is still not practical. To do physics Dr. Gisler has to search through immense 4D data hypercubes in order to make guesses as to what kinds of views will give him the best insights. The 3D visualization facilities are rare and not well connected to the data archive. Unluckily, the software or interactive 3D visualization is still clumsy and slow.

Dr. Gisler's simulations showed that steeper impacts eventually make larger craters. In addition, all runs tend toward symmetric craters. The internal energy produced in the target dominates the kinetic energy for all trajectories. The excavattion and mobilization of the target is greater at higher angles. The tracer particles' ejection in the 15 degrees impact is superficial whereas at 45 degrees, the ejection spreads more symmetrically.

The simulations give an insight in what could have been the orginial impact angle. Dr. Gisler already made clear that the crater size is larger for steeper impacts and that the excavation of granite is greater for steeper impacts.

The ejecta distribution is more asymmetric for shallower impacts and the thermal energy injected into the troposphere is greater for shallower impacts. Given these results, Dr. Gisler guesses that the original would not have been much shallower than 45 degrees.

The speaker regretted that geophysicists are still far short of routine high-resolution 3D computing for complex scientific problems. Computing resources are scarce and not always reliable. The insufficiency of bandwidths to

both scratch and archive the data is also problematic. Still, Dr. Gisler has done a great job, according to the audience's enthusiasm.