Master's thesis presentation. Florian is advised by Mario Wille, and Prof. Dr. Michael Bader.
Previous talks at the SCCS Colloquium
Florian Wunderlich: Simulation of Gravity Waves in the Atmosphere using the PDE Engine ExaHyPE 2
SCCS Colloquium |
The science of atmospheric gravity waves, although going on for many decades, has recently produced promising results in tracking tsunami waves over vast stretches of ocean by using upper atmosphere chemistry. Other recent publications point out possible resonances of the atmosphere with the Earth itself during volcanic eruptions. This makes them a highly interesting research field for computational geophysics.
This work implements an atmospheric model, which supports the generation and propagation of gravity waves in ExaHyPE 2. Since the atmosphere is a layered structure, well-balanced Riemann solvers are necessary to capture these initial steady states. To this end, an f-wave solver for the Euler equations with gravity has been implemented. A second order accurate MUSCL-Hancock scheme was also implemented but the necessary well-balancedness could not be verified. First order tests of the atmospheric model show promising results although numerical issues in the uppermost layers of the atmosphere and too strong velocities still persist.
The coupling to a diffusive model changed the artifacts that appeared but did not ultimately fix the problems. However, reducing the maximum altitude down to 180km was found to be successful in getting rid of artifacts, which leads to the conclusion that the atmospheric background profile in conjunction with the numerics is to be blamed.
In addition to the atmospheric model, a chemical airglow model has also been implemented. The correctness of the production rates and chemical profiles has been verified. The model is also verified in isolation to maintain chemical steady states for hundreds of seconds, which fits the simulation times of gravity waves. Coupling it to the atmospheric model, both the OI and the OH airglow reactions break down and produce errors. Since this does not happen during isolated simulations, coupling effects are the root cause and further investigation into these is necessary. Ignoring the airglow reactions, advection of major species by excited gravity waves shows good results, proving that the coupling mechanism works not only for the diffusive model, but also for the chemical model.
"""