| This thesis presents an idealized theoretical and numerical study of tsunami-induced atmospheric gravity waves. These waves propagate rapidly from the ocean surface up to the ionosphere, where they are well known to leave a detectable fingerprint in air-glow patterns and other remote sensing observables. Accurate modeling of the wave propagation is a prerequisite for being able to detect and decode this transient observational fingerprint by remote sensing methods.;The thesis starts from investigating the effect of non-uniform stratification, which takes into consideration the partial back-reflection of waves in the vertical. The key factor is to properly select the radiation condition to get rid of ill-posedness and unphysical waves. This is illustrated by comparison with two standard approaches of ray tracing, and using ground-level Dirichlet-to-Neumann map in a simple model for stationary gravity waves.;The thesis develops the time-resolving model by formulating the initial-value problem for linear waves forced by an idealized tsunami at the lower boundary and then employing a semi-analytic Fourier--Laplace method to solve it. This approach allows computation of the detailed time evolution of the waves while ensuring that the correct radiation condition in the vertical is satisfied at all times, a non-trivial matter for these transient waves.;In the thesis, the predictions of an anelastic model with that of a fully compressible model are compared in order to discern the importance of acoustic effects. The results demonstrate that back-reflection at the tropopause is a significant factor for the structure of these waves, and that the earliest observable signal in the ionosphere is in fact a fast acoustic precursor wave generated by the nearly impulsive formation of the tsunami itself. |
