According to the Protocol on Environmental Protection to the Antarctic Treaty seismic surveys in the Southern Ocean south of $60\deg$S are exclusively dedicated to academic research. The seismic surveys conducted by the Alfred-Wegener-Institute for Polar and Marine Research, Bremerhaven, Germany during the last 20 years focussed on two areas: The Wedell Sea and the Amundsen/Bellinghausen Sea. Histograms of the Julian days and water depths covered by these surveys indicate that maximum activities occurred from January to March, and most lines were collected either in shallow waters of 200 to 600 m depth or in deep waters of 3000 to 4000 m depth. To assess the potential risk of future seismic research surveys on marine mammal populations an acoustic wave propagation modeling study is conducted for the Weddell and the Amundsen/Bellinghausen Sea. A viscoelastic 2.5D finite-difference code is used, which allows to simulate the spherical amplitude decay of point sources correctly. A sinusoidal wavelet of 50 Hz dominant frequency serves as source wavelet. Based on CTD measurements, sediment core samplings and sediment echosounder recordings two horizontally-layered, range-independent generic input models are established for the Wedell and the Amundsen/Bellinghausen Sea, one for shallow (400 m) and one for deep water (3000 m). They indicate that the vertical structure of the water masses is characterized by a 100 m thick, cold, low sound velocity layer ($\sim$1440 - 1450 m/s), centered in 100 m depth. In the austral summer it is overlain by a warmer, 50 m thick surface layer with slightly higher sound velocities ($\sim$1447 - 1453 m/s). Beneath the low-velocity layer sound velocities increase rapidly to $\sim$1450 - 1460 m/s in 200 m depth, and smoothly to $\sim$1530 m/s in 4700 m depth. The sea floor is mainly covered with soft fine-grained clayey or silty sediments, so that P- and S-wave velocities of 1600 and 200 m/s and a wet bulk density of 1450 kg/m${3}$ are assumed. In a first step the acoustic impact of one seismic line of 10 km length is computed for the two generic models, assuming a shot interval of 15 s and a ship speed of 5 kn. The acoustic impact is determined by running the finite-difference scheme once, shifting the resulting wavefields in space and time according to the movement of the ship and the shot interval, and summing-up the appropriate sound exposure levels derived from the synthetic seismograms. As results, time-dependent contour maps of the cumulative sound exposure levels are derived. From these contour maps time-dependent exposure histories of the received sound exposure levels are extracted for animals staying at fixed depth and range positions along the seismic line.