33 
a better understanding of tsunami modifications on the continental slope, some reference 
computations using baroclinie, three-dimensional, non-hydrostatic models would be useful. 
6.1.5 Propagation and modification on the shelf 
Wave propagation and modification on the shelf excluding near-shore waters are simulated 
adequately by non-linear hydrostatic models like those used at the BSH provided that a 
suitable grid resolution is available which accurately represents the topography and 
coastline. 
Figure 6.1.1 shows water levels at Helgoland comparing hydrostatic (black) and non 
hydrostatic (red) computations of the Federal Waterways Engineering and Research Institute 
(BAW) using the MARTIN model (Milbradt 2002). The model uses a finite element method 
with a higher resolution close to the coast. According to the estimations in section 5.1, there 
are only minor differences between the hydrostatic and non-hydrostatic simulations at this 
location. 
However, a definite improvement of the simulation results is achieved by reducing grid 
spacing in the entire gridded area from 9 km to 2.5 km maximum grid length (Fig. 6.1.2). 
Fig. 6.1.1: Water level curves at Helgoland in a hydrostatic (black) and a non-hydrostatic 
(red) simulation during a 12 h period (input signal: 1 positive signal, period 1800 s, 
wave height 3m at the northern boundary of the North Sea, MARTIN model, PluB 
2005, personal communication) 
A Cm] 
n 
-1 
.500 
.100 
-0.700 
-0.300 
0.100- 
0.500- 
0.900- 
1.300- 
1.700- 
2.100 
2.500 
2.900 
3.300 
. 
Fig. 6.1.2: Water level distribution 13 hours after simulation start (input signal: 1 positive 
signal, period 1800 s, wave height 3 m at the northern boundary of the North Sea, 
MARTIN model), coarse grid max. 9 km grid length (left), fine grid max. 2.5 km 
(right) (Lehfeldt et al. 2007).