Skip to content

17 July 1998 Papua New Guinea Tsunami

Preliminary Simulation Description

We have made several MOST (Method Of Separation Tsunami) numerical model runs for initial conditions corresponding to different fault plane parameters, with strike, dip and rake corresponding to the Harvard CMT fault plane solution.

Figure 1. Summary of Maximum Wave Height Distributions Along Coast for Cases

 

Parameter	Case 1   FP 2	Case 2   FP 2	Case 3   FP 1	Case 4   FP 1
Strike (deg)	292	292	147	147
Dip (deg)	79	79	14	14
Rake (deg)	82	82	124	124
Slip (m)	1	1	1	1
Length (km)	50	70	70	100
Width (km)	25	40	40	50
Top Depth (km)	5	5	5	5
Energy (N-m * 10**19) (Assumes rigidity = 2*10**11 N/m**2)	25	56	56	100
Animations	Case 1	Case 2	Case 3	Case 4
Maximum Computed Wave Heights		Case 2	Case 3

Note the following important caveats:

1. The Harvard CMT energy estimate is 4.3*10**19 N-m, so the energy of these sources is much larger. Energy estimates, however, can easily be off by very large factors. And, had we matched the Harvard CMT energy exactly, we may not have been able to track the evolution of the tsunami waves. This is because we lack high-resolution bathymetry for the simulation, and are using a relatively coarse 2-minute grid. Furthermore, a small source would not be resolved well on this coarse grid.
   
   
2. The absolute values of the wave amplitudes in shallow regions are underestimated, since the waves are too short to be resolved by the 2-minute grid. In deeper water, the wavelengths are longer, can be resolved by the grid, and the computations are more accurate.
   
   
3. Only the relative distribution of maximum wave-heights along the shore are meaningful, not the absolute values.
   
   
4. No inundation computations were made, and the coastal boundary is assumed to be a solid wall at the 10-meter bathymetric contour.
   

In terms of results, we also note that:

1. Both animations indicate wave focusing in the Sissano area, but also display focusing and relatively large amplitudes at the islands across from Sissano. These islands are Wuvulu, Ninigo, Aua and Hermit. (Wuvulu and Aua may be inhabited, since our 1992 map seems to indicate the presence of villages.)
   
   
2. The distribution pattern of maximum values along the coast (Figure 1) are very similar for all cases simulated, i.e., there is a relative insensitivity to these particular source mechanisms.
   
   
3. Finally, we point out the obvious. These sources, with much larger energy than indicated by the Harvard CMT estimate, fail to reproduce the eyewitness reports of 7-10 m wave heights. This is not surprising, in view of similar preliminary computations after the Nicaragua, Flores and Okushiri tsunamis, for example. Either energy was concentrated at much smaller spatial scales than indicated by the single fault plane source mechanism, or other generation mechanisms, such as landslide(s) or some type of resonance phenomena played a role. As noted above, a coarse grid will lead to underestimates in wave heights, and this may also contribute to differences between modeled and observed. However, we suspect that, even with finer grids, the sources would not result in 7-10 m wave values.
   

DISCLAIMER: The numerical results are very preliminary, and cannot be used for emergency or mitigation planning, but should be considered as only a qualitative indication of the PNG tsunami energy distribution pattern.