3D - Field case, Port Valdez, AK 1964

Benchmark Problem 7: Field Case: Slide at Port Valdez, AK during 1964 Alaska Earthquake

The present benchmark consists of simulating the extent of the inundation for two slide events at Port Valdez (one at the head of the bay and another at the Shoup Bay Moraine), based on before and after bathymetry data, and performing a comparison of the simulated results with eye-witness observations of the event, and observed run-up distribution.We have performed simulations for 4 different scenarios, two consisting in single slide events: HPV (for Hotel Port Valdez) and SBM (for Shoup Bay Moraine); and two dealing with combined HPV/SBM scenarios, one using the same model parameters as for the single events initially simulated and a second one using new parameters to fit the combinedscenario to the observed data as a one single scenario.This latest case is the one better reproducing the described observed data.

HPV scenario

A range of model parameters has been considered in order to perform a large set of numerical simulations.The aim of this approach was to find the optimal values for the parameters considered (angle of repose, ratio of densities, friction between water and sediment and friction between water and bottom surface).In the case of HPV scenario, the parameters were taken in the following intervals:

• Angle of repose (a): [4º- 6º]
• Ratio of densities (fluidized sediment/water): [0.25 - 0.45]
• Friction water/sediment: [0.005 - 0.01]
• Friction water/bottom: 0.005

For this scenario the optimal parameters found were (among other posible choices):

• Angle of repose (a): 6.0
• Ratio of densities (fluidized sediment/water): 0.35 
• Friction water/sediment: 0.005
• Friction water/bottom: 0.005

(a) HPV: Initial condition

(b) HPV: Maximal free surface elevation

(c) Location of selected points (37, 38, 39, and VH) (d) Inundation at Valdez 

Figure 16. HPV scenario. Water elevation time series at Valdez locations.

Figure 15.a shows the initial condition considered for this scenario, and Figure 15.b the maximum water surface elevation reached along the whole duration of the simulation. Figure 15.d presents the inundation extension at Valdez. It can be observed that McKinley Street is reached by the inundation. Figure 16 shows the time series of the water height  at 4 selected locations at Valdez (shown in Figure  15.c), including Valdez Hotel. It can be observed that  at location 38 several waves are numerically reproduced, but only one wave is observed at Valdez Hotel and location 39.

SBM scenario

In this second single slide scenario the varying intervals  considered for the 4 model parameters were:

• Angle of repose (a): [4º - 9º]
• Ratio of densities (fluidized sediment/water): [0.25 - 0.45]
• Friction water/sediment: [0.02 - 0.04]
• Friction water/bottom: [0.005 - 0.01]

In this case the optimal parameters found were: 

• Angle of repose (a): 5.0
• Ratio of densities (fluidized sediment/water): 0.45
• Friction water/sediment: 0.03
• Friction water/bottom: 0.005

 

(a) SBM: Initial condition

(b) SBM: Maximum free surface height

(c) Run-up at Anderson Bay (d) Inundation at Valdez

(e) HPV: Elevation time series at Navigational Light

Figure 17: SBM scenario

Figure 18: SBM scenario Wave height time series at selected locations at Valdez.

 

Figure 17.a. shows the initial condition considered in this scenario, and Figure 17.b the maximum water surface elevation reached along the whole duration of the simulation. Figure 17.c depicts the run-up values computed for Anderson Bay (they are larger than 20$m). Figure 17.d presents the inundation around Valdez. It can be observed that McKinley street is reached by the inundation.

The wave height time series obtained at the navigation light is depicted in Figure 17.e, where a maximum value over is 10m is obtained. Finally, in  Figure 18 wave height time series at selected locations are depicted. A 60cm wave arrive at Hotel Valdez, fulfilling the recommended benchmark objectives.It can be observed that  at location 38 three waves are numerically reproduced, but only one wave is reproduced at Valdez Hotel and location 39.

 

SBM+HPV combined scenario 

Now we consider a combined scenario, where both HPV and SBM slides are triggered simultaneously. As was the case for the two single scenarios, the combined scenario also fulfill the combination of all the required benchmark recommendations. Despite that this HPV+SBM combined scenario fulfills all the required items, we cannot consider this experiment setup optimal as, for example, in Anderson Bay numerical results do not fit accurately the observations. 

The reason explaining this not optimal behavior is that, in the SMB case, it is recommended to reproduce 20+ m run-up at the Anderson Bay and an inundation larger than 0.5 m at Valdez Hotel. We do think that the requirement on the inundation at Valdez Hotel should be removed from SBM scenario and should be included in HPV case. The first argument to propose this change is that it seems to be more consistent that HPV slide be the main responsible of inundation at Valdez. Our second argument is based on the numerical simulations performed for the SBM case. We have observed that, when the reproduced inundation at Valdez Hotel reaches this 0.5m value, the corresponding run-up at Anderson Bay is much larger than the real observed run-up of 20+m (see Fig.19 where 30+m runup is simulated).  This is also the reason for having an overestimated runup at Anderson Bay in the combined scenario.

Figure 19. HPV+SBM combined scenario. Runup at Anderson Bay.

Consequently, we have considered a new SBM* redefined experiment where we retain the condition over the run-up at the Anderson Bay, without trying to fulfill the condition on the inundation at Valdez Hotel, while we try to fulfill this last condition for the HPV scenario. The other requirements are retained as they were defined. Then, with the new parameters determined for SBM* scenario, a new combined scenario (named as HPV+SBM* combined experiment) has been considered. This numerical test fully achieves all the requirements while, at the same time, reproduces a better simulated runup at Anderson Bay than all previous scenarios. 

Figure 20.a shows the inundation around Port Valdez as simulated in the optimal combined scenario. It can be observed that McKinley street is reached by the flooding wave. Figure 20.b depicts run-up values computed at Anderson Bay. Run-up values in this area are above 20m. The wave-height time series reproduced at the Navigational Light is depicted at Figure 20.d. A maximum height over 12m is obtained. Finally, Figure 21 presents the wave height time series of the combined scenario at the selected locations in  Fig.15.c compared with the two single scenarios, HPV and SBM*. It can be observed that, in the case of the combined scenario, two neat waves are simulated at Valdez Hotel and at location 39. They are separated by approximately 4 min. The  first wave is 20cm high, arriving to location 39 2.5 min after landslide triggering and about 4 min to Valdez Hotel. The second wave reaches more than 50cm at Valdez Hotel arriving there 7 min after trigering. It is curious how, roughly speaking, the HPV component of the combined scenario is the responsible for the first simulated wave and while the second larger wave  does not appear in the SBM scenario simulation.  Therefore, this second wave is produced as result from the nonlinear combination of the waves produced by both slides occurring at the same time. The main discrepancy with the description provided for this combined scenario is the time between the two waves. The reported time is 10-15 min, while the simulated elapsed time at location 39 is around 5 min. The only way to obtained two waves separated by the reported time should be simulating two asynchronous slides separated by 5-10 min time. First the HPV slide and then, for example 5 min later, the SBM slide.

Acknowledgements

 

This research has been partially supported by the Spanish Government Research project SIMURISK (MTM2015-70490-C2-1-R and MTM2015-70490-C2-2-R) and the University of Málaga, Campus de Excelencia Internacional Andalucía Tech.

The multi-GPU computations were performed at the Unit of Numerical Methods (SCAI, University of Málaga).

 

(a) HPV+SBM*: Inundation at Valdez (b) HPV+SBM*: Run-up in Anderson Bay area

(b) Maximum wave height and inundation

Figure 20: HPV+SBM* (combined optimal) scenario

Figure 21: HPV+SBM* scenario. Wave height time series at selected locations. Comparison with HPV and SBM* scenarios