Simulations of Tsunami Triggered by the 1883 Krakatau Volcanic Eruption: NH41A-1698 Implications for Tsunami Hazard in the South China Sea

YEN JOE TAN1, 2 JIAN LIN1 1Woods Hole Oceanographic Institution, Woods Hole, MA, USA 2Lafayette College, Easton, PA, USA, [email protected] 7. Tsunami hazard in the South China Sea Abstract 2. Modeling results 5. Eruption deposit (ring) vs. collapse (hole) The 1883 Krakatau eruption in Indonesia was one of the largest recorded 30 volcanic eruptions in recent history. The volcanic eruptions and the associated Hole and ring Location 1 Location 3 t = 10 min t = 20 min t = 30 min Reported Calculated Reported Calculated 2 tsunamis claimed about 36, 000 lives and recorded run-up heights more than 30 travel travel travel travel wave maximum wave Hole only 30˚ Location Ring only m along the coastal regions in the Sunda Strait between the Indian Ocean and time from time from height amplitude 20 Ryukyu Trench the South China Sea. In this study, we investigate the source mechanisms of origin (hr) origin (hr) (m) (m) Observed max. amp. 1. Merak, Indonesia 0.75 0.7 35 - 41 4.9 Observed arr. time 1 this tsunami using the non-linear approximation approach of the Cornell 2. Anjer, Indonesia 0.6 0.5 > 10 6 10 Multi-grid Coupled Tsunami Model (COMCOT). Model results reveal that a 3. N. Watcher Island, Indonesia 1.5 1.9 2.4 0.9 donut-shape “hole and ring” initial condition for the tsunami source is able to 4. Java-Batavia, Indonesia 2.5 2.7 1.8 0.6 20˚ explain the key characteristics of the observed tsunami. A “hole” of about 6 km 5. Tyringin, Indonesia 0.5 0.4 15 6 0 Philippine 6. Vlakke Hoek, Indonesia unknown 0.6 15 6 0 Trench in diameter and 270 m in depth corresponds to the collapse of the Krakatau t = 40 min t = 50 min t = 60 min 7. Port Blair, Andaman islands 5.5 4.6 0.2 0.01 Manila Trench volcano on August 27, 1883, while a “ring” corresponds to the eruption deposit 8. Galle, Sri Lanka 6.5 4.7 unknown 0.05 from the event. We found that the shallowness and narrowness of the entrance 9. Negombo, Sri Lanka 6.5 5.2 unknown 0.02 2 Philippines 10. Arugam Bay, Sri Lanka 5 5.6 unknown 0.09 Location 2 Location 4 10˚ pathway of the Sunda Strait limited the northward transfer of the tsunami energy Amplitude (m) ManilaSubmarine Trench Trench 11. Madras, India unknown 5.7 0.56 0.06 8 from the source region into the South China Sea. Instead, the topographic and 12. Negapatam, India unknown 5.8 unknown 0.07 landslides bathymetric characteristics favored the southward transfer of the energy into -15 0 15 m 13. Cossack, W. Australia 5.25 5.4 1.5 0.1 14. Geraldton, W. Australia 9 4.9 1.8 0.3 1 the Indian Ocean. This might explain why Sri Lanka and India suffered 4 casualties from this event, while areas inside the South China Sea, such as 0˚ Singapore, did not record significant tsunami signals. Modeling results further Fig. 5. Time series of modeled wave propagation in the source region in the first 60 minutes after the Sunda Trench Table 1. Compiled comparison of reported (Symons, 1888) and suggest that eruption deposit from the event is more capable in generating large eruption. This shows that the tsunami waves entered the Indian Ocean earlier than the South China calculated travel time from origin and wave height. 0 0 Krakatau wave amplitudes compared to depression from caldera collapse. Sea (SCS), and the maximum amplitude over the Indian Ocean region is slightly greater than that in the SCS. 0 1 2 3 4 0 2 4 6 -10˚ 1. Source region model setup 3. Linear vs. non-linear simulations Arrival time from origin (hr) Fig. 9. Computed tsunami waves at locations #1-4 using the “hole and ring”, “hole only”, and “ring only” model. The “hole 80˚ 90˚ 100˚ 110˚ 120˚ 130˚ 140˚ Observed max. amp. only” model consistently produces smaller wave amplitude. This suggests that the ring of eruption deposit plays a larger India South China Sea 30 role in generating the large wave amplitude observed during the tsunami event. Fig. 11. For countries inside of the SCS, tsunami risk from the Sunda, Java, Ryukyu, and Linear 10 2 southern Philippines trenches as well as the Krakatau volcanic chains is limited by the narrow 3 Non-linear and shallow entrance pathways. Therefore, the primary hazard for large tsunami in the SCS 20 -5.5º Observed could come from tsunami sources inside the SCS, including earthquakes from the Manila and arr. time Sri Lanka 1 6. Two-stage eruption process northern Philippines trenches and potential submarine landslides along continental margins 10 inside the SCS. 0 Single event One hour separation -6º 4 2 1 Amplitude (m) Tsunami source 0 2 4 0 2 Conclusions Location 1 Location 2 Location 3 Obs. points 0 1. A “hole and ring” model is able to explain the observed tsunami propagation Indonesia 5 0 -6.5º 0 2 4 0 2 4 0 2 4 characteristics at the five observation sites near the Sunda Strait. Arrival time from origin (hr) Fig. 6. Comparison of the computed tsunami waves between models using linear vs. non-linear approximations for -2 Expected arrival time -2 0 25 50 km from 1st stage eruption 2. Eruption deposit from the event (“ring”) is more capable in generating large Australia Indian Ocean locations #1-3 shows that waves generated from linear approximation are less stable. A linear approximation is -4 Expected arrival time wave amplitudes compared to depression (“hole”) from volcanic collapse. 105º 106º unsuitable for our study as the source region is very shallow relative to the tsunami wave height. from 2nd stage eruption Fig. 1. Map of modeling region (google.map.com). Figure to the left Fig. 3. Sites of historical records around the source region. (1) Merak; (2) Two hours separation Six hours separation 3. The shallowness and narrowness of the entrance pathway of the Sunda Strait shows the Krakatau Island before and after the 1883 eruption. Anjer; (3) N. Watcher Island; (4) Java-Batavia; and (5) Tyringin. 4. South China Sea vs. Indian Ocean 2 2 might have limited the propagation of tsunami energy into the South China Sea Amplitude (m) compared to the more open path of tsunami propagation into the Indian Ocean. -5.9º South China Sea South China Sea Sumatra Island, Indonesia Fig. 4. Donut-shape “hole and Indian Ocean 2 4. Non-linear approximation is more suitable for the modeling in our study due to ring” initial sea surface 0 0 South China Sea -5.5º the shallowness of the source region relative to the generated tsunami wave -5.5º displacement used in modeling. -6º A’ amplitude. “Hole” represents caldera that is -6º 1 Ring 6 km in diameter and 270 m -2 -6º -2 5. There is a possibility of interference between the tsunami waves generated by deep. “Ring” represents Amplitude (m) multiple eruption events and the interference effect depends on the time Java Island -6.1º eruption deposit from the event. -6.5º A 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 separation between the consecutive eruptions and the distance from the source. Krakatau Hole This was constructed based on -6.5º Bathymetry (km) Arrival time from origin (hr) estimated volume of deposition 0 25 50 km Indian Ocean Indian Ocean -2.3 -1.1 0.1 and change in surrounding Fig. 10. Comparison of computed tsunami waves at location #1 for a single event and a two-stage eruption process -6.2º 0 1 2 3 4 5 Initial sea surface bathymetry after the 1883 105º 106º with different separation time. The single event has the same source energy as the combined energy of the two-stage displacement (m) Arrival time from origin (hr) 104.5º 105º 105.5º 106º 106.5º eruption event (Verbeek, 1884; Fig. 8. Calculated tsunami time series at A and A’. The eruption while the two-stage eruptions have two events of the same energy. We can see that there is interference Acknowledgements Fig. 7. A-A’ profile shows two points of equal distance between the tsunami waves generated by two separate events, which could result in a second wave of higher -150 0 150 Sigurdsson et al., 1883). shallowness and narrowness of the entrace pathway into We express gratitude to the Woods Hole Oceanographic Institution (WHOI) Summer Fig. 2. GEBCO 30” bathymetric map of source region used for from the tsunami origin, with A located in the Indian the SCS could have slowed and limited the tsunami amplitude despite a second eruption of the same magnitude. The constructive interference effect is negligible for a Student Fellowship Program for financial support, as well as to AGU and WHOI for travel tsunami modeling. Grid lines represent model resolution. 105.2º 105.3º 105.4º 105.5º 105.6º Ocean and A’ located in the South China Sea. energy transfer into the SCS relative to the Indian Ocean. two-stage eruption with two or more hours of separation. grants.