1 Introduction

The Tsunami Commission was established within the International Union of Geodesy and Geophysics (IUGG) following the 1960 Chile tsunami, generated by the largest (M w 9.5) instrumentally recorded earthquake. The 1960 tsunami propagated throughout the entire Pacific Ocean, affecting areas located far from the source, and demonstrated the necessity of international cooperation. As organized, the Tsunami Commission is supported by two International Union of Geodesy and Geophysics (IUGG) associations: the International Association of Seismology and Physics of the Earth’s Interior (IASPEI) and the International Association for the Physical Sciences of the Oceans (IAPSO). Since its foundation, the Tsunami Commission has held biannual International Tsunami Symposia and published special volumes of selected tsunami papers (Satake et al. 2007, 2011a, b; Cummins et al. 2008; 2009).

Two recent volumes (Satake et al. 2013a, b) were associated with the 25th International Tsunami Symposium, held in Melbourne, Australia from 1 to 4 July, 2011, just 4 months after the catastrophic Tohoku tsunami of 11 March 2011. This tsunami was generated by the largest instrumentally recorded earthquake (M w 9.0) in the history of Japan, causing nearly 20,000 tsunami casualties and devastating tsunami damage on the Pacific coast of Honshu and Hokkaido islands. Numerous papers have already been published on the Tohoku earthquake and tsunami, including special issues in Earth, Planets and Space (Vol. 63, No.7, Editors: Kanamori and Yomogida 2011), Geophysical Research Letters (Vol. 39, No. 7, 2012), Coastal Engineering Journal (Vol. 54, No.1, Editor: Sato 2012), Earthquake Spectra (Vol. 29, No. S1, Editors: Mori and Eisner 2013), among others. Thus, it was natural that the entire Volume I of the 2013 PAGEOPH topical tsunami issue was on the 2011 Tohoku event (Vol. 170, Nos. 6–8, Editor: Satake et al. 2013a). However, results from new investigations into the 2011 Tohoku followed, requiring additional publication outlets. Additionally, in the remainder of 2011 through 2012 several other tsunami events occurred (Fig. 1). While these tsunamis were not as strong or devastating as the 2011 Tohoku event, they were nevertheless scientifically interesting and societally important.

Figure 1
figure 1

Red stars show the epicenter locations of tsunamigenic earthquakes occurring in 2011 and 2012; green diamonds show the locations of open ocean DART tsunameters

Full size image

These days, we are living in a new, “instrumental” era of tsunami science with ever more detailed (and sometimes overwhelming!) amounts of data available after each event. The types of data include precise seismological data from broadband IRIS and open-ocean bottom seismographs, GPS measurements of ground deformations on land, hydroacoustic and satellite altimetry data, not to mention the results of tsunami field surveys and high-resolution coastal tide gauge records. Most importantly, however, deep-ocean “tsunameters” now routinely provide direct measurements of tsunami waves in the open ocean, significantly improving our understanding of tsunami physics, the quality of tsunami modelling and the reliability and accuracy of tsunami warnings (Mofjeld 2009; Titov 2009; Rabinovich 2014). When a tsunami in the open ocean was first recorded (Filloux 1982; Dykhan et al. 1983), it was a scientific sensation. Nowadays, DARTFootnote 1 stations located throughout the Pacific, Indian and Atlantic Oceans (Fig. 1) record every tsunami no matter how small. The low level of background noise in the deep ocean and the high precision of DART instruments enable us to reliably identify and examine tsunamis with wave heights of only a few millimeters. Moreover, in recent years, Japan has developed a dense and effective network of cable tsunameters comprised of scores of stations (Kawaguchi et al. 2012; Rabinovich 2014). These stations provide invaluable information about tsunamis generated by earthquakes in this region (e.g., Matsumoto and Kaneda 2013; Satake et al. 2013c). Similarly, in Canada, the University of Victoria, BC, established in 2009 the NEPTUNE-CanadaFootnote 2 geophysical observatory on the southwestern shelf of Vancouver Island, which has already recorded more than ten tsunami events (Thomson et al. 2011; Rabinovich et al. 2013).

In general, we can say that the tools available for tsunami research have dramatically improved over the last 35 years. While the 1980s were rife with scientific ideas and theories about tsunami waves, there was very little empirical data; now we have an abundance of observational data, but sometimes not enough ideas to explain the new findings! In any case, the ongoing publication of high-quality scientific results is very important, and for that reason, the IUGG Tsunami Commission decided to prepare the additional, “inter-symposium” topical tsunami issue presented here. Altogether, 30 papers were submitted with 21 selected for publication.

2 The 2011 Tohoku Tsunami

This volume presents 12 new papers discussing the 2011 Tohoku tsunami: three papers deal with field data including previously unpublished data from surveys in the Fukushima exclusion zone, three papers investigate the seismic source, two papers deal with tsunami data recorded off the coast of Russia and four papers look at the tsunami waves recorded in the far-field (Marshall Islands, Hawaii, California and basin-wide).

Starting the group of field survey papers, Tsuji et al. (2014) compare the results from the Tohoku post-tsunami field surveys to runup and inundation data from past great earthquakes along the Honshu coast. They show that along the Sanriku coast, tsunami heights from the 2011 event were positively correlated with the previous Sanriku tsunamis, indicating that for near-field tsunamis, local variations in tsunami height resulting from the irregular coastline may be more dominant than the earthquake location, type, or magnitude. Along that coast, correlations to heights from the far-field Chilean tsunamis (1960, 2010) were less significant due to differences between the local and trans-Pacific tsunamis. However, along the Ibaraki and Chiba coasts, wave heights from the 2011 Tohoku and the Chilean tsunamis are positively correlated, showing a general decrease toward the south with small local variations such as large heights near peninsulas.

Shimonzono et al. (2014) next describe in detail the maximum runup height of nearly 40 m measured in a funneling coastal valley of the Aneyoshi district north of the entrance to Yamada Bay. Wave records from offshore GPS-buoys are introduced in a numerical model to analyze the measured localized runup amplification. The results indicated that a spectral component with a relatively short dominant period of 4–5 min in the leading wave plays a key role in the tsunami runup amplification by local bathymetry and topography. A final field survey paper by Sato et al. (2014) reports on detailed field data collected from inside the former 20-km exclusion zone around the Fukushima Dai-ichi nuclear power plant (NPP). The field surveys in the exclusion zone were delayed for more than a year due to elevated radiation levels caused by the multiple meltdowns at the NPP. Detailed distributions of the measured tsunami heights are presented in combination with observed infrastructure damage. Large tsunami heights exceeding 10 m were limited to areas within 500 m from the shoreline, while onshore profiles of the maximum inundation levels were dependent on inland topography. Tsunami flood levels in the coastal plains were affected by the extent of seawall damage, and remnant seawalls that survived the tsunami overflow provide valuable design lessons.

Next, we present two papers dealing with the seismological and wave generation aspects of the 2011 earthquake. First Rodkin and Tikhonov (2014) use data from the Japan Meteorological Agency (JMA) to examine the recent historical seismicity of the Tohoku earthquake source region. They find that this earthquake occurred in a region where over the previous 7 years (2005–2011), there was a considerable decrease in the number of earthquakes and associated b-values. The authors then describe some short-term (20 days) precursor anomalies of the Tohoku earthquake, in particular, an increase in the number of earthquakes and a decrease in hypocentral depths in a 120 km radius around the epicenter of the 2011 mega event. The unexpectedly high tsunami waves associated with a rather small rupture area of the Tohoku earthquake demonstrate that similar events could occur in some other subduction zones where such cases had previously been considered impossible.

This is followed by a review paper by Pararas-Carayannis (2014), who revisits the earthquake and tsunami source mechanism of the great Tohoku-Oki earthquake of 11 March 2011. An examination of the source mechanism determined that the great tsunami resulted from a combination of crustal deformations of the ocean floor due to up-thrust tectonic motions augmented by additional uplift due to the quake’s slow and long rupturing process, as well as to large coseismic lateral movements that compressed and deformed the compacted sediments along the accretionary prism of the overriding plate. The author suggests that as with the 1896 Sanriku tsunami, additional ocean floor deformation and uplift of the sediments was responsible for the higher waves generated by the 2011 earthquake.

One of the most critical issues relevant to tsunami forecasting, and tsunami modeling in general, is the characterization of the tsunami source. A tsunami source can be modeled based on data from seismometers, GPS measurements of crustal deformation, deep-ocean tsunameters, and other advanced instruments that provide near real-time information. Using data from the 2011 Tohoku earthquake, Wei et al. (2014) applied several different source models, based on different data sources, to simulate the generation, propagation and inundation of tsunami waves in the near field. They compare their results to measured data, including inundation extents and water level records from nearby tsunameters. Their results highlight the critical role these deep-ocean tsunami measurements play in the rapid determination of an approximate tsunami source suitable for issuing reliable and accurate forecasts. They also show that results from tsunameter-derived source models are compatible with results based on other real-time geodetic data, and may also provide a more nuanced understanding of tsunami generation from both tectonic and, non-seismic processes such as submarine landslides.

Our last cluster of papers on the 2011 Tohoku tsunami highlights the findings of researchers who studied tsunami records from various locations around the Pacific. Two papers from the Russian coast give a regional perspective, while four other papers examine tsunami waves at far-field locations or over the Pacific basin as a whole. This group of papers starts with the work of Razjigaeva et al. (2014), who examined tsunami sedimentation in the bays of the southernmost group of the Kuril Islands (the Lesser Kuril Islands). They report on the unusual situation where the sea during the tsunami was covered with ice, and describe changes in the erosional capacity of the tsunami due to ice fragments carried along with the waves. Their paper is a continuation of their previous work (Razjigaeva et al. 2013), which was based on the direct post-event studies. Their newer studies showed well-preserved tsunami deposits, evident one-and-a-half years after the event. A comparative analysis between tsunami deposits in the Kuril Islands relative to near-field (Honshu Island) sites indicated that differences in runup heights contributed to considerable differences in erosion, sedimentation, distribution of tsunami deposits, the formation of sedimentary structures, grain-size composition, and diatom and foraminifera assemblages. On the coasts of the Lesser Kuril Islands, the mud layers in sections of coastal lowlands in closed bays contained the grain-size composition of the 2011 mud, and preserved detailed geological records of paleotsunamis.

This is followed by the contribution of Shevchenko et al. (2014), who examine deep-water and coastal observations and data of the 2011 tsunami along the Far East coast of Russia. Key findings from their study show that tsunami waves reached up to 2.5 m along the Kuril Islands and that tsunami waves affecting the region northeast of the source exhibited much longer periods than waves propagating directly to the east. The authors estimated the major characteristics of the recorded waves and compared the observed tsunami series to the results from numerical models. In contrast to the open-ocean stations where the first waves were the highest, at Russian coastal sites, the highest waves occurred several hours after the arrival of the first tsunami wave.

Heidarzadeh and Satake (2014a) present evidence that extremely long modes of the Pacific Ocean were excited by the 11 March 2011 Tohoku tsunami. A numerical approach was employed to calculate the basin-wide modes of the Pacific Ocean, resulting in 49 modes in the range of 2–48 h. Spectral analysis of tide-gauge records showed that some of the calculated basin-wide modes were indeed excited by the Tohoku tsunami. The tide gauge signals were classified into three groups: (1) basin-wide modes (> 1.5 h), (2) the tsunami source periods (20–90 min), and (3) local bathymetric effects (<20 min). The average contributions to the total tsunami energy were 6.4 % for the basin-wide mode, 64.1 % for the tsunami source, and 29.5 % for the local bathymetry. Simulations suggest that the amount of contribution of basin effects to the total tsunami energy depends on the location of the tsunami source.

Next, Ford et al. (2014) report on an interesting mid-basin data set of tsunami waves from the 2011 Tohoku event. In their study, tide gauge data and numerical simulations supplement a serendipitous recording of the tsunami on an array of bottom-mounted pressure sensors at Majuro and Kwajalein Atolls in the Republic of the Marshall Islands. They show that tsunami oscillations in the lagoon were substantially more energetic and longer lasting than observed on the reefs or modeled in the deep ocean, and that the tsunami excited the normal modes of the atoll lagoons. They also showed that the propagation of the tsunami across the reef flats is tidally dependent, with amplitudes increasing/decreasing shoreward at high/low tide. Most importantly, from a hazard management viewpoint, they showed that while the peak wave heights in the Marshall Islands coincided with low tide, the observed amplitudes could have caused inundation had they coincided with a higher tide stage.

In the far field, Zhou et al. (2014) and Admire et al. (2014) investigate both tsunami wave heights and currents. Firstly, Zhou et al. (2014), examine the effects of frequency dispersion on tsunami waves and currents from the 2011 Tohoku event. In a series of numerical experiments, they compare the trans-oceanic propagation and tsunami dynamics in the vicinity of the Hawaiian Islands using both a weakly dispersive Boussinesq model and a shallow-water model that neglects dispersion effects. The model results indicate that dispersion effects generally result in reduced amplitudes of the leading tsunami waves. They also show that a model neglecting dispersion effects could underestimate wave heights and current speeds of the trailing waves developing in coastal waters.

This is followed by Admire et al. (2014), who present some relatively rare tsunami current speed data from northern California, including instrumental data from the 2010 Chile and 2011 Tohoku tsunamis in Humboldt Bay and current speeds derived from video camera footage at the entrance to Crescent City Harbor during the 2011 Tohoku tsunami. During the 2011 event, the tsunami signal was evident for more than 40 h in Humboldt Bay, with a peak current speed of 0.8 m/s occurring approximately 1 h after arrival. At Crescent City, within the first 3 h of tsunami activity, peak surface currents were estimated to have exceeded 4.5 m/s on one wave cycle and 3 m/s on six others, and were the cause of most of the damage experienced there. Their study also presents numerical model results in an effort to calibrate to measured data.

3 New Tsunami Events

Four new tsunami events occurred in 2012, and in this volume we present papers describing aspects of each. In August 2012, tsunamis occurred in El Salvador and Nicaragua on August 27 and in the Philippines on August 31. Borrero et al. (2014) conducted a post-tsunami reconnaissance survey of the El Salvador event and measured tsunami heights of up to 6 m with a maximum inundation distance of over 300 m. This is an important finding in that the causative earthquake for this event was relatively small, with a moment magnitude of only 7.3. Seismological analyses of the earthquake showed that it was a ‘slow’ or ‘tsunami’ earthquake (Kanamori 1972), making it the second such event in this region in 20 years—an important consideration when assessing tsunami hazards for this region. Heidarzadeh and Satake (2014b) also investigated aspects of the El Salvador tsunami as well as the August 31 Philippines tsunami. In their study, they focused on the far-field signature of these events as recorded on coastal tide gauges and deep-water tsunameters. They performed a detailed spectral analysis (using Fourier and wavelet techniques), and showed that the spectral content of the tsunami signal is related to the location of the recording station relative to the orientation of the tsunami source.

On 27 October 2012, a relatively large (M w 7.7) earthquake occurred off the west coast of Haida Gwaii in the Canadian province of British Columbia. Cassidy et al. (2014) provide an overview of the seismo-tectonics of the fault region and put this event in context. They explain that although the earthquake source region is predominantly a strike-slip transform fault boundary, there is a component of oblique convergence between the Pacific and North America plates off Haida Gwaii. The October earthquake, the second largest instrumentally recorded earthquake in that region, had a primarily thrust mechanism and was responsible for generating a tsunami that caused significant local runup. Leonard and Bednarski (2014) specifically examined the near-field effects of the 2012 Haida Gwaii tsunami. Despite a lack of evidence suggesting damaging waves along the coast of British Columbia (largest amplitudes of ~0.5–1 m were measured in Hawaii), field surveys conducted in the weeks and months following the earthquake revealed that much of the unpopulated and un-instrumented coastline of western Haida Gwaii was impacted by significant tsunami waves reaching a maximum height of 13 m with runup exceeding 3 m along 200 km of the coastline. The greatest impacts were evident at the heads of narrow inlets and bays on western Moresby Island, where natural and anthropogenic debris was found on the forest floor and caught in tree branches, suggesting flow depths up to 2.5 m. The authors indicate that lessons learned from their study of the impacts of the Haida Gwaii tsunami may prove useful to future post-tsunami and paleotsunami surveys.

Finally, Bernard et al. (2014), report on a small tsunami produced by a M W 7.3 earthquake offshore of Japan, adjacent to the source region for the 2011 Tohoku event. They present the deep-water tsunameter data from this event, recorded on instruments that were deployed just 2 weeks before the causative earthquake. They show that the data recorded by these two instruments helped to improve the speed and accuracy of tsunami forecasts for the coast of Japan and the Pacific Ocean in general.

4 Tsunami Science

Four papers in this issue relate to general topics in tsunami science, warning and hazard mitigation, and use examples from recent events to provide new insights or answers. In particular, Kim and Whitmore (2014) examined the relationship between tsunami maximum amplitude and signal duration based on 89 historical data sets from 13 tsunamis occurring between 1952 and 2011. The problem of event duration is quite important for effective tsunami warnings, hazard assessment and emergency response. They evaluated the tsunami sea level time series and used a linear least squares fit to get a quantitative estimate of amplitude decay function for their study region. The confidence interval was found to be roughly 20 h over the range of maximum tsunami amplitudes; this relatively large interval likely resulted from local resonance, late-arriving reflections, and other effects.

Nyland and Huang (2014) discussed specific problems related to tsunami warnings and advisories based on the West Coast and Alaska Tsunami Warning Center (WCATWC) experience during the 2011 Tohoku tsunami event. Specifically, they looked at the problem of how long a tsunami warning or advisory should be in place. To address this issue, the WCATWC developed a technique to estimate the amplitude decay of tsunami waves recorded at tide stations within the WCATWC Area of Responsibility (AOR). Based on an analysis of the real-time tide gauge records, they estimated exponential decay curves, which were then combined with an average West Coast decay function to provide an initial tsunami amplitude-duration forecast.

Nosov et al. (2014) used detailed slip distribution data from recent earthquakes and the Okada (1985) formulae to calculate the vector fields of coseismic sea-floor deformations, the displaced water volume and the gravitational potential energy of the tsunami source. They suggest that in many cases, the horizontal components of the bottom deformation provide an additional contribution to the displaced water volume without reducing the contribution of the vertical component. This increase varies from 0.07 to 55 % with an average increase of 14 %. The authors go on to examine measures of a tsunami’s initial energy as functions of the moment magnitude, and show that a tsunami captures from 0.001 to 0.34 % of the earthquake energy, and on average 0.04 %.

And finally, closing this volume, Kaistrenko (2014) describes a Poissonian probability model for tsunami runup heights, with emphasis on a tsunami recurrence function. He shows that a general tsunami recurrence function contains at least two important scaling parameters: (1) the asymptotic frequency of big tsunamis f at a study site, and (2) a characteristic tsunami height H* for that site. These parameters and their variations are statistically evaluated using observational data from tsunami catalogues. The author also considers theoretical and applied problems related to tsunami recurrence, shows an example of a two-parameter tsunami hazard map, and discusses issues related to probabilistic tsunami hazard estimation.