The 2011 Tohoku tsunami generated major environmental changes in a distal Canadian fjord



[1] Tsunamis triggered by powerful earthquakes cause extensive damage and loss of life within many regions of the World Ocean. Although coastal inundation from major tsunamis is becoming increasingly well understood, we know little about the broader aspects of such events on distal marine systems. Here we use time series from moored oceanic sensors to show that the Tohoku tsunami generated by the magnitude 9.0 earthquake off eastern Japan in March 2011 caused days of surge-like currents and turbulent mixing in the inner basin of an anoxic Canadian fjord located over 7000 km from the seafloor rupture zone. Mixing, combined with the inflow of more oxygen-rich water from the adjoining outer basin, led to abrupt changes in the hydrodynamics, bottom sedimentation, and zooplankton behavior in the basin. These findings help define mechanisms by which major transoceanic tsunamis can significantly alter coastal marine environments located far from the source area.

1 Introduction

[2] The Mw 9.0 earthquake that struck the Tohoku District of northeastern Japan at 05:46 UTC on 11 March 2011 was the strongest in the nation's history and one of the strongest on record [Saito et al., 2011; Simons et al., 2011]. The major tsunami generated by the earthquake propagated throughout the Pacific Ocean (inset, Figure 1a), arriving at the southwest coast of Vancouver Island, British Columbia approximately 9 h and 30 min after the main shock. Approximately 10 min after reaching the tide gauge at Bamfield in Barkley Sound, the tsunami entered Effingham Inlet, where wave properties were subsequently recorded by a suite of instruments on mooring EF04 located at the entrance to the inner basin of the inlet (Figure 1b). According to the model simulations presented in Figure 1a (insert), maximum wave amplitudes (zero-to-peak) were roughly 15 cm in the offshore region seaward of southern Vancouver Island, 45 cm at the entrance to Barkley Sound, and 72 cm at the entrance to Effingham Inlet.

Figure 1.

The study region on the west coast of Canada. (a) Location of Effingham Inlet in Barkley Sound, southwest coast of Vancouver Island (depth contours are in meters). The inset shows the maximum wave heights (η, cm) for the 2011 Tohoku tsunami in the Pacific Ocean based on the Institute of Ocean Sciences tsunami model (courtesy of Isaac Fine); (b) a chart of Effingham Inlet showing the location of mooring EF04 (red triangle) and the two sills; and (c) a depth profile along the centerline of the inlet showing the mooring location and instrument depths (dashed line with symbols).

[3] Effingham Inlet is a 1 km wide, 17 km long, glacially excavated western Canadian fjord with distinct inner and outer basins separated by shallow bedrock sills with minimum depths of 40 and 65 m, respectively (Figure 1c). Water depths range from 120 m in the inner basin to 210 m in the outer basin. The poorly ventilated 7 km long inner basin is characterized by an anoxic 40 (±10) m thick bottom water layer that sits above well-preserved and annually varved seafloor sediments. The sediment record has been the subject of numerous biological, chemical, and paleoclimate studies [Patterson et al., 2004; Dallimore et al., 2005; Dallimore and Jmieff, 2010; Wright et al., 2005; Hay et al., 2007; Chang et al., 2013]. A 40.9 m piston core collected in 2002 by the French drillship Marion Dufresne extends the sediment record back to the late Pleistocene [Ivanochko et al., 2008].

[4] As part of a continuing effort to study the effects of oxygen renewal events on the geochemistry, sediment formation, and biological productivity of the inlet, we maintain a well-instrumented current meter mooring at site EF04 anchored at 110 m depth at the northern end of the narrow channel leading into the inner basin (Figures 1b and 1c). Time series of temperature, salinity, current velocity, and acoustic backscatter intensity from the mooring are augmented by shipboard water property surveys employing water bottle samplers and conductivity-temperature-depth (CTD) profilers equipped with dissolved oxygen and light attenuation sensors. Here we examine time series data collected during the mooring period 3 May 2010 to 21 May 2011 to demonstrate the marked impact of the 2011 Tohoku tsunami on the water property structure, bottom sediments, and zooplankton distribution in the inner basin of the inlet. Profile data used in the study are those closest to the times before and after the tsunami event.

2 Observations

2.1 Background Conditions in the Inlet

[5] Temperature and salinity profiles (Figure 2a) reveal that the inner basin of Effingham Inlet is well stratified to 70 (±10) m depth and weakly stratified below this depth. Vertical gradients are highest within the thin (~10 m) brackish surface layer formed by discharge from the small (flow rate ~10 m3/s) rain-fed Effingham River at the head of the inlet. With the exception of the ~0.1 m/s tidal currents over the sill during spring tides and ~0.1 m/s estuarine outflow in the top few meters during periods of heavy rainfall, currents in the inner basin are very weak.

Figure 2.

CTD profiles taken at the center of the inner basin on 21 September 2010 (thin line) and 21 May 2011 (thick line). Depth ranges recording an increase in a property value from 2010 to 2011 are shaded red, those experiencing a decrease in property value are shaded blue (we chose not to color code the LAC plots). The data have a vertical resolution of 1 m. (a) Temperature (T) and salinity (S); and (b) dissolved oxygen concentration (DO) and light attenuation coefficient (LAC) expressed as a percentage of clear water. The symbols in the DO profiles are from titrated water bottle samples which confirm zero oxygen values in the bottom anoxic layer at the time of the CTD casts (squares and triangles denote water sample values from the 2010 and 2011 surveys, respectively).

[6] Because of the weak currents and upper layer stratification, the inner basin is typically anoxic below about 70 m depth (Figure 2b). Near-bottom water samples collected simultaneously with the SeaBird 9-plus CTD profiles typically have zero oxygen concentrations and commonly give off the distinct odor of hydrogen sulfide (H2S) gas when first opened onboard ship. Thus, the small near-bottom oxygen values of around 0.1 mL/L recorded by the electronic probes in Figure 2b are not real and an artifact of the sensor response. In contrast to the inner basin, the deep waters of the adjoining outer basin are generally suboxic (~1 mL/L of dissolved oxygen below 90 m depth). At depths shallower than 90 m, the outer basin is well oxygenated (> 2 mL/L) due to frequent inflow of oceanic water from the Pacific coast. Only rarely does relatively oxygen-rich water from the outer basin penetrate into the inner basin to aerate the anoxic bottom waters [Dallimore et al., 2005; Chang et al., 2013].

[7] Water clarity profiles (Figure 2b) collected using 660 nm wavelength transmissometers integrated into the CTD systems show that, in addition to a turbid surface layer, there is a thin (~2 m) layer of highly opaque (< 25% light transmission) water immediately above the oxic-anoxic (redox) interface. Relatively high water clarity values of ~55% are observed at intermediate and near-bottom depths. Previous studies [Hurtgen et al., 1999] reveal that the layer contains particulate matter formed at the redox boundary when dissolved constituents such as iron contact the oxygenated water above. It is also possible that the redox interface is blanketed by a mat of filamentous chemosynthetic Beggiatoa bacteria, similar to that observed in studies in Saanich Inlet [Juniper and Brinkhurst, 1986], a 200 m deep anoxic basin on the opposite side of Vancouver Island relative to Effingham Inlet (Figure 1a).

2.2 Time Series Measurements

[8] At the time of the March 2011 tsunami, mooring EF04 consisted of single-point Nortek 2 MHz Aquadopp acoustic current meters (ACMs) positioned at elevations of 5 and 35 meters above bottom (mab), a profiling upward looking Teledyne-RDI 307 kHz Acoustic Doppler Current Profiler (ADCP) at 7 mab, and Seabird SBE37 CTDs at 5 and 35 mab (Table 1). The mooring was deployed on 3 May 2010 and recovered on 21 May 2011. The ACMs measured in situ temperature, pressure, and three-dimensional current velocity at 60 min intervals, while the SBE37s recorded in situ temperature and conductivity (salinity) at 20 min intervals. The ADCP recorded vertical profiles of acoustic backscatter intensity and current velocity every 20 min over twenty 4 m thick vertical bins spanning the upper 100 m of the water column (bottom temperature and pressure were also recorded). Velocity data from the ADCP were unreliable at the time of the tsunami—possibly because of the highly erratic turbulent-induced motions of the acoustic scatterers and the large volumes of water being ensonified by the ADCP—and have been ignored. The ACMs performed well due to their much higher frequency and ensonification of a much lower volume of water close to the acoustic transducers [Emery and Thomson, 2001].

Table 1. Instruments on Mooring EFO4 at the Time of the March 2011 Tohoku Tsunamia
InstrumentDepth (m)Elevation (mab)Δt (min)Parameters
  1. a

    Columns 2, 3, and 4 denote the instrument depths, height in meters above bottom (mab), and sampling interval, respectively. SBE 37 denotes a Seabird Electronics Microcat CTD, ACM, a Nortek acoustic current meter, and T-RDI ADCP, a 307 kHz Teledyne-RDI acoustic Doppler current profiler. Parameters u,v, and w are, respectively, the cross-channel, along-channel, and vertical components of current velocity, T is temperature, S is salinity, P is pressure (equivalent to water depth), and ABS stands for acoustic backscatter intensity.

  2. b

    Data not reliable.

Nortek ACM753560(u, v), T, P
SBE 37 CTD753520T, S, P
T-RDI 307 kHz ADCP; 4 m bins104720(u, v, w)b, T, P, ABS, tilt and roll
Nortek ACM105560(u, v), T, P
SBE 37 CTD105520T, S, P

[9] Selected time series from EF04 for March 2011 are presented in Figure 3, along with detided sea levels from the 1 min recording Bamfield tide gauge. Tsunami waves recorded at Bamfield had a leading crest amplitude of ~27 cm and attained a maximum trough-to-crest height of 80 cm roughly 3.5 h after arrival of the leading wave (Figure 3a). In addition to the tsunami signal, the sea level records from mooring EF04 originally contained low-frequency oscillations generated mainly by winds along the outer coast. These atmospheric contributions have been filtered out using a high-pass filter with a 4 h cutoff period. We have converted pressure variations to equivalent sea level variations.

Figure 3.

Time series from Bamfield and mooring EF04 over an 11 day period (7–18 March 2011) spanning the tsunami arrival. Time is in UTC (local Pacific Standard Time + 8 h). Vertical lines denote the time of the earthquake (E) and the first tsunami wave arrival (T); Δt is the instrument sampling interval. (a) Sea level variations at the Bamfield tide gauge and the mooring site (from ADCP and ACM pressure data). The original records were detided and then high-pass filtered to remove atmospheric effects; (b) the along-channel (v) and cross-channel (u) current velocity components at 35 mab (75 m depth) and 5 mab (105 m depth) from the acoustic current meters (ACMs). Vectors denote the mean currents averaged over 3 days before and 3 days after the tsunami arrival; and (c) temperature (T) and salinity (S) at 35 and 5 mab from the SBE 37 s.

[10] Except during the tsunami, currents observed at 75 and 105 m depth were primarily to the north along 3.1°T (True compass direction) and 356.8°T, respectively; i.e., within a few degrees of the orientation of the narrow channel containing the sill. Following arrival of the tsunami, the primary flow current direction at 75 m depth rotated abruptly to 327°T (the orientation of the inner basin), while that at 105 m depth rotated only slightly to 350°T (see mean current vectors in Figure 3b). Thus, the bottom flow was guided at all times by the local topography whereas the tsunami currents at mid depth were partially guided by the strike of the inner basin.

[11] Because of aliasing effects, maximum recorded values are expected to be smaller than actual values. This is especially true of the hourly current speeds. To estimate the actual maximum tsunami wave height and current velocity at EF04, we used the high-resolution sea level time series, η(t), for Bamfield as a reference. Specifically, we decimated the 1 min Bamfield tsunami record (η1) to 20 and 60 min increment sea level records (η20 and η60, respectively) using the same 3 min burst-sampling protocol used by the ACMs and ADCP. Ratios of these values were then used to “correct” (scale-up) the observations for aliasing. For example, the maximum tsunami wave height obtained from the Bamfield record was 79.5 cm, while the decimated values for 20 and 60 min sampling rates are 52.0 and 26.8 cm, respectively. The scale-up ratios are then η1/η20 ~ 1.5 and η1/η60 ~ 3. A similar procedure was used to evaluate tsunami wave heights for the 2004 Sumatra tsunami along the Atlantic coast of South America from tide gauge records with different sampling intervals [Candella et al., 2008].

3 Results and Discussion

[12] The leading wave from the Tohoku event reached the Bamfield tide gauge at 15:35 (±1 min) and the EF04 pressure gauges at 15:50 (±10 min) (Figure 3a). Tsunami waves at the mooring site had an initial displacement (η) of +37.6 cm (wave crest) and a maximum height (trough-to-crest difference) of 61.9 cm. The observed wave periods of 40–60 min in the 20 min ADCP record represent an aliased version of the dominant wave periods of 18–33 min observed in the Bamfield record. These relatively long periods are consistent with the open ocean frequency content of the Tohoku tsunami [Rabinovich et al., 2013]. After adjustment for aliasing, the estimated maximum wave height from the ADCP is 95 cm, placing it among the highest recorded tsunami waves on the British Columbia coast. The tsunami signal persisted for 5 days.

3.1 Changes in Water Column Hydrodynamics

[13] Arrival of the tsunami at the mooring site resulted in strong oscillatory currents with maximum recorded speeds (combined u and v components) of over 0.45 m/s (Figure 3b). Peak speeds were near the beginning of the event, with the leading wave crest generating strong inflow into the inner basin. In comparison, maximum background current speeds were around 0.05 m/s during the 1 year observation period. After adjusting for aliasing, we estimate that current speeds at the two current meter depths would have been as high as 1.3 m/s. Even stronger currents would have occurred in the constricted channel immediately to the south of the mooring site. Comparison of the observed 60 min along-channel velocity (v) record with the 20 min pressure (p) records from the ADCP indicates that the strong negative (southward) bias in the tsunami-induced flow is the result of aliasing. Specifically, if we assume v ~ p (the case for propagating waves), the more rapidly sampled pressure records indicate that current variations would have had nearly equal inflow and outflow speeds, thereby accounting for much of the bias in the current records. However, irrespective of the exact speeds and structure, it is clear that the tsunami-induced currents were at least a factor of 10 greater than background currents and that v ~ 1 m/s is a reasonable value for the flow speeds during the first few days of the event.

[14] Observed tsunami current velocities were of almost identical amplitude and phase at the two current meter depths, indicating that the tsunami-induced flow in the lower half of the water column was strongly barotropic. However, the observed flow was also sufficiently baroclinic that the mean square vertical shear (dv/dz)2 of the currents increased to 22.6(±36.2) × 10−3 s−2 during the first 2 days of the tsunami, compared to a background level of ~5.4(± 8.0) × 10−3 s−2 . Based on the water density (ρ) profile obtained from the September 2010 CTD survey, the gradient Richardson number (Ri) criterion for shear-generated instability in a stratified shear flow, Ri = N2/(dv/dz)2 < ¼ [Kundu, 1990], was readily satisfied at all depths at the time of the tsunami, which was not the case prior to the event for the upper layer; here N2(z) = − (g/ρ)dρ/dz  is the Brunt-Väisälä frequency and g is the acceleration of gravity (N varies from 10 × 10−2 s−1 in the stratified surface layer to 0.7 × 10−2 s-1 in the well-mixed bottom layer). The enhanced vertical shear at the time of the tsunami increased the kinetic energy available for mechanical mixing by at least a factor of 4 above background levels.

[15] As shown by Figure 3c, the leading wave caused a sudden increase in temperature (T) and a drop in salinity (S) at both the 5 and 35 mab instrument depths. This was followed by large-amplitude T and S fluctuations (roughly ±0.1°C and ±0.05 psu at 75 m depth) indicative of intense mixing within the inner basin. Superimposed on these short-period variations were positive post-tsunami offsets of +0.03°C in mean temperature and +0.06 psu in mean salinity at 75 m depth, indicating that vertical mixing was accompanied by intermediate depth intrusions from the outer basin (consistent with the opposing T and S gradients in Figure 2a, vertical mixing can cause the mean temperature or mean salinity to increase at mid depth but not both simultaneously, as was observed). The intrusions would have delivered more oxygen-rich water to the inner basin.

[16] The large T and S fluctuations observed at 105 m depth (5 mab) show that hydrodynamic effects from the tsunami penetrated to the bottom of the inlet. However, unlike the case for intermediate depths, the post-tsunami, near-bottom temperature, and salinity offsets (T increased, S decreased) indicate that mixing, rather than intrusive flow, was the dominant process (downslope intrusions would have produced either colder or saltier water compared to pre-tsunami bottom water conditions, contrary to what is observed). Also, unlike at mid depth where the water property offsets persisted for several months, the bottom salinity diminished to pre-tsunami values in about 2 weeks.

[17] The small (−0.02 psu) shift in salinity recorded at the mid depth (35 mab) instrument on 9 March, several days prior to the tsunami (Figure 3c), is likely the result of a weak mid-depth intrusion. That the event was weak is supported by the lack of corresponding response in the other sensors, especially temperature. Similar events are observed occasionally throughout the yearlong salinity record and presumably help maintain positive dissolved oxygen values at mid depth within the inner basin. However, the only strong intrusion affecting the entire water column was that which occurred on 11 March after the arrival of tsunami waves.

3.2 Changes in Bottom Sediments and Zooplankton Behavior

[18] Figure 4 presents time series of acoustic backscatter intensity and intensity anomaly from mooring EF04 for March 2011. Because we are primarily interested in relative changes in backscatter, the acoustic data have not been adjusted for range-related attenuation arising from acoustic absorption and geometrical spreading [Emery and Thomson, 2001]. The frequency f = 308 kHz ADCP ensonifies targets with characteristic length scales L ≅ ¼ (c/f) ≥ 1 mm, where c ≅ 1500 m/s is the speed of sound in water. Except near the bottom, where the ADCP observed resuspended bottom sediments and neutrally buoyant particulates (opaque layer), the backscatter returns are from zooplankton and pelagic fish [Flagg and Smith, 1989; Fielding et al., 2004; Thomson and Allen, 2000; Burd and Thomson, 2012]. To generate (Figure 4c), we determined the mean backscatter intensities for 10 days before and 10 days after the tsunami arrival. The mean pre-tsunami value was then subtracted from the mean post-tsunami value.

Figure 4.

Acoustic backscatter intensity from the upward looking 307 kHz ADCP moored at 7 m above bottom at site EF04. Each pixel spans a depth range of 4 m and time step of 20 min. Time is in UTC (local Pacific Standard Time + 8 h). (a) Backscatter time series for the period 9–16 March 2011; (b) same as in Figure 4a except that the record average values for each of the 25 separate acoustic bins have been subtracted from the original time series; and (c) the step-like change in acoustic backscatter intensity caused by the tsunami; areas shaded red denote an increase in backscatter and blue a decrease in backscatter between the two periods. The vertical lines in Figures 4a snd 4b denote the time of the earthquake (E) and the first tsunami wave (T). Note that 0 and 12 h UTC correspond to 16 h (4 P.M.) and 04 h (4 A.M.) local time, respectively. Time series for bins 24 and 25 are not valid since they are contaminated by side lobe acoustic reflections from the ocean surface [Emery and Thomson, 2001].

[19] The inner basin is characterized by three basic acoustic features: (1) a bottom layer extending up to (and including) the strong backscatter peak centered at bin 6 at 80 m depth near the redox boundary; (2) an intermediate layer extending from the top of the redox layer to the base of the surface estuarine layer (bins 7 to 22); and (3) an upper layer as sensed by bin 23 centered at roughly 12 m depth at low tide. Bin 23 has the largest acoustic range because daily variations in tide height cause the bin to move in and out of the highly reflective surface layer.

[20] The acoustic record reveals that the tsunami-induced motions strongly affected the biology and bottom sediments of the inner basin. Specifically, they are as follows: (1) the backscatter layer at the redox boundary around 80 m depth was intensified during and after the tsunami, although the layer did not alter its mean depth; (2) prior to the tsunami, there were distinct scattering layers centered at 30 and 50 m depth, whereas after the tsunami, there was only a single peak layer centered at 55 m depth. Subtraction of the backscatter distributions averaged over 10 day periods before and after the tsunami (Figure 4c) shows that the major post-tsunami drop in the acoustic signal between depths of 10 to 35 m was accompanied by a major increase in the signal centered at 55–60 m depth. Thus, large numbers of zooplankton abandoned the upper layer during and after the tsunami and moved to greater depth. Once established, this new “resting depth” persisted for at least several months; (3) normal diel zooplankton migration, consisting of the upward migration from depth around dusk (presumably to feed on phytoplankton in the surface euphotic zone) and downward migration to depth before dawn (likely to avoid predators) [Thomson and Allen, 2000], was a dominant feature prior to the tsunami. (The pre-tsunami migration behavior gives rise to the near-vertical streaks in backscatter intensity patterns observed after midday UTC on 9–11 March in Figures 4a and 4b, roughly after midnight local time.) At that time, zooplankton generally descended to 65 to 75 m depth and, on occasion, reached or passed through the redox interface around 80 m depth. This migratory pattern was disrupted by the tsunami and did not return to pre-tsunami conditions during the remaining 2 months of the record; (4) the post-tsunami diel cycle reveals a downward migration of animals to depths of around 40 to 70 m depth around dusk, attaining highest concentrations around midnight local time. The presence of these mid-depth scattering layers during darkness suggests that fish, which would have normally fed on zooplankton in the upper layer, may have been migrating downward to reach the new zooplankton resting levels. This pattern ended about a month after the tsunami; and (5) bottom sediments were lifted rapidly upward by as much as 40 m during the initial period of strong flow but settled back to the seafloor within about 1.5 days.

[21] Depending on the thickness of the material disturbed, past tsunami-induced bottom sediment resuspension events, such as those observed here, could be interpreted in sediment cores as bottom water renewal events. However, it is also possible that the depositional “fingerprint” associated with the short-lived resuspension may be sufficiently distinct from other processes recorded in the annually laminated cores (such as the 15 cm thick gravity-flow turbidites generated by the June 1946 earthquake in central Vancouver Island [Dallimore et al., 2005; Hay et al., 2007]) to allow detection of historical megathrust tsunamis. Sediment cores in the inlet do provide evidence for the last major (Mw ~ 9) earthquake and tsunami that occurred along the Cascadia Subduction Zone on 27 January 1700 [Satake et al., 1996]. However, bottom water renewal events appear to have been more frequent after the 1700 tsunami, resulting in fewer laminated units and encouraging bioturbation and mixing of underlying sediments [Hay et al., 2007].

4 Summary and Conclusions

[22] The observations presented here show that the 2011 Tohoku tsunami generated several days of relatively intense currents and turbulent mixing within the highly sheltered and normally quiescent inner basin of Effingham Inlet on the Pacific coast of Canada. These dynamical processes led to pronounced changes in the water column properties and to an abrupt but temporary resuspension of the bottom sediments. Downward mixing of well-oxygenated water from the upper water column, combined with tsunami-induced intrusions of more oxygen-rich water from the adjoining outer basin, would also have elevated dissolved oxygen (DO) concentrations at intermediate depths within the inner basin. Shipboard profile data collected several months before and after the tsunami (Figure 2b) indicate that the mid-depth DO in the basin increased by as much as 2 mL/L. Based on the abrupt changes in acoustic backscatter distribution that followed the tsunami arrival (Figure 4), it appears that the increased DO concentration enabled macrozooplankton to comfortably descend as much as 40 m deeper in the water column. After the tsunami had dissipated, zooplankton again ascended to the upper layer in the evening to feed, but were then able to descend to a much greater depth in the early morning to avoid the upper ocean predators. To our knowledge, the findings from Effingham Inlet are the first direct evidence that major transoceanic tsunamis can generate pronounced hydrodynamic and biological changes in marine environmental systems located many thousands of kilometers from earthquake source regions.


[23] The authors have benefited from the advice and support from Maxim Krassovski, Isaac Fine, Steve Mihály, Earl Davis, and Robie Macdonald. Patricia Kimber helped prepare the figures. We also thank the Editor, Pete Strutton, and the two anonymous reviewers for their useful comments and suggestions. A. Rabinovich was partly supported by the Russian Foundation on Basic Research grants 12-05-00733-a and 12-05-00757-а.

[24] The Editor thanks Kenji Satake and Francois Schindele for their assistance in evaluating this paper.