The 2016 Valentine s Day Mw 5.7 Christchurch earthquake: Preliminary report

The 2016 Valentine s Day Mw 5.7 Christchurch earthquake: Preliminary report A. Kaiser, C. Holden, I. Hamling, S. Hreinsdottir, N. Horspool, C. Massey, P. Villamor, D. Rhoades, B. Fry, E; D Anastasio, R. Benites, A. Christophersen, J. Ristau, W. Ries, T. Goded, G. Archibald, C. Little & S. Bannister GNS Science, Lower Hutt. 2016 NZSEE Conference Q. Ma University of Auckland, Auckland. P. Denys & C. Pearson University of Otago, Dunedin. M.Giona-Bucci & P. Almond Lincoln University, Lincoln. S. Van Ballegooy & S. Wallace Tonkin + Taylor Ltd ABSTRACT: The Mw 5.7 Valentine s Day earthquake struck just offshore from Christchurch on February 14 th February 2016, more than four years after the last major earthquake of the Canterbury sequence (the Mw 5.9 on 23 rd December 2011). This moderate-sized earthquake caused further rockfall from known susceptible slopes in the Port Hills and liquefaction in some parts of eastern Christchurch and Kaiapoi. Permanent ground displacement as a result of fault movement measured up to 11 cm in the eastern suburbs and recorded peak ground acceleration ranged up to 0.36 g. Response spectra highlight that some buildings may have experienced ground shaking exceeding a serviceability limit state event, and up to approximately 60% of an ultimate limit state even, although significant damage is considered unlikely. Here, we present a summary of the physical impacts, earthquake source models, ground motions and aftershock statistics in the context of the ongoing Canterbury earthquake sequence. 1 INTRODUCTION On February 14 th 2016 a magnitude 5.7 earthquake struck just off the eastern shore of Christchurch (Figure 1). This earthquake was the latest significant aftershock of the Canterbury earthquake sequence, which began in 2010 with the Mw 7.1 Darfield Earthquake. The Christchurch region had not experienced an earthquake of this magnitude or greater in over four years (i.e. since the Mw 5.9 earthquake on 23 December 2011). Prior to this event, GeoNet recorded the most recent aftershock greater than magnitude 5 in May 2012. However, the probability of an earthquake of magnitude greater than 5 within the Canterbury region for the next year was approximately 50% prior to this event. While not as damaging as the four major earthquakes of the sequence in 2010 2011 (e.g. Gledhill et al.; Kaiser et al.), this earthquake provides a timely reminder of the ongoing potential for strong shaking in Canterbury. It also provides a useful opportunity to assess the impact of a moderatesized aftershock on the recovering city, given that the probability of another aftershock of magnitude 5 or greater occurring in the future is high (see discussion on GeoNet website: http://info.geonet.org.nz/x/xieo). We present initial observations compiled from a range of

contributors summarising the physical impacts, earthquake source models, and ground motions, as well as discussing the probability for future ground shaking in the region. Figure 1 - Location of the 2016 Valentine s Day earthquake (orange star) in the context of the previous events of the Canterbury earthquake sequence. 2 PHYSICAL IMPACTS The earthquake initiated further rockfalls, cliff collapses and cliff top recession from susceptible slopes in the Port Hills. Initial observations suggest that affected areas were contained within the Canterbury Earthquake Recovery Authority s official red-zones and the Christchurch City Council s hazard zones. Working with LINZ geotechnical engineers, the GeoNet landslide team re-surveyed the main cliffs and re-mapped the cliff-top cracks in the Sumner area, to quantify cliff top cracking and recession, and the volumes of debris that fell from the cliffs in response to the earthquake. These surveys comprised terrestrial laser scanning (TLS) of the cliff faces and remote controlled aerial footage of the inaccessible cliff edges (via UAV). Areas were also identified where blocks in the slope have moved and not fallen, indicating the likelihood of further boulders falling off in the short-term. At Whitewash Head (Figure 2) for example, new cracking, in response to the 14/02/2016 earthquake was found to extend laterally across the cliff top, where homes had formerly been situated prior to removal after red-zoning. Based on our preliminary observations and measurements, the already damaged cliffs have been further weakened by the earthquake-induced cracking and dilation of the rock masses caused by the Valentine s Day earthquake. Greater volumes fell from cliffs closer to the epicentre (i.e. Richmond Hill Road and Whitewash Head) compared to those further away (i.e. Redcliffs and Shag Rock Reserve). The volumes that fell from the cliffs in this earthquake are comparable to those volumes that fell in previous earthquakes (post the destructive Mw 6.2 February 2011 earthquake), at similar levels of peak ground acceleration. The volumes of debris falling from these cliffs are currently larger than the long-term average volumes, estimated from the accumulation rate of debris at the cliff bottoms prior to the 2010 Darfield mainshock. These data suggest a period of elevated rockfall and cliff collapse hazard that could take many years to reduce to those levels that occurred before the 2010-2011 Canterbury Earthquake sequence. 2

Figure 2 - Photograph of Whitewash Head taken on the 17/02/2016. Note the large crack in the cliff face. Ground surface manifestations of liquefaction occurred in parts of the eastern suburbs of Christchurch (Bexley, North New Brighton, Parklands, Spencerville and Brooklands) and as far north as Pines Beach in the east of Kaiapoi township (Figure 3). The liquefaction manifestations were more localised compared to those observed following the major events of the sequence in 2010 2011. This is consistent with the fact that PGA in the affected areas was close to liquefaction triggering thresholds, such that only the localised areas with the loosest soils liquefied. Only one minor instance of lateral spreading was observed along the Avon River in North New Brighton. Liquefaction also occurred at the time that the GNS Science-Lincoln paleoliquefaction team were analysing trenches opened across the 2010-2011 liquefaction features at Pines Beach. Pines Beach had relatively extensive liquefaction for a small event and at a distance of ~ 13 km from the epicentre compared with areas closer to the source such as Brooklands, highlighting the fact it is a highly susceptible site. This can be attributed to the presence of thick deposits of loose-to-medium dense non-plastic fine-grained sediments close to the surface, and the very shallow water table (within 1 m of the surface). The 14 February liquefaction features have helped understand the effects of liquefaction in the soil profile in coastal settings, which will help identify similar prehistoric features elsewhere. Figure 3 - Extent of observed liquefaction 3

Preliminary GPS and InSAR observations (Figure 4) show that the ground onshore was permanently displaced up to 11 cm in the horizontal direction and up to 6 cm in the vertical direction. Based on this initial result there may be up to 6 cm of uplift in the New Brighton area with 1-2 cm of subsidence towards Brooklands. The latest data indicate there is little significant change in the ground level of flood prone areas around Aranui-Bexley. However, there may be local changes associated with liquefaction or spreading which have not yet been assessed. 3 SOURCE MODELS The Valentine s Day earthquake focal mechanism (inset Figure 4) shows oblique-reverse dextral fault motion. The faulting style and P-axis orientation of the mechanism is consistent with previous mechanisms observed in Pegasus Bay during the Canterbury earthquake sequence (e.g. Ristau et al. 2013). Initial analysis of the spatial distribution of aftershocks does not clearly delineate the orientation of the fault that ruptured. We also note, that variations in focal mechanism were observed in the largest (Mw 4+) aftershocks. Source modelling based on both geodetic displacements (measured by GPS and InSAR data) and GeoNet strong motion data, favour the approximately East-West orientation of the fault plane, with dip towards the south. Preliminary geodetic modelling suggests an approximately 8 km-long fault (see the surface projection marked as blue line in Figure 4) with ~30 cm of slip. A preliminary model based on strong motion data from 13 near-field stations is shown in Figure 5, suggesting a maximum slip of nearly 1 m. Figure 4 - InSAR and GPS data showing ground displacement associated with the earthquake. Arrows indicate GPS measured displacements and colours indicate displacements measured from InSaR data. The preliminary geodetic source model gives a best-fitting fault plane with surface projection indicated by the blue line. The focal mechanism of the Valentine s Day earthquake is shown to the right. Inset shows an example of data logged every second from permanent GPS station GWMR (2.5km NW of the epicentre). The site has been permanently displaced ~10 cm towards N, and ~4 cm towards E and up. However, during the earthquake, GWMR moved up to ~50 cm in the N-S plane and ~30 cm in the E-W plane, before reaching its final position. 4

Figure 5 - Preliminary source model based on strong motion data. Top left: Map view of fault slip and GeoNet strong motion stations (red diamonds); the yellow star is the GeoNet hypocentre. Bottom left: View of fault plane model looking north with slip indicated by colour scale (in metres). Right: Observed (black line) and synthetic (red line) velocity seismograms filtered 0.1-1Hz and computed for the slip model described below; the duration is 50 seconds. Values above the traces are the maximum observed absolute velocities (m/s). The results favour an East-West fault plane (approximately: strike 54, dip 65, rake 120 ). The modelled rupture starts at a depth of 15 km and propagates up to 4 km, with a maximum slip of nearly 1 meter. The estimated moment value is 7.9x10**17Nm. 4 GROUND MOTIONS New Zealand Modified Mercalli Intensity (Dowrick 1996) was approximately MM5 for much of Christchurch, but ranged up to MM8 for isolated suburbs, e.g. Scarborough and Phillipstown. These MMI values were extracted from a community intensity map based on GeoNet felt reports and derived following the method of Sbarra et al. (2010). The maximum recorded Peak Ground Acceleration (PGA) reached 0.36 g in the vertical direction at New Brighton (station NBLC) and 0.29 g in the horizontal direction at station PRPC in eastern Christchurch. The maximum recorded PGA was significantly lower than that recorded during the February and June 2011 earthquakes (over 2 g) and somewhat lower than that recorded during the Mw5.9 December 23 rd 2011 earthquake (0.7 g horizontal). PGA (horizontal) has also been estimated for the entire Christchurch region using ShakeMap in Figure 6. The ShakeMap software, developed by the USGS (Wald, 1999, Worden, 2012), and calibrated for New Zealand (Horspool, 2015) is used as the basis for estimating ground motion intensity at each location. ShakeMap combines observed ground motions recorded at strong motion stations with ground motion prediction equations (GMPE) to estimate ground motions across a region. ShakeMap provides estimates of peak ground acceleration (PGA), spectral acceleration at 0.3s, 1.0s and 3.0s, peak ground velocity (PGV) and MMI intensity as well as corresponding uncertainties. Near strong motion stations the uncertainty will approach zero, and will reach the uncertainty of the GMPE away from the stations. The ShakeMap for the Valentine s Day earthquake estimates PGA in excess of 0.15 for eastern Christchurch and as far north along the eastern shoreline as Kaiapoi. 5

Figure 6 - Map of estimated horizontal PGA during the Valentine s Day earthquake. Triangles represent the locations of GeoNet strong motion stations and are coloured according to actual observed recordings. Response spectra calculated from strong motion recordings from the Valentine s Day aftershock highlighted that some buildings may have experienced ground shaking exceeding a serviceability limit state event, and up to approximately 60% of an ultimate limit state event. These are likely to be stiff structures situated towards the east of Christchurch or near the foothills with structural period of less than 0.5 second. An important feature to note is that the ground motion history is relatively short duration ( 5 second) and thus significant damage is unlikely. Figure 7 presents a number of the response spectra compared to the ultimate limit state design spectrum for site class D condition. More comparisons are available at the NZSEE website (Link: http://www.nzsee.org.nz/information-anddata-on-valentines-day-earthquake/). The authors note that a more accurate building-specific comparison requires knowledge of i) the ground condition, ii) specific building period, and iii) the age of design and construction. A number of strong motion stations (e.g. NBLC New Brighton Library, HPSC - Hulverstone Drive Pumping Station) close to the epicentre also showed amplification of long period motion consistent with significant site effects and possible liquefaction. 5 AFTERSHOCK STATISTICS In the three weeks following the Valentine s Day earthquake, aftershocks ranged up to magnitude 4.5, with four aftershocks of magnitude greater than 4 recorded in the vicinity of the earthquake. The earthquake increased the short-term probability of further earthquakes greater than magnitude 5 in the Canterbury region (see discussion of the latest earthquake probabilities on the GeoNet website (http://info.geonet.org.nz/x/xieo). However, Figure 8 demonstrates that the Valentine s Day earthquake has only a small effect on the expected annual rates of such earthquakes in the context of the Canterbury sequence as a whole. The annual probability of further MM7 ground shaking (i.e. similar to the maximum experienced during the Valentine s Day earthquake) calculated in the week following the event ranged up to approximately 50% in parts of Christchurch and was highest in the eastern and central suburbs. 6

Spectral Acceleration (g) 0.9 5% Damped elastic acceleration response spectra 0.8 0.7 Thick blue line represents Design Spectrum from NZS1170.5 Site Class D, Z=0.3 R=1 Sp=1 Dotted black line represents 33% of the thick blue line (i.e. 33% code) 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 3 Period (s) DALS-N00E DALS-N90W GODS-N00E GODS-N90W OPWS-N00E OPWS-N90W SUMS-N00E SUMS-N90W DHSS-N00E DHSS-N90W LCQC-N00E LCQC-N90W MORS-N00E MORS-N90W SACS-N00E SACS-N90W GOVS-N00E GOVS-N90W Figure 7 - Response spectral accelerations calculated at a selection of GeoNet stations compared to design spectra in Christchurch. Note, that locations closest to the epicentre (NBLC and HPSC) not included here show higher levels of long period motion. Figure 8 - Effect of the Valentine s Day earthquake on the expected annual rate of magnitude > 5 earthquakes within a given ~5km-by-5km area in the centre of Christchurch. 7

6 CONCLUSIONS We present a preliminary report on the 14 th February 2011 Valentine s Day earthquake that struck a few kilometres offshore from eastern Christchurch. Physical impacts of this moderate-sized earthquake included rockfall from susceptible cliffs, localised liquefaction in parts of eastern Christchurch and Kaiapoi, and permanent ground displacement up to 11 cm in parts of the eastern suburbs. Preliminary source models based on geodetic and strong motion data suggest oblique-reverse slip of up to 1m or less on an ~8km-long fault plane oriented approximately ENE-WSW. Peak ground accelerations are estimated to have exceeded 0.15 g in the eastern suburbs of Christchurch, with the highest PGA of 0.36 g recorded in the vertical direction at New Brighton. Response spectra highlight that some buildings may have experienced ground shaking exceeding a serviceability limit state event, and up to approximately 60% of an ultimate limit state even, although significant damage is considered unlikely. This earthquake was an aftershock of the ongoing Canterbury earthquake sequence. Although the Valentine s Day earthquake increased the probability of earthquakes in the region in the short-term, the effect on expected annual rates of earthquakes greater than magnitude 5 was small in the context of the Canterbury sequence as a whole. 7 ACKNOWLEDGEMENTS We acknowledge the New Zealand GeoNet project and its sponsors EQC, GNS Science and LINZ, who provided essential data and support for this preliminary report. The GPS data in Figure 3 is derived from projects involving GeoNet, GNS Science, LINZ, University of Otago, Global Survey, GeoSystems and Eliot Sinclair. We are very grateful to Paula Gentle (LINZ) for her efforts in obtaining the GPS data. We also thank Neville Palmer for conducting GPS fieldwork. We thank J. Kupec, C. Mangos and C. Gibbons (Aurecon NZ Ltd.) and Mark Yetton (Geotech Consulting Ltd.) for their field observations and contribution to landslide assessment. Julian Thomson and Carol Smith were part of the Pines Beach liquefaction field team. M. Jacka and J. Carter were part of the Tonkin + Taylor liquefaction mapping effort. Silvia Canessa developed the Python code to obtain the community MMI values quoted here. This paper presents a brief summary of many key aspects of the earthquake science response, but many others have been involved in the overall response effort. REFERENCES Dowrick, D.J. (1996). The modified Mercalli earthquake intensity scale; revisions arising from recent studies of New Zealand earthquakes, Bulletin of the New Zealand National Society for Earthquake Engineering 29, 92 106. Gledhill, K.R.; Ristau, J.; Reyners, M.E.; Fry, B.; Holden, C. 2011 The Darfield (Canterbury, New Zealand) Mw7.1 earthquake of September 2010 : a preliminary seismological report. Seismological Research Letters, 82(3): 378-386; doi: 10.1785/gssrl.82.3.378 Horspool, N.; Chadwick, M.; Ristau, J.; Salichon, J; and Gerstenberger, M.C. 2015. ShakeMapNZ: Informing post-event decision making. Proceedings, 2015 Conference of the New Zealand Society for Earthquake Engineering, Rotorua, April, 2015. Kaiser, A.E.; Holden, C.; Beavan, R.J.; Beetham, R.D.; Benites, R.A.; Celentano, A.; Collet, D.; Cousins, W.J.; Cubrinovski, M.; Dellow, G.D.; Denys, P.; Fielding, E.; Fry, B.; Gerstenberger, M.C.; Langridge, R.M.; Massey, C.I.; Motagh, M.; Pondard, N.; McVerry, G.H.; Ristau, J.; Stirling, M.W.; Thomas, J.; Uma, S.R.; Zhao, J.X. 2012 The Mw 6.2 Christchurch Earthquake of February 2011 : preliminary report. New Zealand Journal of Geology and Geophysics, 55(1): 67-90; doi:10.1080/00288306.2011.641182 Sbarra, P., P. Tosi and V De Rubeis (2010). Web-based macroseismic survey in Italy: Method, validation and results, Natural Hazards 54, 563-581. Wald, D.J., Quitoriano, V., Heaton, T. H., Kanamori, H., Scrivner C.W., and Worden B.C. 1999. TriNet "ShakeMaps": Rapid generation of peak ground-motion and intensity maps for earthquakes in southern California. Earthquake Spectra 15, 537-556. Worden, C.B., Gerstenberger, M.C., Rhoades, D.A., and Wald, D.J. 2012. Probabilistic Relationships between Ground Motion Parameters and Modified Mercalli Intensity in California. 8