Introduction
The M7.8 Kaikoura earthquake of 14 November 2016 generated very large peak ground accelerations (PGA’s) that resulted in the closure of the coastal sections of both State Highway 1 and the Main North Railway Line due to slope failures originating from above the road and rail corridor at more than 50 locations north and south of Kaikoura (Figure 1).
The North Canterbury Transport Infrastructure Recovery (NCTIR) was set up by the government under the Hurunui/Kaikoura Earthquakes Recovery Act 2016 to repair and re-open the earthquake-damaged road and rail networks between Christchurch and Picton by the end of 2017. NCTIR is an alliance partnership between the NZ Transport Agency, KiwiRail, Fulton Hogan, Downer, HEB Construction and Higgins.
The Main North rail line was reopened in September 2017 for limited services. The State Highway was re-opened (with restrictions) in December 2017 and fully re-opened on 30 April 2018.
A large number of engineering geologists, geotechnical engineers and (some) geomorphologists drawn from many different organisations are working on the recovery project. This short summary, which reflects the work of the whole team, outlines our approach to assessing the landslide hazard to the road/rail corridor and the key understandings that have developed from this work.
Figure 1: Location of coastal areas affected by landslides and slope failures burying SH1, Ohau Point
Earthquake effects
The 2016 Kaikoura earthquake caused significant damage over a very large area, with large debris slides on the coastal slopes closing both State Highway 1 (SH1) and the Main North Rail Line (MNL) between Picton and Christchurch (eg. see Figure 1).
Once post-earthquake LiDAR information was available it did not take long to realise that many old landslides and creeping slopes are concealed in the bush and scrub-covered slopes above the transportation corridor (Figure 2).
Helicopter inspections immediately after the earthquake showed that many of the ridge crests were intensely shattered and dilated due to topographic amplification effects from this and previous earthquakes. Many of the earthquake-induced landslides had originated in these areas.
DEM’s based on LiDAR data identified that most of the large failures occurred in accumulations of debris from large old (ancient) paleo-landslides and/or thick accumulations of landslide-derived colluvium, or as failures of parts of the adjacent ridges (eg. Figure 3). Some of the paleo-landslide features (or parts of them) have been highlighted by tension cracks that developed along their margins as a result of the Kaikoura Earthquake; others show little sign of response to the earthquake (eg. see Figure 2).
In addition, many of the debris/colluvium slopes have localised earthquake-induced tension cracks, scarps and/or shallow failures, in some cases reaching to the top of the ridge above the transportation corridor.
Figure 2: Site SR30A is an old landslide at Tunnel 8 portal (Kaikoura South) that did not react to the earthquake. The hillshade model shows geomorphic detail and clear indications of on-going creep movement.
Current situation
The landslide materials have been cleared from the road and rail formations, structures have been repaired or replaced, protective works have been designed and constructed at many locations, and sections of the road and rail have been, or are being, realigned to increase resilience of these important transportation links (for examples, see Figures 4 and 5).
Figure 3: Slip area P3, Okiwi Bay (top) has secondary features that did not reach the highway. Site SR14 at (above) is an example of localised failure of an old landslide
Since the earthquake quite small rainfall events have initiated debris flows from the secondary failures and erosion on the primary landslide slopes. The debris flows have deposited new debris on the road and rail links and discharged via culverts onto the foreshore. The rainfall associated with ex-Cyclones Cook and Debbie (April 2017) emphasised that debris flows will continue to be a problem until source areas revegetate. This concern was reinforced by ex-Cyclone Gita in February 2018 (see above).
Hazard and Risk Assessments
Field mapping following the earthquake revealed extensive ground cracking, tension cracks and scarps within, around and in some cases well beyond these features (for example, see Figure 3). The mapping also highlighted that despite extensive damage, only a small proportion (~5% by area) of the slopes had actually failed.
The focus of the original mapping programme was on the large landslides that blocked the road and rail corridor. This mapping identified additional pre-existing (paleo) landslides that had not failed, and a number of ‘secondary’ failures that had not reached the transport corridor but potentially threatened the road and rail corridor in the event of future movement. It was decided that these features should also be assessed as the risk to rail and road users could be significant. By reviewing a number of hillshade models, we identified over 50 geomorphic features that have the appearance of paleo-landslides that could affect the road and rail corridor. These features range from 10,000 m2 to 100,000 m2 in area and are distributed throughout the project corridor between Oaro and Clarence.
Figure 4: Site P8, December 2016 and in September 2017 after clearing debris and realigning rail. Temporary road alignment shown.
Figure 5: Site P2, December 2016 and in October 2017 after clearance. Road realignment underway. Old road is used for temporary access, protected by barrier.
Landslide classification
The geological mapping undertaken by NCTIR since the 2016 Kaikoura Earthquake has faced many challenges. Access to many of the sites is extremely difficult because the slopes are steep, heavily vegetated and often mantled with loose rock. Helicopters were used to transfer teams to and between sites to improve the efficiency of mapping excursions. Efforts were concentrated on headscarp and toe areas because many of the landslides are too steep to safely traverse on foot.
Field mapping utilised the mobile phone or tablet based ArcGIS Collector App to collect geological data. The system allows users to find their position on aerial models stored on their mobile device. The device does not require mobile phone coverage and relies on the GPS satellite network to calculate user location. This technique was particularly useful for mapping the shape and distribution of tension cracks and scarps beneath heavy bush cover and for recording structural geological data.
The combination of DEM analysis and field mapping recognised extensive areas of past landsliding and mass movement as well as the recent earthquake effects. This has led to the development of a landslide classification scheme in which the pre-existing landslides have been classified as follows:
- Ancient landslide: The slope had previously failed and has a debris pile at the base, or
- Mass movement: The slope shows evidence of long term creep or repeated minor movements rather than obvious discrete slope failures (eg. Figure 2).
The earthquake-induced slope failures affecting the road/rail transportation corridor along the north coast section have been classified as:
- Primary landslide: Landslide debris reached and covered the road and/or rail corridor (eg. Figures 1, 4 and 5).
- Secondary landslide: Landslide debris did not reach the road and/or rail corridor in significant volumes, but has potential to significantly affect the corridor (eg. see Figure 3).
Age of landsliding
Rapid coastal erosion was likely occurring as sea level was rising until about 6,000 years before present. This toe erosion is inferred to have had a major destabilising effect on the coastal slopes, initiating the landslides and keeping them active. Relative uplift of the coastline in the last 6,000 years is estimated to be about 18 m. Hence, within the last 6,000 years, tectonic uplift (associated with earthquakes similar to the Kaikoura earthquake) has stranded the landslides above the eroding effect of the sea and removed the destabilising effect. Consequently, the landslides are now much less active than they have been in the past, and their behaviour is driven by climatic and tectonic events rather than coastal erosion.
Identification of landslide activity from LiDAR data
LiDAR datasets were collected by NCTIR in December 2016 and May 2017. These have been used to assist with evaluation of slope instability hazards and in re-design of the road and rail corridors. Digital terrain models and various hillshade models developed by the NCTIR GIS and mapping teams have allowed the identification or improved understanding of areas of previously hidden rock outcrop or colluvium, possible headscarps, tension cracks, alluvial terraces and debris flow channels.
An initial screening assessment of the identified landslide features was undertaken using the simple matrix presented as Figure 6. The objective of this screening assessment was to prioritise field mapping efforts and ensure that any slope that is likely to present high hazard to the road and rail corridor is inspected by a mapping team. In particular, a LiDAR difference model developed using the December 2016 and May 2017 LiDAR data sets has been used to identify any significant changes in the slope morphology that could indicate on-going, post-earthquake landslide movement.
This approach also allowed a cross-check of the risks posed by the slopes in the ARL and/or KiwRail Slope Hazard Rating assessment as areas of erosion and debris accumulation clearly showed up on those slopes sensitive to rainfall effects (Figure 7) and the LiDAR difference models also have the potential to identify any areas with significant ongoing mass movement.
Figure 6: Likelihood and priority assessment matrix for slopes based on hillshade maps and LiDAR differences.
Assessment
- A Highest risk/ priority
- B Moderate risk/priority
- C Low risk/priority
- D Very low risk/priority
Hillshade categories
- H1 Sharp morphology and/or active
- H2 Moderately sharp morphology and/or minor recent activity
- H3 Subdued morphology and/or not recently active
Difference map categories
- D1 Evidence of widespread surface change
- D2 Evidence of localised surface change
- D3 No clear evidence of surface change
The slope shown in Figure 7 is classified as H1/D2 (B – moderate) in terms of the criteria outlined in Figure 7. However, because the road/rail corridor is over 50m from the toe of the slope, the site is assessed to pose a low risk to users of the transportation corridor.
Effects of Cyclone Gita
Ex-tropical Cyclone Gita struck the Kaikoura Coast on 20-21 February 2018 and rainfalls ranging from 100 to almost 250mm were recorded over a 24 hour period. At Rakautara, 215 mm of rain was recorded over an 18 hour period, which is considered to have a return period on the order of 1 in 30 to 1 in 50 years based on NIWA’s High Intensity Rainfall Design System (HIRDS).
There was a 3 to 4 hour period of intense rainfall (with a peak intensity of 45 mm/hr) and the 6 hour rainfall accumulation was considered to be on the order of a 1 in 80 to 1 in 100 year event.
Most of the slopes where cleanup and stabilisation works had been undertaken survived the event with remarkably little damage but large debris flows originated from remote sources in streams crossing the transport corridor at a number of locations, particularly north of Kaikoura. The most dramatic of these was at Jacobs Ladder where a very large granular debris flow inundated both the rail and road and buried them with debris (ranging from silt and fine sand to boulders >1 m diameter) that was many metres thick in places, and the MNL was pushed out across the road (Figure 8). The deposit is estimated to have been around 100,000 m3 with approximately 18,000 m3 requiring removal to expose the road and rail. The road was closed for 10 days and the rail for 15 days. The nearby Black Miller Stream was similarly affected by Cyclone Alison in 1975, at which time a train was partially buried in the debris (David Bell, pers. comm, March 2018; see Figure 9).
Figure 7: Example of aerial photo and hillshade map showing LiDAR differences, Okiwi Bay. The dark blue in the difference map represents erosion, the yellow and orange areas indicate deposition.
Figure 8 (left): Debris flow covering road and rail at Jacobs Ladder, Cyclone Gita, Feb 2018.
Figure 9 (right): Debris flow at Black Miller Stream, Cyclone Alison, March 1975.
Conclusions
Most of the old landslides identified along the coastal corridor were probably creeping before the earthquake. As a result of the earthquake they have been subjected to shallow failures affecting parts of them (or the adjacent ridges). Although damaged, they are not thought to be moving significantly faster than before, which is consistent with international experience.
The hillshade DEM models have allowed good geomorphological mapping of the landslides and other slopes along the Kaikoura coast, and we have a level of engineering geological mapping that is sufficient to allow an assessment of hazard and risk for most slopes.
Despite extensive engineering works including debris removal, stabilisation and protection works and/or realignments to improve resilience, on-going instability of the slopes above the Kaikoura Coast transportation corridor is expected due to the damage caused by the earthquake. The ongoing instability is most likely to consist of rockfalls, soil falls and debris flows triggered by rainfall events. It is expected that for frequent events these can largely be contained by the protective works undertaken, but it is recognised that there will be larger, less frequent events (including future earthquakes) that will result in road and/or rail closures for varying lengths of time.
Assessment of the mapping and LiDAR information, supported by rockfall modelling for engineering design, has enabled the slopes to be classified according to their relative activity and risk to the road and rail links. The asset owners, KiwiRail and NZTA, have specified acceptable outage levels and NCTIR will develop observational monitoring and maintenance strategies to manage such events.
