In October 2019, a spectacular landslide destroyed nearly 300m of State Highway 4 at a remote site near Te Ore Ore, about 20km south of Raetihi in the central North Island. The slip caused major disruption to the local community and to goods and tourist traffic through the area. Eleven weeks later a new temporary route across the slide was opened in time for the Christmas holiday season. This paper describes the landslide failure mechanism and sets out the measures adopted to manage ongoing landslide risk to the temporary crossing.
On Sunday 29 September last year cracking in the pavement of State Highway 4 was noticed by maintenance crews from network maintenance contractor Higgins Contractors (Higgins) who manage the route on behalf of Waka Kotahi NZ Transport Agency (Transport Agency). On the night of 1 October an earthflow originating from saturated ground in the floor of the valley covered the road during the night. This was cleared and a geotechnical inspection undertaken on Tuesday 2 October. Open tension cracks were found in the pavement both to the north and south of the earthflow (Figure 1). The road was closed following this inspection.
Figure 1. Tension cracking several days prior to the landslide
Figure 2. UAV imagery of Te Ore Ore Slip shortly after failure, October 2019
Figure 3. A road to nowhere…the edge of the carriageway at the north end of the landslide
Over the next day a large headscarp developed upslope of the highway. At some time during the night of 3 October a 12 hectare area failed rapidly toward the Mangawhero River, completely removing about 300m of highway and partially blocking the river channel (Figure 2 and Figure 3).
Figure 4. Site plan with pre-landslide imagery showing location of the 2019 landslide in relation to the original pre-historic landslide
The 2019 landslide was found to be within a much older and larger pre-historic landslide (around 32 Ha) that had previously failed north-east toward a bend in the Mangawhero River (Figure 4). The failed block is bounded to the south and east by two intersecting topographic lineaments. Sub-horizontal bedding in the underlying Tertiary siltstone may have provided a low angle basal shear surface. Similar landslides at even larger scales are apparent in the wider area and have formed as a result of rapid downcutting and incision of the drainage features in geologically recent time, causing removal of toe support to natural slopes.
Displacement of the older landslide block toward the north-west has left behind it a poorly drained “pull-apart” depression parallel to southern scarp (Figure 4). Springs and swampy ground are apparent in the floor of this valley.
The earthflow that preceded the 2019 landslide suggests that the slide mass had slowly become saturated over the past winter (or winters) and this is likely the primary cause of the failure. Satellite imagery commissioned by the Transport Agency also suggested a gradual build-up of soil moisture in the valley floor. Rainfall for the 5 days prior to the first observed movement was around 50mm, which is relatively unremarkable for the site and may be expected to occur on one or more occasions in a typical year. This suggests the landslide was not a direct result of a single rainfall event.
Figure 5. Post-landslide imagery showing October 2019 slip extents with principal elements highlighted
The lower section of the 2019 landslide comprises a debris fall zone (Figure 5). The base of the landslide is hypothesized to “daylight” above river level with material in the lower reaches of the landslide having “collapsed” into the river. Further upslope, movement was largely translational, with the ground surface remaining broadly intact but severally deformed through tension cracks, pressure ridges and scarps. Downslope displacement ranged from around 27 m just above the debris fall zone, steadily decreasing to 1-2 m at the upper extent of movement. Shallow secondary landslides have formed from the southern slopes due to removal of toe support as a result of the main slide. A small prominent hill in the centre of the valley floor represents a semi- intact block of siltstone that is being “rafted” within the slide debris.
Following the landslide, traffic was diverted to Fields Track (Whangaehu Valley Road), a detour of approximately 25 kilometers along rural roads unsuited to heavy vehicles. These had to make an even longer detour via State Highway 1. The landslide isolated the township of Raetihi from through-traffic and significantly affected access to the scattered rural community over a large area. This directly impacted school pupils, rural business, tourists and access to emergency services. Reconnaissance-level geotechnical mapping was carried out, with repeat photogrammetric surveys based on drone data capture initially used to assess the degree of ongoing movement. Once ground access on to the slip was available an array of survey prisms was set up to monitor ongoing ground deformation. Initial survey results indicated a section of the slip immediately upslope of the debris collapse zone was continuing to move at up to 10 mm per day (see Figure 8). No ongoing deformation was recorded in the upper reaches of the landslide.
An early assessment of options for reinstating SH4 concluded that there was no short-term solution to bypass the slip. Possible alternative alignments both east and west of the Mangawhero River were unable to provide required gradients for heavy vehicles without major earthworks. These options were further constrained by the presence of other landslides. Investigation works to assess the feasibility of stabilising the slip would also require many months to complete and even longer to implement.
In the absence of these alternatives and given the imperative to reopen the highway for the local community the decision was made to temporarily reinstate the highway across the slip. This involved rebuilding the highway slightly upslope of its original alignment requiring significant earthworks, surface drainage works, landscaping and pavement reconstruction. The adopted construction methodology sought to manage the effects of the works on stability as far as possible but did not include further measures to stabilise the landslide. This solution therefore required the risk to the construction team and later to road users to be closely monitored and proactively managed. The risk that the route might become untenable during or at any time after construction in the event of renewed instability was also acknowledged.
A key component of formalizing landslide risk management is the Trigger Action Response Plan (TARP). TARPs were used successfully during the reinstatement of SH 1 following the 2016 Kaikōura Earthquake.
The TARP sets out:
Triggers – events which indicate an increase in level of risk which requires further action. These may be physical observations of changed conditions, instrument responses or wider-field adverse events such as earthquake or intense rainfall. In some cases, triggers may also be anticipated. For example, a forecast for abnormally heavy rain may constitute a trigger.
Actions – a series of activities initiated by triggers to further investigate the level of risk and inform decision-making. The party responsible for each action is defined.
Responses – measures taken to directly mitigate elevated risk. The most extreme is temporarily closing the route but this may also include partial restrictions such as temporary speed limits, daylight only operations or restricted hours of operations. The response may also require additional monitoring and/or an increase in monitoring frequency.
Actions and responses are grouped into three TARP levels based on defined triggers:
Level 1 – normal operations in accordance with the temporary traffic management plan;
Level 2 – a state of heightened awareness requiring for example detailed inspection, additional monitoring, with the option of partial restriction of service.
Level 3 – closure of the route followed by further monitoring and investigations.
Key aspects of the TARP include:
- Triggers must be related to measurable criteria. Where these are quantitative, such as rates of movement or rainfall intensity, alert levels for the various responses must be defined. Where these are observational, such as observed ground movement on site, clear descriptive criteria need to be provided.
- Trigger levels need to be reviewed frequently as more is learnt about the behavior of the landslide and the level of “noise” inherent in the various monitoring instrumentation. If they are set too conservatively, false alarms will be generated which rapidly degrade the effectiveness of the TARP and can quickly cause stakeholders to disengage. Insufficiently conservative trigger levels obviously risk missing a key precursor to renewed landslide movement. As our understanding of the movements across the slip increased, the TARP trigger levels were adjusted accordingly.
- Every aspect of the TARP needs to be owned by a specifically identified person with at least one alternate provided in the event the primary point of contact cannot be reached.
- The TARP should define de-escalation processes and responsibilities as well as escalation (for example, who has to sign-off re-opening the road following a level 3 response?).
Figure 6. Drone imagery showing temporary reinstatement of SH4 just prior to re-opening (December 2019)
Figure 7. Southern section of the temporary alignment from one of the remotely monitored cameras
Temporary reinstatement and monitoring
Higgins, with delivery partners I.D. Loader Ltd and Goodman Contractors Ltd commenced construction of the temporary route on behalf of the Transport Agency on 19 November 2019. SH4 was reopened to public traffic as a fully chipsealed, two lane carriageway on 20 December, in time for the holiday period (Figure 6 and Figure 7). A 30 km/h speed limit was in place with temporary lighting provided using portable plant. The design of the works was by Higgins, with design review and TARP development by Beca.
Monitoring throughout the construction period comprised:
- Daily re-survey of the surface prism array and re-calculation of vectors of movement.
- Daily walk-over pre-start inspection of the working area for any evidence of cracking or deformation.
- A site rain gauge, supplemented by an existing Horizons gauge located just 2 km to the south of
- Establishment of a series of waratah stake extensometers across key tension cracks
- Regular geotechnical inspections.
- Monitoring focused on two areas:
• The southern section of the realignment due to post-landslide movement with potential for damage to the carriageway, and
• The slopes above the northern end of the alignment where extensive tension cracking had occurred during the main landslide event.
During construction, movements of up to 10 mm per day continued at the southern end of the route . No movement was recorded on the northern slopes. A number of alerts occurred under the construction TARP requiring further investigation, but only one day’s work was lost due to rainfall exceeding the Level 3 threshold.
A revised regime of monitoring was put in place once the road was open, including:
- The site was manned full time and drive-over inspections of the carriageway were done every 2 hours.
- Eight automated surface extensometers and five tiltmeters were deployed at key location across the site. The instruments are connected via telemetry
to a satellite communications unit which provides near real time data availability on a Beca-hosted instrumentation portal. This system generates automatic alarms if pre-set movement thresholds
- Two remotely accessed video cameras were installed providing full visual coverage of the route within the landslide limits. These can be viewed remotely in real time and can also be panned and zoomed to inspect features of interest.
- A new rain gauge was set up to provide hourly data via telemetry.
- Regular geotechnical inspections continued.
Figure 8. Horizontal movement recorded from the prism array since construction commenced. Points showing movement are all located in the zone of post-event movement shown in Figure 5. No movement beyond survey measurement error has been registered since the beginning of February.
A new TARP was developed to cover road operations, considering the revised monitoring programme above and learnings from the construction period. These included:
- There was no evidence that the slip had responded to rainfall over the construction period. The TARP trigger levels for rainfall were therefore increased.
- Survey monitoring had shown a steady decline in the rate of movement over the construction period (Figure 8) allowing survey frequency to be reduced to weekly.
- No evidence was seen to suggest that the construction earthworks had any effect on the rates of slip movement.
On 24 December a crack was noted in the new seal toward the southern end of the route. A geotechnical inspection was immediately undertaken (under TARP Level 2) and additional monitoring stakes were installed across the crack. While the crack was able to be mapped east and west of the road, the movement rates were found to be consistent with the ongoing survey monitoring results. The crack defines the upslope extent of the zone of post event movement shown on Figure 5. Given the consistency in movement rates, the road remained open and there was no interruption to service. The crack continues to be monitored through telemetered surface extensometers and regular geotechnical inspections to give the operations team the ability to close the road at short notice and maintain safety of the travelling public.
The operational TARP was further revised after approximately two months operation. During this time landslide displacement had dropped below the detection threshold for the survey method and none of the other instrumentation was recording any movement. Similarly, no effect was recoded from the Magnitude 5.4 earthquake which occurred 50 km northwest of Paraparaumu on 25 January (135 km from the site). The coming winter will provide further opportunity to observe any response to seasonally elevated groundwater levels and any learnings will be incorporated in later versions of the TARP.
The monitoring and TARP will remain in place until the solution for the permanent reinstatement of SH4 is developed – work which the Transport Agency is currently progressing.
The decision to temporarily re-instate a key transport route across a recently active landslide required timely information on the landslide’s behavior and a clear framework for risk management and decision making. To date the insights provided by remote and on-site monitoring coupled with an active TARP have allowed SH4 to be safely re-established and operated at this site. In the event of any change in landslide behavior the monitoring regime and the TARP can be quickly updated. In the right circumstances this approach offers a potential way to manage natural hazards that cannot immediately be circumvented or remediated.
The authors wish to thank Waka Kotahi NZ Transport Agency and Higgins Contractors for permission to publish this article. Valuable assistance and peer review have been provided by the North Canterbury Infrastructure Rebuild Alliance (NCTIR) and the Transport Agency’s technical review team. Information exchange with Mount Messenger Alliance which is investigating options for permanent reinstatement of SH4 is gratefully acknowledged.