Search

As part of a GNS Science (GNS) landslide response to Cyclone Gabrielle (12–16 February 2023), GNS scientists were asked by the National Emergency Management Agency (NEMA), via a request from Gisborne District Council (GDC), for help to investigate the Tokomaru Bay / Mangahauini landslide dam (‘Mangahauini dam’). The Mangahauini dam formed during Cyclone Gabrielle when a landslide blocked the Mangahauini River, approximately 3 km upstream of Tokomaru Bay township. The landslide dam was identified by GDC as being a potential hazard that could pose a risk to life, property and infrastructure downstream if an outburst flood (from a sudden dam breach) were to occur. The hazard that such dams pose can be highly uncertain given the large number of factors that influence their stability and subsequent debris inundation if/when they breach. The landslide response consisted of two site visits, surveying and landslide dam-breach inundation simulations. Prior to an initial aerial reconnaissance survey of the dam on 20 February 2023, the dam had partially breached and a spillway had formed. Nonetheless, following this visit, numerical modelling was undertaken to indicate the extent of dam-breach inundation if the remaining dam material and lake were to fail suddenly. An additional field survey was carried out on 3 March 2023 and included Global Navigation Satellite System (GNSS) and Remotely Piloted Aircraft System (RPAS)-mounted LiDAR (Light Detection and Ranging) surveys of the landslide, landslide dam and lake to estimate their volumes more accurately. Dam-breach models were re-run with the updated volume estimates, and the results were provided to GDC. At the time of surveying (3 March 2023), the Mangahauini River was flowing through the dam along a relatively wide spillway, partly developed along State Highway 35, and there was only a small, impounded lake behind the dam with an estimated volume of ~80,000 m3 . Based on our field surveys and numerical simulations, two potential end-member scenarios regarding the hazard (not the risk) posed by the dam were identified: 1. Lake continues to gradually drain via the spillway through the dam – the current scenario, and thought to be the ‘best case’ scenario, with the following conditions and impacts:The lake continues to drain through the existing channel without sudden and complete failure of the dam and lake; Water gradually erodes and incises the existing dam debris and spillway; Floodwater and debris inundation downstream outside of active river channel less likely to occur. 2. Re-damming of the river due to remobilised landslide debris or wood, perhaps in a future storm – thought to be the ‘worst case’ scenario, with the following conditions and impacts: Up to 320,000 m3 of potentially unstable debris remains on the slope above the Mangahauini River valley, and; If this debris from the slope above the dam or farther upstream were to remobilise and re-dam the river, then the lake could increase in size, which; Could result in potential for larger (rapid) dam failure, which could lead to inundation downstream by dam-breach floodwater and debris; This scenario is highly uncertain, as it depends on the volume of debris remobilised, which affects the volume and geometry of any potential dam and impounded lake, and the (unknown) stability of the dam material.

Capture of accurate, quantifiable and repeatable data – typically ‘point clouds’ and their derivative products, such as topographic surfaces – are needed to monitor coastal cliff retreat. Currently, rates of coastal cliff change are often spatially and temporally averaged. The purpose of this report is to demonstrate how Remotely Piloted Aircraft System (RPAS)-mounted Light Detection And Ranging (LiDAR) can be used to capture these data and how the data can be processed to create topographic surface models that can be used by scientists and non-professionals alike. The aim of this report is to develop methodologies and workflows for: (1) capturing and processing RPAS-mounted LiDAR data; (2) using the processed data to create surface models; (3) creating change models by comparing the surface models from multiple RPAS-mounted LiDAR surveys captured at different times ‘epochs’; and (4) utilising the change models to calculate the locations and volumes of the changes, as well as the magnitude-frequency of such volume changes. These methods and workflows were then tested and further developed by applying them to a 1.1 km section of coastal cliff at Cape Kidnappers, Hawkes Bay. In addition to this work, we also compared the surface models developed from RPAS-mounted LiDAR with those developed from other survey techniques, including Airborne Laser Scanning (ALS), Terrestrial Laser Scanning (TLS) and RPAS-mounted photogrammetry. Cape Kidnappers in Hawkes Bay was selected as the site to test and further develop the workflows, as many cliff collapses have occurred from these cliffs and people have been hit and injured by debris falling from these cliffs. The walk along the beach below the cliffs is famous, as it allows access to a Gannet colony, which many thousands of people a year visit. Hastings District Council (HDC) and the Department of Conservation (DOC) are responsible for managing the cliff collapse hazards, and other hazards, along the beach, and the risk such hazards pose to people. This report will be presented to both HDC and DOC upon completion. Methodology of how to capture and process RPAS-mounted LiDAR data is discussed, as well as how the data can be processed to present change models, volume changes and magnitude-frequency graphs. Two lines were flown in parallel to the cliff face at 7.5 m/s; this was determined as the most effective way to gain coverage with both RPAS-mounted LiDAR and imagery. The data were ‘cleaned’ by removing data artifacts and removing vegetation to produce a ground model; this was done using Leica Cyclone 3DR. Surface models were created from these ground models, and change analyses were completed between the two data epochs. One change model shows overall change (using Leica Cyclone 3DR) and one shows individual rockfall events (using the Rockfall Activity Morphological Bigdata Optimiser [Rambo]). RPAS height and speed tests were completed on site to confirm the manufacturer’s specifications, and results were tabulated. Sensitivity analyses were tested to establish the best settings to use in Rambo for this project, and it was determined that this may be necessary to do with all future locations where Rambo is used. Advantages over other remote-sensing techniques, such as RPAS-based photogrammetry, ALS and TLS are explained, as well as what limitations come with RPAS-mounted LiDAR. It is concluded that RPAS-mounted LiDAR is the superior solution for medium-scale (between 0.1 and 5 km2) sites with complex topography, due to the coverage and point density achieved and the speed at which data can be captured. (The authors)

The Kāpiti Coast and Horowhenua districts are areas of rapid growth and development with expansion of housing and construction of major traffic infrastructure. Extending through these regions is the 60 km-long Northern Ōhāriu Fault, which is thought to be the northern extension of the Ōhāriu Fault. Little work on the Northern Ōhāriu Fault has been done since the early 2000s and given the potential for continuous ruptures along the Ōhāriu and Northern Ōhāriu faults, as well as increasing development in the Wellington-Levin corridor, further study is warranted. In this report we present the results of two paleoseismic trenches and a slip rate pit excavated across a branch of the Northern Ōhāriu Fault and an offset alluvial terrace, respectively. The work has been funded by the It’s Our Fault programme. Evidence of earthquakes spanning the Holocene and into the Late Pleistocene is identified from two paleoseismic trenches. Radiocarbon dating in one trench constrains the ages of the two most recent earthquakes to 120–273 cal yr BP and 1063–1271 cal yr BP and these are used to develop a preliminary recurrence interval of 400–2,300 years. Individual earthquakes in the other trench cannot be dated but suggest the recurrence interval is closer to the longer end of this range. At this stage, a slip rate cannot be derived from the pit excavated in an offset terrace due to uncertainty in when displacement began. The earthquake timings and recurrence interval of the Northern Ōhāriu Fault determined here are in good agreement with that of the Ōhāriu Fault. The results of this work suggest that either the Northern Ōhāriu and Ōhāriu faults form a single continuous structure along which earthquakes can rupture (with a preliminary maximum magnitude of Mw 7.9), or that earthquake activity on each fault is highly influenced by the other causing both faults to fail in earthquakes together or very close in time. Furthermore, earthquakes on the Northern Ōhāriu and Ōhāriu faults overlap in time with earthquakes on the Wellington and Rangatira faults, confirming earlier suggestions that earthquakes in the lower North Island may not be restricted to single faults and instead may occur as multi-fault ruptures. Results described in this report are preliminary and will be developed further with additional analysis and age modelling. However, these results demonstrate that the Northern Ōhāriu Fault is more active than previously thought and have implications for multi-fault rupture in the lower North Island, highlighting the importance of further research to understand the earthquake timing of this and other faults in the area. (The authors)

The New Zealand Paleoseismic Site Database (new term, new database) contains paleoseismic data (grouped into Slip Rate, Earthquake [EQ] Timings and Recurrence Interval [RI], and Single-Event Displacement [SED]) collected at specific sites along active faults throughout New Zealand. The database was developed as part of the New Zealand National Seismic Hazard Model 2022 Revision Project (NSHM 2022). The primary purpose is to compile paleoseismic data at specific sites to be used either as inputs into, or to constrain/validate outputs from, the Seismicity Rate Model. This report describes the purpose and design of the database, as well as the compilation process and contents of the first edition (Version 1.0). The New Zealand Paleoseismic Site Database has two components: a Microsoft Excel spreadsheet, which is a stand-alone database and the focus of this report, and a Geographic Information System (GIS) feature class dataset that is a subset of, and is intended to be entered back into, the AF.Points (Active Fault Point features) layer in the Active Faults Database of New Zealand. The Excel Slip Rate worksheet was initially adapted from the UCERF3 Geologic-Slip-Rate Data spreadsheet, and the EQ Timings RI and SED worksheets were adapted from the Slip Rate worksheet. Each worksheet has a high-level division into two sets of attributes – Fault Data and Site Data. The Fault Data attributes pertain to faults in the New Zealand Community Fault Model (CFM) and are the data most likely to be used for the Seismicity Rate Model. The Site Data component includes the site-specific data, and a site is defined as a location where paleoseismic data, generally field data, has been obtained. A site must therefore have a grid reference, but some published combined records are also included where data from two or more sites have been aggregated (e.g. a composite earthquake record derived from two or more trench sites). Most rows in the spreadsheet are therefore Site Data, but, in the EQ Timings RI worksheet, each row is an inferred earthquake. The definitions, formats and guidelines for compiling each attribute in both the Excel spreadsheet and GIS feature class are described in a companion Data Dictionary (Litchfield 2022). Version 1.0 of the New Zealand Paleoseismic Site Database was compiled by NL, JH, RM, RL and GC in 2020 and 2021, with input from RVD and other members of the New Zealand paleoseismology community. Version 1.0 contains both published and unpublished data, mostly onshore, and many sites were relocated using high-resolution Light Detecting and Ranging (LiDAR) Digital Elevation Models and orthophotos. The total number of sites is 2136, with 68 combined records, which have a reasonable geographic spread, particularly across onshore faults with slip rates of ≥1 mm/yr. The Slip Rate worksheet contains 862 sites situated on 189 CFM faults. The EQ Timings RI worksheet contains 304 sites and 953 records (earthquakes and combined records), situated on 99 CFM faults. The EQ Timings RI worksheet also includes Last Event (314) and previously documented (Reported) RI (100) records. Ninety-nine EQ Timings sites have three or more earthquakes that are currently being used for RI calculations in the NSHM 2022. The SED worksheet contains 970 sites and 17 combined records, situated on 90 CFM faults. The majority of SED sites are field-based displacement measurements for historical earthquakes, dominated by the 2010 Darfield and 2016 Kaikōura earthquakes. Version 1.0 is considered as complete as possible in the time available for data compilation, but there are known data that could be compiled in future versions, including student theses and data currently being obtained and published. New versions are contingent upon funding but could be partial (e.g. regional) or full (~5 yearly?) updates. Two known issues that could be addressed in future versions are calibration of radiocarbon ages with consistent calibration curves and inclusion of more SED data from cumulative displacements. (The authors)

The goal of the Stability of Land In Dynamic Environments (SLIDE) research project is to improve the resilience of New Zealand’s buildings and infrastructure through better knowledge of the behaviour of slopes and the development of strategies for more robust remediation approaches. One aspect of the SLIDE research is the development of a geomorphology map for terrain within the Wellington urban area in order to help identify those slopes that have been anthropogenically modified and are subject to failure. Many hundreds of slope failures affect Wellington’s roading network each year, and most of these failures occur on slopes that have been modified by urban development. One of the SLIDE project’s aims is to better understand how anthropogenically modified slopes may perform in future strong earthquakes and in heavy rain events. Recent heavy rain events and earthquakes, especially the M7.8 14 November 2016 Kaikōura earthquake, have underlined how vulnerable some of Wellington’s slopes may be in such events. However, it should be noted that not all anthropogenically modified slopes are potentially unstable and likely to fail in future earthquake and/or rain events. This report presents Version 1.0 of the geomorphology maps and the methodology used to carry out the mapping. We have classified the geomorphology of the Wellington urban area (the study area), at a regional scale of nominally 1:500, into two main layers: 1) surface morphology and 2) near-surface materials. An additional third layer comprising anthropogenically modified ground, subdivided into: a) cut slopes and b) fill bodies, was also created. Multiple datasets were used to complete this work, including historical and recent aerial photographs, LiDAR-derived digital elevation models and photogrammetrically-derived digital surface models, which were subtracted from each other to create surface difference models. The interpretations were selectively ‘ground truthed’ using both field assessments and Google Earth ‘Street View’ imagery. The surface difference models and aerial photographs were successfully used to identify the primary areas of anthropogenically modified ground within the study area. The oldest suitable photo survey available was undertaken in 1938, and this covers the Wellington City to Miramar Peninsula area. The remainder of the study area from the city westwards and northwards is covered by a photographic survey undertaken in 1945. These surveys, along with other historical photographs, form the ‘baseline’ for our interpretations of landscape modification. These areas of modified ground were interpreted as being mainly due to slope cutting (negative surface changes) and the formation of fill bodies (positive surface changes) for the construction of buildings, roads and other infrastructure. The approximate date of construction of the mapped cut slopes and fill bodies has also been estimated, primarily using the multiple epochs of aerial photographs available for the study area. Our mapping is limited in time to the last available aerial photo dataset (2013), and modifications since this time are not captured in Version 1.0. Other anthropogenic features, such as retaining walls, were also identified. Although several thousand retaining walls were mapped, identifying all retaining walls within the study area from remotely sensed data, their type of construction and whether they were engineered or not, was not possible and was outside the scope of this research project. The mapping presented in this Version 1.0 of the geomorphology database is an interpretation that is designed to support general concepts of how anthropogenic landscapes differ from those of the natural environment. The digital geomorphological layers presented here have subsequently been used in the assessment of landslide hazards within the study area (e.g. Massey et al. 2019). (auth)

The Auckland region does not have a historic record of any significant tsunamis, but it is exposed to a number of local, regional and distant tsunami sources. Of these potential sources, the largest tsunamis are expected to be generated by large earthquakes on the Kermadec Trench. Tsunami models of large Kermadec Trench earthquakes (M 8.5–9.4) suggest parts of the eastern coastline of the Auckland Region could be exposed to tsunamis with wave amplitudes of up to 10 m on Great Barrier Island and 1–5 m on the mainland coastline. The primary objective of this EQC project is to contribute to the calibration of tsunami modelling by identifying potential paleotsunami deposits in the Auckland region that can inform us of the run-up heights, inundation distances and recurrence of tsunamis in the prehistoric period (pre- AD 1850s). This report presents a review of existing paleotsunami information in the Auckland region, methods and results of new field studies on paleotsunami in the Auckland region (primarily on Great Barrier Island) and outlines future steps toward a better understanding of tsunami hazard in Auckland and the upper North Island. We reviewed evidence for paleotsunamis at 18 sites within the Auckland region and our review found three sites (Tawharanui, Whangapoua Beach and Harataonga Bay) have robust evidence of paleotsunami. The dating of the inferred paleotsunamis at all sites is relatively poor, and it is currently hard to evaluate if there are temporal correlations (similarities in age) between the records. To undertake new field studies of paleotsunami we evaluated 12 coastal areas and selected sites based on previous paleotsunami research, tsunami modelling, suitability of the coastal depositional environments and accessibility of the site. Field reconnaissance was undertaken at 8 coastal sites but only two had likely evidence of past paleotsunamis, these were the previously identified sites of Whangapoua Beach and Harataonga Bay (Tawharanui could not be revisited). Further data were gathered at Whangapoua Beach and Harataonga Bay to better constrain the age of the inferred paleotsunami deposits. Most sites on the Auckland mainland did not have ideal depositional environments for capturing and preserving paleotsunami sediments, so the lack of paleotsunami information on the mainland is more a reflection of the environment than lack of past tsunami inundation. We recommend future research to review paleotsunami records in the neighbouring regions of Northland and Coromandel Peninsula, coupled with multidisciplinary paleotsunami field studies with iwi and archaeologists. We also recommend a general investigation to evaluate the potential of sheet gravels within sand dunes as paleotsunami indicators, and tsunami modelling at specific sites to understand local effects on tsunami amplification. Further field studies at three sites on Great Barrier Island and two sites on the Auckland mainland are recommended in order to improve the age precision and reliability of the Auckland paleotsunami record. The coastline of Auckland and its neighbouring regions offer our most promising sites to better understand the size and frequency of large to great Kermadec Trench earthquakes and this information could be of critical importance for understanding tsunami risk in New Zealand. (auth)

The MW 7.8 14 November 2016 Kaikoura Earthquake generated more than 20,000 mapped landslides and about 200 significant landslide dams. Besides the immediate hazard from the landslides, cracked ground, landslide debris and landslide dams also pose longer-term risk to infrastructure, because if the landslides and debris remobilise and the dams breach, they could generate future debris flows and floods. These in turn, could sever transport routes and further damage infrastructure, adversely impacting the post-earthquake economic recovery of the region. The goal of this short-term project was to collect perishable (ephemeral) data on landslides and landslide dams generated by the earthquake. Currently there are more than 20,000 landslide source areas in the landslide inventory. Key findings from the landslide inventory are: 1) the number of large landslides (with source areas ≥10,000 m2) triggered by the Kaikoura Earthquake is fewer than the number of similar sized landslides triggered by other similar magnitude earthquakes in New Zealand; 2) the largest landslides (with volumes from 5 to 20 M m3) occurred either on or within 2,500 m of the 24 mapped faults that ruptured to the surface; and 3) the landslide density within 2,500 m of a mapped surface fault rupture is as much as three times higher than those densities farther than 2,500 m from a ruptured fault. Around 200 significant landslide dams generated by the earthquake have now been mapped and combined with data from past New Zealand and overseas landslide dams. These data have been used to develop a regional-scale, empirically-based tool to assess the post-formation likelihood of dam failure (breaching), which can be used for future landslide dam generating events. Many of the larger dams have also been investigated in detail and dam-breach models have been calibrated from back-analysing their failure. (auth)

An approximately 25,500 m³ debris avalanche occurred on the 23rd January 2019 from the coastal cliffs of Cape Kidnappers, Hawke’s Bay. The debris avalanche was observed by multiple bystanders, and seriously injured two tourists. The tourists were walking along the beach, which forms the public accessway to the Gannet Colony at Cape Kidnappers – a popular tourist attraction in the Hawke’s Bay. The cliff collapse occurred within a conglomerate unit located in the upper 50 m of the cliff. The site has been the location of previous failures with a significant debris avalanche occurring from the lower half of the cliff in the 2015. The remnants of the 2015 debris deposit were still present at the base of the cliff on the 23rd January 2019, with the 2015 deposit acting as a ramp, allowing the 23rd January 2019 debris avalanche debris to travel further. ‘Small’ precursory rockfalls were observed sporadically through the week prior to the 23rd January 2019 event. There was no discernible trigger for the debris avalanche, with no seismicity and limited rainfall recorded in the week prior. Marine erosion at the toe of the slope may have been a contributing factor in the 2015 failure. The most likely cause of 23rd January debris avalanche is upward propagation of cliff failure through time. The final cause of failure would be the culmination of cliff material weakening through time (due to weathering processes such as saltwater wetting and drying and failure surface development in response to ongoing stress relief). As the strength of the material decreases and the failure propagates, ‘low’ apparently benign environmental stresses can act as a trigger for final failure. Several smaller rockfalls have occurred after the 23rd January 2019 debris avalanche, including a 10,000 m³ debris avalanche located to the east of the 23rd January debris avalanche. This subsequent cliff failure narrowly missed two tourists who were walking along the beach below (the beach and accessway were closed at the time). Anecdotal evidence, field observations, and aerial LiDAR analysis all indicate that landslides of a similar size, or smaller, occur regularly along this 7 km long section of coastal cliffs between Clifton and Black Reef. However, the baseline risk users of the beach accessway are exposed to is unknown. As such, the logical next step would be to undertake a quantitative risk assessment to quantify this baseline risk. (auth)