The reconstruction of Holocene relative sea level (RSL) is important to understand and project future sea-level rise. However, there is a lack of sea-level records developed in far-field regions such as Singapore, which presently experience minimal glacial isostatic adjustment (GIA) and are tectonically stable. These regions can provide robust constraints on ice-volume equivalent sea level. Here, we aim to produce standardised sea-level index points (SLIPs) that consider vertical and temporal uncertainties from sea-level indicators to constrain Holocene RSL in Singapore. We obtained sediment cores from two locations in Singapore. (1) A series of sediment cores with basal mangrove peats were collected from an upland-mangrove transition area in Pulau Ubin, northeast of mainland Singapore. We sampled plant macrofossils for radiocarbon dating. The mangrove peat spans 5528 ± 60 – 140 ± 131 cal yr BP with elevations from 0.45m MTL to -0.53m MTL. (2) Additional sediment cores were obtained from Jurong Lake in southwestern Singapore. Organic units within the Jurong Lake cores, the dating of which is ongoing, range from -1.9m MTL to -6.7m MTL in elevation. Research is underway to constrain the timing and magnitude of Holocene RSL change. Using laboratory techniques such as pollen analysis, we aim to establish the indicative meaning of mangrove peats to constrain their vertical position and uncertainty with respect to the tidal frame. Their temporal position and uncertainty are obtained through radiocarbon dating. A decompaction model will also be used to correct sample elevations for sediment compaction processes. Together with existing sea-level records, these SLIPs will be used to constrain GIA models and assess the magnitudes, rates and processes driving Holocene RSL changes in Singapore.

Relative sea-level (RSL) change in the Philippines is driven by a combination of eustatic sea level, glacial isostatic adjustment (GIA) and tectonics. The contribution of each component is however poorly studied. Coral microatolls robustly track RSL changes and can be used to infer and understand the processes that drive them. To constrain RSL and land-level change in northwestern Luzon, Philippines, we extracted slabs from fossil coral microatolls from several sites along the coastline. Here we report preliminary results from the Cabugao site in Ilocos Sur province. In Cabugao, all slabbed fossil coral microatolls have distinctly cup-shaped morphology, defined by lower centers and higher outer rings, which results from rising RSL during the coral's lifetime. Fossil corals older than ~7180 cal yr BP recorded faster rates of RSL rise (~2.9 ± 0.2 mm/yr) compared to younger corals that grew between ~1100 and 860 cal yr BP (~1.9 ± 0.3 mm/yr). Our preliminary hypothesis is that the differences in rates are dominated by eustatic sea-level rise (due to melting of continental ice sheets in higher latitudes) and GIA prior to 7180 cal yr BP. Corals estimated to be from ~7780 to ~7180 yr cal BP were all found to be 0–1 m higher than their modern equivalents, indicating either that RSL in the absence of tectonics had already reached present levels earlier than predicted by published GIA models, or that this site has experienced meters of tectonic uplift in the past ~7180 yr. Ongoing analysis of the coral microatolls records in Cabugao, a better data on elevated inferred Pleistocene surfaces at the site, and comparisons to RSL reconstructions at sites elsewhere along the coast will help discriminate between these hypotheses.

Remains of organisms that grow within tight elevation ranges (eg. coral microatolls) are proxies for relative sea level (RSL). Sea-level proxy data are commonly collected at sites hundreds of meters wide or more. A fundamental assumption is that sea level itself is level across short distances at a site: if different generations of an RSL proxy, hundreds of meters apart, are found at different elevations, the elevation differences are commonly interpreted as changes in RSL over time. But what if, instead, the differences over short distances were caused by spatial variability in sea level, rather than changes in RSL over time? We explore this possibility here. In Cabugao, Ilocos Sur, we deployed two tide gauges concurrently at two ends of our site, 2.1 km apart. Although both gauges recorded the full tidal range, sea level was consistently higher at the northern gauge by ~7.5 cm, and the difference increased by ~5 cm after the onset of a wind event. A similar pattern emerged from the fossil microatolls that we sampled at Cabugao. In two coeval pairs of microatolls, the northern corals yielded RSL reconstructions that were 2–13 cm higher than the southern corals. This raises concern that spatial differences in sea level recorded even over short distances may map into inferred RSL change. Both sets of evidence imply that sea level is persistently higher in the north than in the south. We suggest this could be due to greater wave setup in the north. Work at this site and further investigation of this issue are ongoing.

Coral microatolls allow for the reconstruction of relative sea level (RSL) and the inference of tectonic deformation along tropical coastlines over the Holocene. Microatolls track RSL with unparalleled vertical precision, and their annual banding allows us to count years precisely over an individual coral’s lifetime; however, absolute age estimates of fossil corals still suffer from limitations in radiocarbon (¹⁴C) and uranium-thorium (²³⁰Th) dating techniques. We extracted multiple samples for ¹⁴C and ²³⁰Th dating from coral microatoll slabs at sites in Ilocos Region, northwestern Luzon, Philippines. The number of annual bands separating any two samples in a coral is precisely countable, therefore, we can readily compare multiple dates from an individual slab. After adjusting for the numbers of intervening bands, all dates from a coral are treated as independent (redundant) estimates of the outer preserved band on each coral slab. After excluding samples flagged by x-ray diffraction for diagenesis, our initial findings highlight that paired ²³⁰Th dates do not overlap within 4σ in 4 of 8 cases, whereas calibrated ¹⁴C dates overlap in their 2σ probability distribution functions in 8 of 9 cases, with the simple assumption that the radiocarbon marine reservoir correction (∆R) = 0 yr. We use the programme OxCal 4.4 (Bronk Ramsey, 1995) and the updated marine radiocarbon calibration curve, Marine20 (Heaton et al, 2020), to apply Bayesian statistics to combine our ¹⁴C and ²³⁰Th dates and calculate ∆R. After setting a uniform prior of -500 to +500 yr for ∆R, we find that the ∆R value overlaps in all but one of our sites, and we calculate a joint ∆R value of -138 ± 10 yrs for all our sites in Ilocos Region. With additional ¹⁴C and ²³⁰Th samples pending, we aim to continue our analysis and further constrain ∆R and its temporal and spatial variability.

We present a workflow exploiting the complementarities that exist between 1D magnetotelluric (MT) and magnetic data inversions from both the petrophysical and structural point of views. We introduce a flexible, sequential workflow that leverages inversions that can be performed separately. It is constituted of three main steps. We first perform 1D probabilistic MT inversion to obtain ensembles of models honouring the data and their uncertainties. Secondly, we use these ensembles to calculate the probabilities of observation of the modelled rock units. We derive domains characterized by positive probabilities to observe the different rock units. Thirdly, these domains are used to define constraints for magnetic data inversion where magnetic susceptibilities are restricted accordingly with the rock units that can be observed within each domain. This is achieved using the alternating direction method of multipliers to enforce bound constraints using multiple disjoint, spatially-varying intervals as implemented in the Tomofast-x inversion platform. We test this workflow using a geologically realistic model obtained using data from a region in the Mansfield area (Victoria, Australia) and apply it to real-world data, with an emphasis on the modelling of the sediment-basement interface. Results indicate that our methodology is useful to refine our understanding of the basement interface and that MT-magnetic data integration may reduce interpretation uncertainty.

The process of 3D modelling in the geosciences involves the collection, aggregation and integration of usually disparate data sources into a structured knowledge-base. Supporting this crucial endeavour is the Geoscience Knowledge Manager (GKM), an ontology-based resource that incorporates knowledge of geological units, structures, properties, events and their relationships within the 'Loop' platform. The geosciences are also familiar with ontologies, with many proponents seeing benefits from a successful implementation of: (1) a shared vocabulary; (2) increased interoperability and (3) automated reasoning. GeoSciML, perhaps the most well-known example of a geoscience data structure, provides a standardized model for geological concepts and an associated format for exchanging data between geoscience databases. In a similar manner, the Loop GKM and its GeoScience ontology (GSO) aim to enhance 3D modelling practices by providing support via the three advantages listed above, but specific to the procedures and expectations of the 3D modeller. To begin testing the GKM for geophysical modelling purposes, we to perform the simple task of assigning plausible petrophysical values and their errors to a 3D model in order to obtain a geophysical forward model. This is achieved by structuring the petrophysical and stratigraphic knowledge using the GSO, and storing it in GKM semantic database for retrieval by 3D modelling tools. This task is by no means difficult, but the use of GKM is superior to the traditional procedure which typically involves obtaining the necessary data from disconnected databases, published papers, archived data stores and when not available, from generic ‘global’ examples. Thus, repeatability is not always ensured. Using an ontology-based GKM can highlight subtle but important relationships between geological entities to improve accuracy and repeatability. An ongoing and consistent record of data use and knowledge gain provides a valuable resource for decision accountability, knowledge-transfer and uncertainty reduction.

Imaging the subsurface with appropriately high resolution has always been challenging. Building 3D geological models is a time-consuming task due to geological complexity and the multiple scales to be considered. A particularly challenging task in exploration geophysics is to generate models of the subsurface consistent with all available datasets. Combining geophysical datasets with different characteristics has become necessary to address such complexity. Geophysical datasets can be integrated using sequential, cooperative, or joint inversions. Method selection depends on the geoscientists’ goal, availability of the data, the physical properties, and geology of rock units in the area. Two units with a single differing physical property do not necessarily indicate differing geological units. However, the difference between two or more physical properties increases the possibility of distinct geological units. This encourages the implementation of cooperative and joint inversion algorithms. Joint modeling of seismic and gravity data reduces the ambiguity of velocity and density when compared to single domain inversions especially when petrophysical measurements are available. However, borehole petrophysical information is insufficient for correlating purposes at larger scales and for deep structures. In addition, inverting seismic jointly with other geophysical datasets is challenging due to the relative sparsity of seismic data and coverage that differs between datasets. Furthermore, modeling is often restricted to refracted seismic raypaths due to the limited computational resources for integrating datasets of differing resolution, especially in hard rock exploration. In this study, we explore and discuss different ways of integration of information from seismic images and gravity data at regional scales. We utilize the results from cooperative and constrained inversion of seismic and gravity data using property and geometric inversion to raise the necessity of new procedures for joint inversion.

The Hamersley Basin of Western Australia, approximately 500km long by 200km wide, is one of the largest Archean-Proterozoic basins in the world. It is also recognised globally for its rich endowment of iron ore resources. Although significant mining operations and exploration programs have been conducted across the basin resulting in substantial knowledge of the stratigraphy and structure a single regional 3D geological model has not been generated. The ability to visualise the 3D geology of the entire basin is important for the understanding of iron ore occurrences and their relationship to ore genesis of the bedded iron ore mineralisation. Recent advances in implicit automated geological software modelling presents an ideal opportunity to develop a 3D model of the entire Hamersley Basin. The innovative Loop software was used to model the stratigraphy and structure of the Hamersley Basin utilising open source geological data sourced from the Geological Survey of Western Australia. However, due to the large scale of the model certain assumptions were made and filtering of the data was performed. For example, the open source outcrop bedding data generally provided good coverage across the basin, however, some areas had very little or no structural measurements, thus, the map2loop software generated interpolations based on the nearest data points. Furthermore, mapped faults measuring less than 1km were excluded from the dataset. Therefore, building a model at this large scale will have an implied expectation that local scale structural and geological variations will not be captured. Furthermore, a level of geological uncertainty will exist in areas of sparse data. This study demonstrates the assumptions made and data filtering applied when generating a large scale 3D geological model of the Hamersley Basin. Furthermore, this study highlights the importance of understanding the implications of a model built at certain scales and the inherent uncertainty.

　Abrupt climate changes are occurred during the late Pleistocene to Holocene period. Basin-fill deposits in inland basins are suitable records of paleoclimate, because the sediment production and sedimentary environment in such basins are regulated by climate change and tectonic activity rather than sea level change (Shanley and McCabe, 1994). We clarify the changes of sedimentary environment and lake-level based on the sedimentary facies analysis, paleosol descriptions, and major elemental analysis for the boring core collected from the Suwa Basin which is an inland basin in central Japan, and discuss potential causes related to climate change during the latest Pleistocene and Holocene.  　In the last glacial period ca. 15,000–27,000 cal yrs BP, meandering fluvial environment was occurred in this site . The lacustrine environment expanded in ca. 12,000–15,000 cal yrs BP coinciding with the abrupt warming (Bølling-Allerød). The lake-level fluctuation with short cycle can be recognized in the sediments between ca. 7,000 and 12,000 cal yrs BP, based on the alternation between paleosol layers and diatom-rich mud in the lacustrine unit and variation of TN. The major lowstand period at ca. 8,200–8,800 cal yrs BP is contemporaneous with the cooling event at 8,200 yrs BP and the weakening of the East Asian summer monsoon. The higher lake-level developing during the period ca. 4,000–7,000 cal yrs BP indicates an increase in precipitation and coincides with the middle Holocene climatic optimum. The fluctuations of lake-level in the Suwa Basin can be linked to climate changes during the late Pleistocene to Holocene.  References: Shanley, K.W., McCabe, P.J., 1994. Am. Assoc. Petrol. Geol. Bull. 78, 544–568.

¹⁷O-excess (¹⁷O-excess = ln(δ¹⁷O+1)-0.528×ln(δ¹⁸O+1)) has been recently considered a new hydrological tracer. Experiments suggest that ¹⁷O-excess in precipitation record the humidity conditions in its moisture source regions. However, observations show that the controls of ¹⁷O-excess are much more complicated than experiments suggested. Before applying ¹⁷O-excess to hydroclimate studies, further investigations are needed to understand what control ¹⁷O-excess, particularly in the tropics where studies of ¹⁷O-excess are sparse. In this study, we collected nearly six-years of monthly precipitation samples in Singapore, and examined their δ²H, δ¹⁸O, and δ¹⁷O to understand the drivers of their variability with a focus on ¹⁷O-excess. We found that in low latitudes, ¹⁷O-excess mostly varies in the range of 10 and 40 per meg, much narrower relative to high latitudes. An anti-correlation exists between d-excess and ¹⁷O-excess, and this correlation breaks down during ENSO years. Both¹⁷O-excess and d-excess are found not to record RH of the moisture source region. Likely, processes during convection significantly affect these two parameters. Further, we found that regional convection associated with ENSO, monsoon and MJO has different impacts on δ¹⁸O, d-excess, and ¹⁷O-excess. Therefore, they exhibit different periodicities over the time. Weak periodic variations were observed in δ¹⁸O, and especially d-excess. On the other hand, ¹⁷O-excess shows strong periodicities in response to MJO, monsoons, and ENSO. This finding implies that ¹⁷O-excess has the potential to be used as a proxy for tropical climate variabilities and thus has significant implications for paleoclimate studies.

Information about the long-term landslide surface displacement and spatiotemporal evolution can improve our understanding of the landslide development process and can help prevent landslide disasters. Based on field investigation, Interferometric Synthetic Aperture Radar (InSAR) as well as Unmanned Aerial Vehicle (UAV) photogrammetry, high-resolution remote sensing imagery and digital elevation model, we addressed the surface displacement, travel distance, topographic changes, and causative factors of the landslides. The results show that InSAR technique has an advantage in terms of the retrieval of pre- and postfailure creep deformation. Mostly, the pre- and post failure spatiotemporal deformation processes and evolutionary patterns of the landslide are different. The combination of ascending and descending orbit datasets can not only be used to monitor the landslide surface displacement, but also to verify mutually the deformation results. We detected the surface travel distance of the landslide and found spatial differences exist in the surface travel distance of the landslide. The frequency distribution of the basic topographic factors before and after the landslide are different, which indicates that the landslide event significantly changed the local topography and geomorphology.

A progressive entrainment model has been incorporated into an energy-based runout model to analyze the kinematic characteristics of debris flow. The progressive entrainment model considers both rolling and sliding motions of granular material during the erosion process. The incorporation entrainment in the runout model is presented in this paper. Measurements from the 1990 Tsingshan debris flow in Hong Kong, China, which had significant entrainment, are used to validate the model. Velocities at different locations along the channel based on the super-elevation method and entrainment depth are used to evaluate the modelling results. Comparison between the calculated and observed maximum velocity and entrainment depth shows that the progressive entrainment model incorporated into the runout model can capture the erosion characteristics of the granular material. It also indicates that the assumption of rolling motions in the progressive entrainment model is reasonable, which is the dominant erosional mechanism in granular material.

The geological hazard chain (debris flow-dammed lake-outburst flood) may result in varying degrees of damage to downstream infrastructure and endanger the safety of human beings, increasing the difficulty of disaster prevention and mitigation. Although considerable progress has been made recently in the theory and modelling of such geohazard chains, the mechanisms that govern the hazard process in a natural environment are still unclear. Meanwhile, on-site monitoring instruments and equipment are usually vulnerable to these hazards and experienced destruction during large hazard events. Recently, indirect methods, such as ground vibration measured by seismometers, provide a potential breakthrough in geological hazard chain measurement. Such indirect methods can be used to monitor the occurrence and invert the mechanics of the hazard chain. In this study, the evolution process of the Danba landslide-dam failure hazard chain is comprehensively analyzed by field surveys, remote sensing, and seismic signals. Numerical simulations of the dam breach-flood routing process are then conducted using dam breach overtopping model and two-dimensional hydrodynamic models. The evolution process of each stage in the hazard chain was then interpreted. The debris flow, landslide, barrier lake outbursting, and flood routing were analyzed to determine their processes. Finally, the reactivation of ancient landslides downstream caused by flood erosion was also identified with stage explanations. This method provides new ideas for interpreting the hazard chain processes of landslide dam failure hazard chains and theoretical guidance for hazard early warning and mitigation.

Debris flow hazard often brings huge economic losses and fatalities downstream. Despite the traditional on-site monitoring system developed, the equipment is vulnerable to be damaged during the hazard process, resulting in the limited in-situ data collected to analyze the dynamic process. Recently, with the development of seismology, using seismic signals recorded from geophones can reconstruct the process of debris flow for hazard assessment. The scientific challenge lies in how to link the diagnosed seismic signal to the basal fluctuating force to reveal the dynamic process of debris flow. In this study, a physical flume experiment is carried out using viscous slurry and glass bead as the debris materials. PIV (Particle Image Velocimetry) method is used to obtain the velocity field during the transportation process. Besides, acceleration and stress sensors are used to obtain the seismic signal and the basal impact force. Results indicated that the variation trend and peak time of basal impact force are similar to seismic signals, which can be used to explore the dynamic process of debris flow. Based on this, the monitoring and early warning work of debris flow hazards can be guided.

Landslides are dangerous natural disasters that can cause significant losses downstream and often hard to predict. Although great research efforts have been made in these hazards, such as physical modeling, numerical simulation, and on-site monitoring, the mechanisms that govern the hazard process in a natural environment are still unclear. Recently, with the development of environmental seismology methods, it is possible to use the seismic signals that are recorded by the seismic stations and Geophones to provide a new geological hazard diagnostic. Previous literature indicated that the force-time function including the key kinematic characteristics of landslides can be calculated from the seismic signal generated by the impact of source material on the ground. In this research, we selected a landslide that occurred in the mountains of southwest China as a case study. The spectrogram is first calculated, which is the vertical component of the seismic records. Then, the seismic signal inversion method is used to calculate the force-function, and the discrete element method (DEM) is used to simulate the dynamic process of the landslide. The input parameters of DEM are first calibrated through a traditional geotechnical laboratory test, a more reliable simulation scenario of the landslide dynamics process is then conducted by comparing the force-time function with the impact force on the bottom surface of the DEM model. Followed by data interpretation, each stage of landslide evolution is analyzed in detail. This study demonstrates a new way to reconstruct the dynamic process of landslides using numerical simulation calibrated from the diagnosed seismic signal, which has important significance for disaster prevention and mitigation in the mountainous area.

The failure of landslide dams is a sudden geological disaster, and thus their formation and failure greatly threaten the security of people’s lives and property. In order to better reproduce the entire overtopping breaching process of landslide dams, we conducted 12 sets of model experiments to explore the influence of the flume gradient, dam height, and downstream slope angle relative to the flume bed on the breaching process of landslide dams. Based on the results of these experiments, we analyzed the characteristics of the longitudinal and transverse evolution, and outburst discharge of landslide dams in detail. It was found that the dam overtopping breaching can be divided into four stages: initiation, head cutting, acceleration, and riverbed rebalancing. The variation in the erosion rate (E_AX and E_BX) and the downstream slope angle relative to the horizontal line (Φ) exhibit clear differences in the different stages. The influences of the flume gradient, dam height, and downstream slope angle on the breaching process of landslide dams are different. Since the dam breach evolution mode and dam overtopping breaching process were mainly affected by the dam height and downstream slope angle relative to the horizontal line, the evolution of the longitudinal section in the process of dam body failure can be divided into four modes based on our experimental results. This preliminary research provides a basis for subsequent studies of dam breaching and a scientific reference for the prevention and mitigation of landslide dams.

Ground shaking from moderate to high magnitude continental earthquakes in mountainous regions can initiate a chain of cascading hazards that are known to far outlast the earthquake itself. One consequence is that reconstruction efforts are faced with a legacy of latent landslide risk for months or years after the shaking has stopped. There remains relatively limited analysis of how this legacy of earthquakes in mountain regions shape the geography of geohazard risk to both people and infrastructure. This is as a result of both limited post-earthquake landslide time-series inventory data that captures how the hazard changes, and is coupled with the limited use of forward-looking modelling of how landslides may evolve and the risk that they could pose to individual households. We address these uncertainties by assessing relative risk to residential buildings from landsliding over the period 2014 to 2019, with a specific focus on the 4 years following the 2015 Mw 7.8 Gorkha earthquake. Using a time-series landslide inventory data mapped across the full footprint of the earthquake, and a distributed physics-based run-out model, we map how the evolution of how potential landslide runout intersects with and poses a continued risk to residential buildings downslope. We observe how the footprint of runout hazard systematically shifts and intersects with concentrations of residential housing, generating a systematic shift in the geography of landslide risk through time. We examine how the character and distribution of this risk has changed since 2015, to identify the timescales over which a return to pre-earthquake landslide risk may occur and to assess who is at risk, when and where. Finally, we use this to assess what actions or decisions would be necessary to reduce exposure to landslide risk in a post-earthquake mountain landscape.

Discharge thresholds for runoff-generated debris flow initiationCommon rainfall intensity-duration (ID) thresholds for debris flow are empirical and developed with historical data, and therefore most applicable to those settings where debris flows have been recorded in the past. We propose a method to derive dimensionless discharge critical values for runoff-generated debris flows by combining process-based numerical modeling and machine learning methods. We train logistic regression functions using monitoring data and hydrologic modeling for debris flows with support vector machines. Our training dataset includes debris flows in the San Gabriel Mountains (USA), Chalk Cliffs (USA), and runoff events in the Venetian Dolomites (Acquabona and Cancia, Italy. Our proposed approach can be used to estimate rainfall ID thresholds in areas with no historical data on runoff-generated debris flow occurrence. This results in a dimensionless discharge threshold consistent with previously derived discharge thresholds for post-fire debris flows in southern California.