Hydrogen leakage through abandoned wells represents a critical risk for underground hydrogen storage (UHS) in depleted gas reservoirs, potentially leading to safety hazards and economic losses. Although reservoir-scale numerical simulations can capture leakage processes in detail, their computational cost limits their use in rapid scenario screening and early-stage design assessment. In this work, a probabilistic surrogate framework is developed to evaluate leakage likelihood under varying geological and operational conditions. The framework is intended as an early-stage screening tool to identify configurations that may pose a leakage risk and therefore warrant further detailed investigation. The surrogate is constructed using multiphase reservoir simulation results and incorporates six continuous variables together with ten families of H2-water relative permeability curves spanning a range of low to high mobility that collectively represent structural uncertainty in gas-water flow. Validation against an independent simulation dataset shows that the surrogate achieves an overall prediction accuracy of 0.967 with an area under the ROC curve (AUC) of 0.995. Results indicate that, within the investigated ranges and under homogeneous reservoir conditions, well spacing and injection rate are the principal controls on leakage probability. Formation permeability does not show a stable directional influence under relatively high-permeability conditions (100-1000 mD), while reservoir depth and porosity are negatively associated with leakage probability. In addition, variability in relative permeability significantly affects predicted risk, with high-mobility curve families increasing leakage likelihood relative to intermediate conditions. This framework provides an efficient basis for leakage-aware screening and supports risk-informed evaluation during the early design stage of UHS deployment.

Underground hydrogen storage (UHS) is a promising solution for large-scale, long-term energy storage by converting surplus renewable electricity into hydrogen for later use. Hydrogen recovery efficiency is an essential indicator for evaluating the performance of UHS. However, the impacts of permeability anisotropy, including horizontal permeability anisotropy (kh,max/kh,min) and horizontal-to-vertical permeability anisotropy (kh/kv), on hydrogen recovery efficiency have not been thoroughly investigated. This study addresses this gap with numerical simulations that couple a simplified box model and a field-realistic model based on the Ahuroa gas storage site. Our results show that kh,max/kh,min affects hydrogen recovery efficiency, and this effect becomes more pronounced at higher permeability. The difference in hydrogen recovery efficiency between kh,max/kh,min=1 and 10 expands from 1.3 percentage points at 100 mD to 4.3 percentage points at 200 mD. Increasing kh/kv can reduce the effect of kh,max/kh,min. In reservoirs with horizontal anisotropy, orienting a horizontal well along the minimum rather than the maximum permeability direction can reduce hydrogen recovery efficiency by up to 8 percentage points (from 78 % to 70 %), with all other conditions held constant. These findings from the box model were subsequently validated through simulations of a realistic geological site (Ahuroa natural gas storage site). This study indicates that permeability anisotropy is an important consideration in UHS site assessment. Ignoring it can result in errors in hydrogen recovery efficiency evaluation, leading to inaccurate predictions of recoverable hydrogen volumes.

Background Wildfire risk is rising under climate change, yet most operational forecasts rely on daily indices that miss rapid weather shifts preceding ignition. Accurate, timely and economically justified forecasts are critical for early warning and resource allocation.Aims This study evaluates a sub-hourly machine-learning (ML) forecasting system for fire potential using weather-station data from three Australian regions (Sunshine Coast, Brisbane, Hobart), and quantifies its economic value using a cost-loss framework.Methods ML classifiers were trained on sub-hourly Automatic Weather Station (AWS) data and benchmarked against the Fire Behaviour Index (FBI). Forecast discrimination was assessed using True Positive Rates (TPRs) and False Positive Rates (FPRs), as well as a Potential Economic Value (PEV) analysis, complemented by learning-curve tests.Key results The ML model improved forecast skill over the FBI by 10-30%, with the ML system doubling potential savings relative to the FBI. Learning-curve diagnostics indicated that stable performance is achieved after similar to 30 fire events.Conclusions The ML forecasting framework demonstrates predictive skill and measurable economic value, highlighting its potential for scalable, cost-effective early-warning systems.Implications By linking forecast skill directly to financial and operational outcomes, this approach provides a quantitative basis for prioritising investment in timely, data-driven fire-warning tools across regions with limited data and constrained budgets.

Hydrogen, a renewable energy source, is expected to see steadily increasing production to meet global net-zero targets. Depleted hydrocarbon reservoirs, with their large storage capacity, offer a promising solution for largescale underground hydrogen storage (UHS). Stored hydrogen (H2) and cushion gases (e.g., CH4) may coexist with residual hydrogen sulfide (H2S) or H2S generated locally through geochemical reactions or microbial activity. As gas composition significantly influences the interfacial tension (IFT) between gas and pore water, the distribution and dynamics of H2and CH4 in depleted reservoirs are affected by H2S under reservoir conditions. This study presents newly determined IFTs of H2-H2S-water and CH4-H2S-water mixtures at 298 K and 343 K with pressures ranging from 10 to 30 MPa using molecular dynamics (MD) simulations. Results reveal that IFT decreases with increasing pressure, temperature, and H2S concentration, with distinct mechanisms driving these variations. Notably, even small amounts of H2S (e.g., 10 mol%) significantly reduce IFT, underscoring its critical role in gas-water interactions. With simulated IFT data, two regression models are developed to predict the IFTs of H2-H2S-water and CH4-H2S-water mixtures under geological conditions. This study provides comprehensive IFT expressions for CH4-H2S-water and H2-H2S-water mixtures and offers insights into the mechanisms underlying IFT variations for UHS applications.

Flood prediction at sub-daily resolution requires models that are accurate and physically plausible beyond observed conditions. We augment a per-basin Long Short-Term Memory network with a linear reservoir constraint (Q = KS), embedding storage-discharge dynamics as a soft physics loss, and evaluate it in hourly and daily streamflow-prediction experiments across New Zealand and Great Britain. At hourly resolution, the physics-informed LSTM improves performance in 99% and 91% of catchments, respectively, increasing median NSE from 0.30 to 0.44 in New Zealand and from 0.61 to 0.70 in Great Britain while learning physically plausible recession timescales. The largest gains occur where test-period flood recessions differ most from training conditions, indicating that the constraint is particularly useful for hydrological behaviour that is weakly represented during training. However, these benefits largely disappear at daily resolution, with improvements observed in only 14% and 21% of catchments in New Zealand and Great Britain, respectively. At hourly resolution, recession limbs span multiple timesteps and directly inform the storage–discharge constraint, which learns physically plausible effective recession timescales (median ~11.5 hours in New Zealand) that vary among catchments independently of catchment size; daily aggregation compresses these dynamics and substantially weakens that information. These results show that the value of physics-informed learning depends on whether the constrained physical process is resolved at the modelling timestep, and suggest that temporal resolution should be considered explicitly when designing and evaluating physics-informed hydrological models.

With the rapid expansion of renewable energy deployment, underground hydrogen storage (UHS) has emerged as a promising large-scale storage option to smooth seasonal fluctuations in electricity supply. However, current geological assessments for UHS primarily focus on reservoir properties such as porosity and permeability, while overlooking the influence of interlayers. To address this research gap, this study systematically investigates how interlayer characteristics, including permeability, thickness, and geometry, affect hydrogen recovery efficiency and the extent of unrecoverable hydrogen. Numerical simulations were conducted using OpenGoSim (OGS) software, beginning with a box model and extending to a realistic geological model based on the Ahuroa gas storage site.Simulation results reveal that, under fully perforated conditions, lower interlayer permeability impedes upward hydrogen migration, thereby reducing the impact of gravity override and enhancing hydrogen recovery efficiency. With decreasing interlayer permeability, hydrogen transport into the interlayer shifts from advection-dominated transport to diffusion-dominated transport, resulting in greater hydrogen volume than predicted by a model that neglects molecular diffusion. The study further demonstrates that optimizing well configuration should consider interlayer properties to maximize recovery efficiency. Consistency between the box model and the realistic geological model supports the generality of the findings.By integrating interlayer characteristics into site evaluation, this work enhances the accuracy of hydrogen recovery efficiency evaluation and advances UHS assessment.

For relatively isolated energy systems, such as for island nations like New Zealand, energy balancing is an important consideration for ensuring system reliability. Underground hydrogen storage (UHS) is one possible technology for addressing seasonal fluctuations of solar, wind and hydropower generation on month to year timescales. Although prior power system modelling has considered UHS, it has typically represented subsurface storage with simplified tank models that neglect expected geological complexity and the operational constraints of managing a subsurface reservoir.Here, we present an energy-balance model of a national power system that incorporates (1) seasonal generation fluctuations derived from New Zealand’s historical records, (2) a UHS facility based on geological characteristics of the Ahuroa gas field (a natural gas storage site in Taranaki, New Zealand), including structure, storage volume, and well configuration, and (3) operational constraints, including reservoir pressure limits, and co-production and treatment of formation water. The model is operated under future multi-year scenarios that incorporate expected growth in renewable generation as well as demand.Our study finds that under typical meteorological conditions, a single UHS site with capacity of 5.6 PJ can buffer median annual electricity fluctuations of 2.6 PJ. This result is robust under a range of future scenarios including variation in electricity mixes and climatic conditions. However, as wind and solar increase to replace fossil fuels, the seasonal balancing requirement exceeds UHS capacity. Due to round-trip conversion losses – power to hydrogen to power – renewable overbuild that provides an additional 3 PJ annually is required to maintain sufficient hydrogen inventory for stable multi-year operation.During meteorological dry years, when hydropower generation is well below average, the UHS is called upon to deliver gas at higher than ordinary rates. This causes low-pressure transients in the reservoir that lead to the gas-water interface moving upward, increased water co-production that exceeds treatment capacity, and hence inability of the UHS to meet the energy shortfall.

Abstract The adoption of new forecasting methods for volcanic eruptions typically emphasizes predictive accuracy, often overlooking their potential to reduce overall life and economic losses. We introduce a Potential Economic Value (PEV) framework that evaluates forecast utility by balancing precautionary actions against avoidable catastrophic losses (such as mass casualties). Using machine-learning forecasts from continuous seismic data at five volcanoes, we show that non-forecasted eruptions (missed) have disproportionate consequences, compared to false alarms, which generate recurring and manageable disruption. Retrospective analyses of the 2019 Whakaari (New Zealand) and 2014 Ontake (Japan) eruptions, along with three additional volcanoes, indicate that losses could have been reduced by 30–90%, despite generating numerous false alarms. Across all case studies, effective forecasting prioritizes reducing missed eruptions over maximizing accuracy. This supports the use of more precautionary warning thresholds, provided that the resulting increase in alert frequency is managed in ways that maintain public compliance and trust.

Geyser eruptions provide a test bed for using geophysical data to forecast eruptions and to understand heat and mass transport in hydrothermal systems. We used time series analyses of seismic data at Steamboat Geyser, Yellowstone National Park, to identify short‐term precursors that are recurrent, detectable in real time, distinctly identifiable, as well as being rare during non‐eruptive periods. We analyzed seismic data from March to December 2018 to identify patterns that occurred before 31 eruptions. Four seismic amplitude measures and 700 time‐series features were computed from the seismic data. A template matching analysis identified an optimal 18‐hr window for detecting precursors. We applied a random forest to classify pre‐eruptive and non‐eruptive data for out‐of‐sample eruptions (eruptions that were not included in the model's training data), showing ability to distinguish between the two. This model performed better than a simpler amplitude‐based approach. Seismic features with the most predictive power include autocorrelations, longest strike above the mean, and change quantiles, particularly within the 4.5–16 Hz frequency range. We applied isotonic regression, a method that converts raw model outputs into calibrated probabilities, to improve the interpretability of eruption forecasting outputs. The likelihood of an eruption reaches 12.6% within 18 hr prior to the event, representing a marginal increase over the static 8% probability derived solely from eruption intervals. Unlike the interval‐based approach, our model does not rely on the time since the last eruption, instead using real‐time seismic features to detect precursory signals. Our study advances Machine Learning methodologies in eruption forecasting by integrating calibrated probability estimation through isotonic regression, which has advantages over traditional approaches for geysers with highly irregular eruption intervals.

Underground hydrogen storage (UHS) in depleted reservoirs is a promising solution for large-scale energy storage and decarbonization. However, abandoned wells within these reservoirs pose significant risks to storage integrity and environmental safety, necessitating thorough assessment of potential leakage pathways and mitigation strategies. This study presents a numerical reservoir simulation of gas leakage from abandoned wells into the overlying aquifer during UHS in a depleted gas reservoir, serving as a reference case for future underground H2 storage projects. Predicting hydrogen leakage over time under varying geological and operational conditions involves considerable uncertainty. To improve model reliability, sensitivity analysis was performed to evaluate the impact of key geophysical and operational parameters on gas leakage including formation permeability, formation porosity, caprock thickness, leakage accessibility (abandoned well permeability), well location (distance to the abandoned well), and injection rate. In the model, the abandoned well is represented as a porous medium with higher permeability than the surrounding formation, consistent with previous studies. The storage operation involves a 3-years injection of cushion gas (CH₄) to establish a pressure plume, followed by 13 years of hydrogen injection. Flow dynamics of gases in the reservoir, abandoned well, and overlying aquifer were successfully simulated using the PFLOTRAN reservoir simulator at the first. Then sensitivity analysis reveals that the most influential parameters affecting leakage into the overlying aquifer are the well pattern, reservoir porosity, and injection rate. Understanding the influence of these parameters is essential for optimizing storage design and ensuring long-term safety, contributing to the development of secure and sustainable underground hydrogen storage systems.

Seismic data recorded before volcanic eruptions provides important clues for forecasting. However, limited monitoring histories and infrequent eruptions restrict the data available for training forecasting models. We propose a transfer machine learning approach that identifies eruption precursors—signals that consistently change before eruptions—across multiple volcanoes. Using seismic data from 41 eruptions at 24 volcanoes over 73 years, our approach forecasts eruptions at unobserved (out-of-sample) volcanoes. Tested without data from the target volcano, the model demonstrated accuracy comparable to direct training on the target and exceeded benchmarks based on seismic amplitude. These results indicate that eruption precursors exhibit ergodicity, sharing common patterns that allow observations from one group of volcanoes to approximate the behavior of others. This approach addresses data limitations at individual sites and provides a useful tool to support monitoring efforts at volcano observatories, improving the ability to forecast eruptions and mitigate volcanic risks.

Limiting global temperature rise to between 1.5 and 2 degrees C will likely require widespread deployment of carbon dioxide removal (CDR) to offset sectors with hard-to-abate emissions. As financial resources for decarbonization are finite, strategic deployment of CDR technologies is essential for maximizing atmospheric CO2 reductions. Carbon capture and sequestration (CCS), using either direct air capture (DACCS) or bioenergy (BECCS) technologies has a particular synergy with geothermal electricity generation. This is because expensive geothermal infrastructure can be leveraged to transport dissolved CO2 for storage in subsurface reservoirs. Here, we present a techno-economic comparison of renewable electricity generation coupled with either BECCS or DACCS at high-temperature, low-gas hydrothermal systems. We use a systems model that quantifies energy, carbon and financial flows through a generic hybrid power plant. At a CO2 market price of $100/tonne, the geothermal-BECCS system has a lower median cost of electricity generation ($88/MWh) than geothermal-DACCS ($181/MWh) and conventional geothermal ($89/MWh). Geothermal-BECCS also had the lowest costs of overall emissions abatement, $122/tCO(2), accounting for carbon removal and assuming displacement of fossil-fuel generation. Abatement costs are even lower, $45/tCO(2), for BECCS retrofit of existing geothermal plants, owing to discounted costs of pre-existing injection wells, steam fields, and plant equipment. For a case study based on a geothermal field in New Zealand's Taup & omacr; Volcanic Zone (TVZ), we determined that achieving CDR rates of 1 MtCO(2)/year via new geothermal-BECCS builds would require 62 standard geothermal wells and 790 kt/year of feedstock and result in 511 MWe in installed capacity. In contrast, geothermal-DACCS would need 49 wells and no external fuel source to achieve 1 MtCO(2)/year scale but result in only 190 MWe in installed capacity. Both pathways are calculated to require similar upfront investment costs at $2.2 billion and $2.3 billion for geothermal-BECCS and geothermal-DACCS respectively. Although geothermal-DACCS removes CO2 at high rates, its high parasitic load increases the overall decarbonization cost ($187/tCO(2)). In contrast, when biomass hybridization is considered, geothermal-BECCS has a lower cost of emissions abatement and produces 20 % more electricity than the benchmark geothermal plant. We conclude that this increase in electricity production makes geothermal-BECCS the more cost-effective geothermal-based CDR configuration. Finally, we argue that revenues from net-negative CO2 emissions and increased power production make geothermal-CDR a cost-competitive decarbonization technology.

Underground hydrogen storage (UHS) in depleted gas reservoirs is a possible solution for large-scale seasonal energy storage. A key challenge with UHS lies in an elevated gas-water contact zone due to prior hydrocarbon exploitation. This condition increases the risk of water invasion during hydrogen injection and withdrawal operations, adversely affecting overall storage performance. However, the mechanisms controlling this phenomenon and effective mitigation strategies remain poorly understood.This study aims to optimize UHS operational strategies by investigating the impact of operational parameters on water invasion behaviour using a model based on an existing underground gas storage (UGS) facility. In this study, numerical simulations were conducted to evaluate the influence of various operational parameters, including cushion gas injection rate and volume, along with hydrogen injection and withdrawal rates, durations, and volumes. Our results demonstrate several key findings. First, hydrogen's high compressibility reduces water invasion effects compared to natural gas when injecting and withdrawing equivalent volumes. Second, extending the duration of withdrawal stage and increasing withdrawal rate significantly increases water production rate. Third, cushion gas volume strongly influences reservoir pressure, with larger volumes reducing water production due to a stabilized pressure baseline. Fourth, the amount of gas injected and withdrawn in each cycle positively correlates with water production rate, while increasing injection and withdrawal frequency mitigates water invasion effects. Finally, cushion gas and hydrogen injection rates have less impact on water invasion.These findings provide a basis for optimizing UHS design parameters in depleted reservoirs by specifying injection and withdrawal schemes and cushion gas management, intending to mitigate water invasion risks.

Underground storage of green hydrogen in depleted gas fields could provide Aotearoa New Zealand (ANZ) with a storage option critical for meeting peak energy demands and realising green hydrogen ambitions. During early de-risking of specific sites, it is important to develop an accurate geological model to test whether the reservoir has the desired containment, volume and hydrogen deliverability. However, where seismic reflection lines and well data are limited and/or the storage system is structurally complex, the resulting geological models may be non-unique. Therefore, injection and withdrawal simulations using different structural end members is critical to constrain how a hydrogen plume may flow within (and out of) the container and interact with existing reservoir fluids. Here we present workflows for modelling a multi-year injection and withdrawal cycle of hydrogen into a depleted gas field. We use data from the Tariki Sandstone Member of the Ahuroa field in the Taranaki Basin, currently used to store natural gas in ANZ. This reservoir is located 2 km deep at the crest of an anticline above a major thrust fault, with marine mudstones forming the top seal and low-permeability fault rock the lateral seal. With only mixed quality 2D seismic reflection lines and a tight well cluster, the precise geometry of the thrust fault and its relations to smaller secondary faults is poorly constrained. To capture this uncertainty in our simulations, we have developed two 3D geological models of the Ahuroa field in Leapfrog Energy software. We use these geological models to conduct dynamic simulation of hydrogen injection and withdrawal using the massively-parallel simulator PFLOTRAN-OGS. We develop simulations that allow us to, over a 10-year cycle, test for closure or spill into adjacent fields, and predict the amount of mixing with remnant natural gas and formation water. During the simulations, we see major differences between the two geological models related to cushion injection and working H2 volumes, rates of water production and impurities due to natural gas. Additionally, one model has high risks of unrecoverable H2 gas loss when over-pressurised. Finally, we reimport the results back into Leapfrog for visualisation of the behaviour of the two hydrogen plumes over time.

Carbon dioxide (CO2) geological sequestration (CGS) involves intricate physical and chemical processes, which is a multifield coupling prediction issue. This complexity presents significant computational obstacles for conventional numerical methods, particularly over extended timescales. The machine learning method represented by the Fourier neural operator (FNO) has significantly enhanced the efficiency of solving partial differential equations (PDEs) in CGS simulations. Nevertheless, existing FNO implementations are predominantly confined to either hydraulic or geochemical fields within 2D models. Thus, an autoregressive FNO (AGFNO) method that couples hydraulic and geochemical fields across spatial and temporal has been developed, enabling precise characterization and reliable predictions of multiphase flow, CO2 dissolution, and mineralization processes in 3D for long- term CGS. By incorporating the temporal decomposition framework, AGFNO extends FNO to 4D spatiotemporal scenarios. Permeability variations serve as intermediate variables transmitted across different timesteps and models, thereby coupling hydraulic and geochemical fields and replicating various CO2 trapping mechanisms. With a prediction time of 0.31 sec/case, AGFNO is significantly faster than conventional simulations (12,000 sec/ case). When evaluated on 22% of the data set, AGFNO achieved high coefficients of determination (R-2) for predicting reservoir pressure (0.958), CO2 saturation (0.932), fluid density (0.978), and mineral volume fractions (0.935-0.964). The mean absolute percentage errors (MAPEs) for predicting residual trapping, solubility trapping, and mineralization trapping using AGFNO were 8.3%, 1.1%, and 8.7%, respectively. These numerical experiments demonstrate AGFNO's advantages in prediction speed, accuracy, stability, and physical significance. This methodology provides an alternative to conventional numerical simulations and the estimation of CO2 sequestration capacity.

Underground hydrogen storage (UHS) in depleted reservoirs presents a promising solution for managing seasonal variations in renewable energy during the global energy transition. However, the impact of reservoir heterogeneity, particularly permeability anisotropy and interlayer characteristics, on hydrogen recovery efficiency remains insufficiently understood. To bridge this knowledge gap and improve storage site selection accuracy, we developed a systematic box model to evaluate the effects of reservoir heterogeneity and validated our findings using New Zealand's Ahuroa gas storage field.Our investigation revealed that permeability anisotropy affects hydrogen recovery efficiency, with variations depending on well patterns. For well patterns with vertical wells only, both lateral (kx/ky) and horizontal-to-vertical (kh/kv) permeability anisotropy enhanced hydrogen recovery efficiency. For combined vertical and horizontal well patterns, the effect varied by anisotropy type. Lateral (kx/ky) anisotropy enhanced efficiency when horizontal wells aligned with the maximum permeability direction. In contrast, when horizontal wells aligned with the minimum permeability direction, kh/kv anisotropy exhibited an optimal ratio, beyond which efficiency began to decline. Analysis of interlayer effects revealed that reducing permeability from 1 mD to 10-3 mD led to an enhancement in hydrogen recovery efficiency, increasing from 61% to 75%. Additionally, our investigation demonstrated that the presence of interlayer pinch-outs and discontinuities along vertical hydrogen migration pathways reduced hydrogen recovery efficiency. A realistic geological model corroborated the box model findings: hydrogen recovery efficiency improved from 66.6% to 77.9% as the kx/ky ratio increased from 1 to 10, and from 66.6% to 76.2% when the kh/kv ratio increased similarly. Furthermore, inaccurate estimation of interlayer permeability could result in an 11.3% deviation in hydrogen recovery predictions.These results underscore the importance of accurately characterizing reservoir heterogeneity, including permeability anisotropy and interlayer properties, to ensure reliable hydrogen recovery predictions and improve site selection for UHS.

Background Rapidly developing pre-fire weather conditions contributing to sudden fire outbreaks can have devastating consequences. Accurate short-term forecasting is important for timely evacuations and effective fire suppression measures. Aims This study aims to introduce a novel machine learning-based approach for forecasting fire potential and to test its performance in the Sunshine Coast region of Queensland, Australia, over a period of 15 years from 2002 to 2017. Methods By analysing real-time data from local weather stations at a sub-hourly temporal resolution, we aimed to identify distinct weather patterns occurring hours to days before fires. We trained random forest machine learning models to classify pre-fire conditions. Key results The models achieved high out-of-sample accuracy, with a 47% higher accuracy than the standard fire danger index for the region. When simulating real forecasting conditions, the model anticipated 75% of the fires (11 out of 15). Conclusions This method provides objective, quantifiable information, enhancing the precision and effectiveness of fire warning systems. Implications The proposed forecasting approach supports decision-makers in implementing timely evacuations and effective fire suppression measures, ultimately reducing the impact of fires.

Hydrogen is projected to account for at least 10% of the global energy system in 20 years and is a critical component of the future zero-emissions energy system. Underground storage of green hydrogen in Aotearoa New Zealand (ANZ) will take advantage of intermittent surplus of renewable electricity at low cost, balance seasonal fluctuations in energy supply and demand, and provide a strategic reserve of energy. This poster is part of a larger research programme primarily focused on investigating the potential for underground hydrogen storage (UHS) in Taranaki, ANZ. Here, we explore the potential for UHS in porous rock formations of depleted gas reservoirs with particular focus on the role of seal integrity for storage. In this project the overarching goal is to improve understanding of whether mudstone seal strata have the potential to prevent leakage of hydrogen from Taranaki reservoirs. The primary focus is to characterise the geometries of fault and fracture systems in seal strata, their impact on its bulk permeability and to identify the pressure conditions required to promote the loss of seal integrity. In this poster we use Formation Micro Imagery (FMI) together with stratigraphic and fault/fracture mapping of core from petroleum wells to identify fracture densities, orientations and properties in both seal and reservoir rocks. Interpretations of seismic reflection lines in Taranaki and analogous outcrop observations are used to understand the geometries and permeability properties of fault zones. Preliminary results indicate that fractures are present in both reservoir and seal rocks. The densities of fractures increase with proximity to regional fold hinges and faults, and with increasing carbonate content. Questions remain about under what conditions fractures are open and capable of transmitting hydrogen. The poster outlines preliminary results, proposed research pathways and invites discussion.

Anticipating volcanic eruptions remains a challenge despite significant scientific advancements, leading to substantial human and economic losses. Traditional approaches, like volcano alert levels, provide current volcanic states but do not always include eruption forecasts. Machine learning (ML) emerges as a promising tool for eruption forecasting, offering data-driven insights. We propose an ML pipeline using volcano-seismic data, integrating precursor extraction, classification modeling, and decision-making for eruption alerts. Testing on six Copahue volcano eruptions demonstrates our model's ability to identify precursors and issue advanced warnings pseudoprospectively. Our model provides alerts 5-75 hr before eruptions and achieving a high true negative rate, indicating robust discriminatory power. Integrating short- and long-term data reveals seismic sensitivity, emphasizing the need for comprehensive volcanic monitoring. Our approach showcases ML's potential to enhance eruption forecasting and risk mitigation. In addition, we analyze long-term geodetic data (Interferometric Synthetic Aperture Radar and Global Navigation Satellite System) to assess Copahue volcano deformation trends, in which we notice an absence of noteworthy deformation in the signals associated with the six small eruptions, aligning with their small magnitude.