Imaging the Earth’s thermochemical structure is crucial for understanding its dynamics and evolution. Moreover, the increased demand for critical minerals and geothermal energy driven by the energy transition has intensified the need for reliable subsurface models. Multi-Observable Thermochemical Tomography (MTT) is a simulation-based, probabilistic inversion platform designed to harness the combined sensitivities of multiple geophysical datasets and thermodynamic modelling. It produces internally-consistent estimates of the Earth’s interior as probability distributions, offering a powerful means for uncertainty quantification. Here, we present an updated MTT formalism and assess its benefits and limitations to image the thermochemical structure of the lithosphere-asthenosphere system. Individual and combined sensitivities of different observables to parameters of interest (e.g. temperature, composition, crustal architecture) are explored using challenging synthetic models. Our findings demonstrate that a judicious combination of observables can retrieve complex thermochemical structures relevant to greenfields exploration. We then apply MTT to study two cratonic regions of geological and economic significance. In the Superior Craton, we jointly invert receiver functions, gravity anomalies, gravity gradients, geoid anomalies, Rayleigh-wave dispersion curves, absolute elevation and surface heat flow. In the North Australian Craton, we incorporate new data from the Ausarray and add teleseismic P- and S-phase travel times to the datasets. The imaged lithospheric architectures provide new insights into the tectonic evolution of these two regions and the physical meaning of geophysical signatures. Additionally, these models offer unique proxies to guide exploration efforts for clean energy and critical minerals and serve as reference models for future high-resolution studies.

Interactions among atmospheric, root-soil, and vegetation processes drive carbon dioxide fluxes (Fc) from land to atmosphere. Eddy covariance measurements are commonly used to measure Fc at sub-daily timescales and validate process-based and data-driven models. However, these validations do not reveal process interactions, thresholds, and key differences in how models replicate them. We use information theory-based measures to explore multivariate information flow pathways from forcing data to observed and modeled hourly Fc, using flux tower datasets in the Midwestern U.S. in intensively managed corn-soybean landscapes. We compare Multiple Linear Regressions (MLR), Long-Short Term Memory (LSTM), and Random Forests (RF) to evaluate how different model structures use information from combinations of sources to predict Fc. We extend a framework for model predictive performance and functional performance, which examines the full suite of dependencies from all forcing variables to the observed or modeled target. Of the three model types, RF exhibited the highest functional and predictive performance. Regionally trained models demonstrate lower predictive but higher functional performance compared to site-specific models, suggesting superior reproduction of observed relationships. This study shows that some metrics of predictive performance encapsulate functional behaviors better than others, highlighting the need for multiple metrics of both types. This study improves our understanding of carbon fluxes in an intensively managed landscape, and more generally provides insight into how model structures and forcing variables translate to interactions that are well versus poorly captured in models.

Our world has been continuously urbanized and is currently accommodating more than half of the human population in cities. Despite that cities cover only less than 3% of the Earth’s land surface area, they emerged as focal points of human activities, and confront numerous environmental challenges as a result of changes in landscapes, hydroclimate, ecosystems, and biodiversity. In particular, the built environment usually experiences exacerbated heat stress induced by global climate and landscape changes, commonly known as the urban heat island effect. Urban irrigation, as a climate adaptation and mitigation strategy, is effective in cooling the built environment, but exhibits large uncertainties in the trade-off between water use and heat mitigation capacity. Here we show the efficiency of cooling effect induced by irrigation of urban vegetation, represented by a novel metric, viz. urban water capacity, analogous to the heat capacity, across the contiguous United States (CONUS) during summertime via numerical simulations. The urban water capacity is calculated as the average irrigation depth per degree of urban temperature reduction; the values are 4.52 ± 0.77 mm day–1 °C–1 and 7.27 ± 1.27 mm day–1 °C–1 (mean ± standard deviation) for surface and near-surface air cooling, respectively, over the CONUS. The robustness of urban water capacity is further exemplified in an extreme heat wave event, during which the warming anomaly is partially offset by the additional cooling from urban irrigation. Estimates of water capacity provide a quantitative metric for evaluating the efficacy of irrigation in urban planning under current heat stress and future warming.

Anthropogenic aerosol emissions from North America and Europe have strong effects on the decadal variability of the West African monsoon. Anthropogenic aerosol effective radiative forcing is model dependent, but the impact of such uncertainty on the simulation of long-term West African monsoon variability is unknown. We use an ensemble of simulations with HadGEM3-GC3.1 that span the most recent estimates in simulated anthropogenic aerosol effective radiative forcing. We show that uncertainty in anthropogenic aerosol radiative forcing leads to significant uncertainty at simulating multi-decadal trends in West African precipitation. At the large scale, larger forcing leads to a larger decrease in the interhemispheric temperature gradients, in temperature over both the North Atlantic Ocean and northern Sahara. There are also differences in dynamic changes specific to the West African monsoon (locations of the Saharan heat low and African Easterly Jet, of the strength of the west African westerly jet, and of African Easterly Waves activity). We also assess effects on monsoon precipitation characteristics and temperature. We show that larger aerosol forcing results in a decrease of the number of rainy days and of heavy and extreme precipitation events and warm spells. However, simulated changes in onset and demise dates does not appear to be sensitive to the magnitude of aerosol forcing. Our results demonstrate the importance of reducing the uncertainty in anthropogenic aerosol forcing for understanding and predicting multi-decadal variability in the West African monsoon.

We develop a Hamiltonian Monte Carlo (HMC) sampler which solves a multi-parameter elastic full-waveform inversion (FWI) in a probabilistic setting for the first time. This gives novel access to the full posterior distribution for this type of highly non-linear inverse problem. Typically, FWI has focused on using gradient descent methods with proper regularization to iteratively update models to a minimum misfit value. Non-uniqueness and uncertainties are mostly in this approach. Bayesian inversions offer an alternative by assigning a probability to each model in model space given some data and prior constraints. The drawback is the need to evaluate a very large number of models. Random walks from Markov chains counter this effect by only exploring regions of model space where probability is significant. HMC method additionally incorporates gradient information, i.e. local structure, typically available for numerical waveform tomography experiments. So far, HMC has only been implented for acoustic FWI. We implement HMC for multiple 2D elastic FWI set-ups. Using parallelized wave propagation code, wavefields and kernels are computed on an regular numerical grid and projected onto basis functions. These gradients are subsequently used to explore the posterior space of different target models using HMC. The free parameters in these experiments are P and S velocity, and density. Although simulating Hamiltonian dynamics in the resulting phase space is approximated numerically, the results of the Markov chain are nevertheless very insightful. No prior tuning of kernels, data or model space is required, under the constraint that the sampler is properly tuned. After a burn in phase during which the mass matrix is iteratively optimized, the Markov chain is run on multiple nodes. After approximately 100,000 samples (combined from all nodes) the Markov chain mixes well. The resulting samples give access to the full posterior distribution, including the mean and maximum-likelihood models, conditional probabilities, inter-parameter correlations and marginal distributions.

We analyze peridotites from a wide range of tectonic settings to investigate relationships between olivine lattice preferred orientation (LPO) and deformation conditions in naturally deformed rocks. These samples preserve the five olivine LPO types (A through E-type) that rock deformation experiments have suggested are controlled by water content, temperature, stress magnitude, and pressure. The naturally deformed specimens newly investigated here (65 samples) and compiled from an extensive literature review (445 samples) reveal that these factors may matter less than deformation history and/or geometry. Some trends support those predicted by experimentally determined parametric dependence, but several observations disagree — namely that all LPO types are able to form at very low water contents and stresses, and that there is no clear relationship between water content and LPO type. This implies that at the low stresses typical of the mantle, LPO type more often varies as a function of strain geometry. Because olivine LPO is primarily responsible for seismic anisotropy in the upper mantle, the results of this study have several implications. These include (1) the many olivine LPO types recorded in samples from individual localities may explain some of the complex seismic anisotropy patterns observed in the continental mantle, and (2) B-type LPO – where olivine’s “fast axes” align perpendicular to flow direction – occurs under many more conditions than traditionally thought. This study highlights the need for more experiments, and the difficulty in using olivine LPO in naturally-deformed peridotites to infer deformation conditions.

Rapid retreat of the Larsen A and B ice shelves has provided important clues about the ice shelf destabilization processes. The Larsen C Ice Shelf, the largest remaining ice shelf on the Antarctic Peninsula, may also be vulnerable to future collapse in a warming climate. Here, we utilize multi-source satellite images collected over 1963–2020 to derive multidecadal time series of ice front, flow velocities, and critical rift features over Larsen C, with the aim of understanding the controls on its retreat. We complement these observations with modeling experiments using the Ice-sheet and Sea-level System Model to examine how front geometry conditions and mechanical weakening due to rifts affect ice shelf dynamics. Over the past six decades, Larsen C lost over 20% of its area, dominated by rift-induced tabular iceberg calving. The Bawden Ice Rise and Gipps Ice Rise are critical areas for rift formation, through their impact on the longitudinal deviatoric stress field. Mechanical weakening around Gipps Ice Rise is found to be a primary control on localized flow acceleration, leading to the propagation of two rifts that caused a major calving event in 2017. Capturing the time-varying effects of rifts on ice rigidity in ice shelf models is essential for making realistic predictions of ice shelf flow dynamics and instability. In the context of the Larsen A and Larsen B collapses, we infer a chronology of destabilization processes for embayment-confined ice shelves, which provides a useful framework for understanding the historical and future destabilization of Antarctic ice shelves.

Improving our understanding of glacial sliding is crucial for constraining basal drag in ice dynamics models. We use icequakes, sudden releases of seismic energy as the ice slides over the bed, to provide geophysical observations that can be used to aid understanding of the physics of glacial sliding and constrain ice dynamics models. These icequakes are located at the bed of an alpine glacier in Switzerland and the Rutford Ice Stream, West Antarctica, two extremes of glacial settings and spatial scales. We investigate a number of possible icequake source mechanisms by performing full waveform inversions to constrain the fundamental physics and stress release during an icequake stick-slip event. Results show that double-couple mechanisms best describe the source for the events from both glacial settings and the icequakes originate at or very near the ice-bed interface. We also present an exploratory method for attempting to measure the till shear modulus, if indirect, reflected icequake radiation is observed. The results of this study increase our understanding of how icequakes are associated with basal drag while also providing the foundation for a method of remotely measuring bed shear strength.

We investigate sharp changes in magnetic field that can produce Geomagnetically Induced Currents (GICs) which damage pipelines and power grids. We use one-minute cadence SuperMAG observations to find the occurrence distribution of magnetic field “spikes”. Recent studies have determined recurrence statistics for extreme events and charted the local time distribution of spikes; however, their relation to solar activity and conditions in the solar wind is poorly understood. We study spike occurrence during solar cycles 23 and 24, roughly 1995 to 2020. We find three local time hotspots in occurrence: the pre-midnight region associated with substorm onsets, the dawn sector associated with omega band activity, and the pre-noon sector associated with the Kelvin-Helmholtz instability occurring at the magnetopause. Magnetic field perturbations are mainly North-South for substorms and KHI, and East-West for omega bands. Substorm spikes occur at all phases of the solar cycle, but maximise in the declining phase. Omega-band and KHI spikes are confined to solar maximum and the declining phase. Substorm spikes occur during moderate solar wind driving, omega band spikes during strong driving, and KHI spikes during quiet conditions but with high solar wind speed. We show that the shapes of these distributions do not depend on the magnitude of the spikes, so it appears that our results can be extrapolated to extreme events.

Projections of the sea level contribution from the Greenland and Antarctic ice sheets rely on atmospheric and oceanic drivers obtained from climate models. The Earth System Models participating in the Coupled Model Intercomparison Project phase 6 (CMIP6) generally project greater future warming compared with the previous CMIP5 effort. Here we use four CMIP6 models and a selection of CMIP5 models to force multiple ice sheet models as part of the Ice Sheet Model Intercomparison Project for CMIP6 (ISMIP6). We find that the projected sea level contribution at 2100 from the ice sheet model ensemble under the CMIP6 scenarios falls within the CMIP5 range for the Antarctic ice sheet but is significantly increased for Greenland. Warmer atmosphere in CMIP6 models results in higher Greenland mass loss due to surface melt. For Antarctica, CMIP6 forcing is similar to CMIP5 and mass gain from increased snowfall counteracts increased loss due to ocean warming.

A document by Jida Wang. Click on the document to view its contents.

Absorption muon imaging is a technique that can measure density variations underground down to a few hundred meters. Then, it can prove to be useful in a mining environment: to help assess the ore bodies volumes and/or to monitor the underground for natural hazards that can happen at the surface because of the mining exploitation. Here we report the result of an experiment designed to test the capabilities and resolution power of a cosmic muon measurement in a mining environment compared to other standard geophysics tools: gravimetry and seismic studies. It consists of three independent measurement of a subset of the decommissioned iron mine of May-sur-Orne (France). The first one was made using a 50x50cm² micromegas based muon telescope installed at the deepest non-submerged level (50m underground) during 3 months. The second one is a gravimetry survey of the surface area inside the muon telescope acceptance cone. And the third one is a study of refracted and reflected seismic waves along a single line above the muon telescope location. The investigation volume was chosen because of the presence of surface risks (neighborhood), the unknown of some uncharted volume and the presence of an ore storage volume of several meter cubed that was used during the mine exploitation and which filling state is unknown. The data analysis showed that while muon tomography is able to detect the negative density anomaly of the storage volume, the gravimetry measurement is not sensible to it. However, the seismic study was able to detect the volume as well and its location and extension is compatible with the muon measurement.

Previous studies have shown that Strong Thermal Emission Velocity Enhancement (STEVE) events occur at the end of a prolonged substorm expansion phase. However, the connection between STEVE occurrence and substorms and the global high-latitude ionospheric electrodynamics associated with the development of STEVE and non-STEVE substorms are not yet well understood. The focus of this paper is to identify electrodynamics features that are unique to STEVE events through a comprehensive analysis of ionospheric convection patterns estimated from SuperDARN plasma drift and ground-based magnetometer data using the Assimilative Mapping of Geospace Observations (AMGeO) procedure. Results from AMGeO are further analyzed using principal component analysis and superposed epoch analysis for 32 STEVE and 32 non-STEVE substorm events. The analysis shows that the magnitude of cross-polar cap potential drop is generally greater for STEVE events. In contrast to non-STEVE substorms, the majority of STEVE events investigated accompany with a pronounced extension of the dawn cell into the pre-midnight subauroral latitudes, reminiscent of the Harang reversal convection feature where the eastward electrojet overlaps with the westward electrojet, which tends to prolong over substorm expansion and recovery phases. This is consistent with the presence of an enhanced subauroral electric field confirmed by previous STEVE studies. The global and localized features of high-latitude ionospheric convection associated with optical STEVE events characterized in this paper provide important insights into cross-scale magnetosphere-ionosphere coupling mechanisms that differentiate STEVE events from non-STEVE substorm events.

Climate change has been a common interest and the forefront of crucial political discussion and decision-making for many years. Shallow clouds play a significant role in understanding the Earth’s climate but they are challenging to interpret and represent in a climate model. By classifying these cloud structures there is a better possibility of understanding the physical structures of the clouds which would improve the climate model generation resulting in a better prediction of climate change or forecasting weather update. Clouds organise in many forms which makes it challenging to build traditional rule-based algorithms to separate cloud features. In this paper classification of cloud organization patterns was performed using a new scaled-up version of Convolutional Neural Network (CNN) named as EfficientNet as the encoder and UNet as decoder where they worked as feature extractor and reconstructor of fine grained feature map and was used as a classifier which will help experts to understand how clouds will shape the future climate. By using a segmentation model in a classification task it was shown that with a good encoder alongside UNet it is possible to obtain good performance from this dataset. Dice coefficient has been used for the final evaluation metric which gave the score of 66.26% and 66.02% for public and private (test set) leaderboard on Kaggle competition respectively.

A three-dimensional physical-biogeochemical ocean numerical model with eddy-permitting horizontal resolution was applied to simulate upper-ocean carbon cycle variabilities in the Tropical Maritime Continent (TMC) over the last decade (2010–2019). Forced by atmospheric and oceanic reanalysis products with high temporal resolution, the model showed robust consistency with the observed seasonality of pCO2 and atmospheric CO2 sink/source characteristics across the modeling domain. Within the TMC, the model results indicated strong CO2 degassing along the west of Sumatra-south of Java associated with the seasonal cycle of the upwelling system in the area. While acting as a full-year atmospheric CO2 source, the TMC exhibited pronounced interannual modulation in both pCO2 and sea-air CO2 flux over the last decade. Large-scale anomalous strong CO2 degassing from 2015 to 2016 in response to the evolution of the 2015/2016 El Niño was observed from the simulation results. Modulations related to the Indian Ocean Dipole (IOD), on the other hand, were confined along the west of Sumatra-south of Java with a higher magnitude compared with anomalies related to El Niño/La Niña. Simulation results also captured the asymmetric response of the upper-ocean carbon cycle to the IOD over the last decade, where anomalies during negative IOD (nIOD) were notably strong despite being indicated as weak nIOD events by the Dipole Mode Index.

The higher-order turbulence scheme, Cloud Layers Unified by Binormals (CLUBB), is known for effectively simulating the transition from cumulus to stratocumulus clouds within leading atmospheric climate models. This study investigates an underexplored aspect of CLUBB: its capacity to simulate near-surface winds and the Planetary Boundary Layer (PBL), with a particular focus on its coupling with surface momentum flux. Using the GFDL atmospheric climate model (AM4), we examine two distinct coupling strategies, distinguished by their handling of surface momentum flux during the CLUBB’s stability-driven substepping performed at each atmospheric time step. The static coupling maintains a constant surface momentum flux, while the dynamic coupling adjusts the surface momentum flux at each CLUBB substep based on the CLUBB-computed zonal and meridional wind speed tendencies. Our 30-year present-day climate simulations (1980-2010) show that static coupling overestimates 10-m wind speeds compared to both control AM4 simulations and reanalysis, particularly over the Southern Ocean (SO) and other midlatitude ocean regions. Conversely, dynamic coupling corrects the static coupling 10-m winds biases in the midlatitude regions, resulting in CLUBB simulations achieving there an excellent agreement with AM4 simulations. Furthermore, analysis of PBL vertical profiles over the SO reveals that dynamic coupling reduces downward momentum transport, consistent with the found wind-speed reductions. Instead, near the tropics, dynamic coupling results in minimal changes in near-surface wind speeds and associated turbulent momentum transport structure. Notably, the wind turning angle serves as a valuable qualitative metric for assessing the impact of changes in surface momentum flux representation on global circulation patterns.

Flow recession analysis, relating discharge Q and its time rate of change -dQ/dt, has been widely used to understand catchment scale flow dynamics. However, data points in the recession plot, the plot of -dQ/dt versus Q, typically form a wide point cloud due to noise and hysteresis in the storage-discharge relationship, and it is still unclear what information we can extract from the plot and how to understand the information. There seem to be two contrasting approaches to interpret the plot. One emphasizes the importance of the ensembles of many recessions (i.e., the lower envelope or a measure of central tendency), and the other highlights the importance of the event scale analysis and questions the meaning of the ensemble characteristics. In this study, we examine if those approaches can be reconciled. We utilize a machine learning tool to capture the point cloud using the past trajectory of discharge. Our results show that most of the data points can be captured using 5 days of past discharge. We show that we can learn the catchment scale flow recession dynamics from what the machine learned. We analyze patterns learned by the machine and explain and hypothesize why the machine learned those characteristics. The hysteresis in the plot mainly occurs during the early time dynamics, and the flow recession dynamics eventually converge to an attractor in the plot, which represents the master recession curve. We also illustrate that a hysteretic storage-discharge relationship can be estimated based on the attractor.

Supervised deep learning models have become a popular choice for seismic phase arrival detection. However, they don’t always perform well on out-of-distribution data and require large training sets to aid generalization and prevent overfitting. This can present issues when using these models in new monitoring settings. In this work, we develop a deep learning model for automating phase arrival detection at Nabro volcano using a limited amount of training data (2498 event waveforms recorded over 35 days) through a process known as transfer learning. We use the feature extraction layers of an existing, extensively-trained seismic phase picking model to form the base of a new all-convolutional model, which we call U-GPD. We demonstrate that transfer learning reduces overfitting and model error relative to training the same model from scratch, particularly for small training sets (e.g., 500 waveforms). The new U-GPD model achieves greater classification accuracy and smaller arrival time residuals than off-the-shelf applications of two existing, extensively-trained baseline models for a test set of 800 event and noise waveforms from Nabro volcano. When applied to 14 months of continuous Nabro data, the new U-GPD model detects 31,387 events with at least four P-wave arrivals and one S-wave arrival, which is more than the original base model (26,808 events) and our existing manual catalogue (2,926 events), with smaller location errors. The new model is also more efficient when applied as a sliding window, processing 14 months of data from 7 stations in less than 4 hours on a single GPU.