Detecting the source edges from potential field data is crucial in geological interpretations since it helps to identify contacts, faults, and other tectonic features. In recent years, many high-resolution filters have been introduced to generate sharp signals over the edges. In this study, we review 16 recent high-resolution filters, and introduce two frameworks for most of these filters. The filters are compared and discussed in terms of their accuracy and sharpness in detecting the edges using synthetic data. It is found that the filters based on the horizontal gradient are helpful in reducing the false information, while the filters using the second vertical derivative in the numerator can provide the edges with the highest resolution. The findings also show that the performance of many of the newer filters based on horizontal gradient or modified horizontal gradient versions is similar or lower compared to the previous horizontal gradient-based filters. Finally, we estimate and discuss the edges obtained from applying the filters to aeromagnetic data in the Montresor area (Canada), comparing the inferred edges with known geological structures. The obtained results demonstrate that the horizontal gradient-based filters are a useful tool for interpreting potential field data.
Climate-related sea level changes near the coast may deviate from what we observe in the open ocean. In effect, coastal changes result from the superposition of the global mean sea level rise, regional sea level trends, plus local contributions related to processes specific to nearshore areas. The latter operate on a broad range of timescales, from subdaily to multidecadal. They include ocean tides, atmospheric forcing, wind and waves, trapped coastal waves, coastal currents, river discharge in deltas and estuaries, natural climate modes and remote steric and mass effects. Changes in coastal ocean circulation driven by bathymetry or in coastal morphology and changing forcing factors can also impact coastal sea level. Such coastal processes not only produce coastal sea level changes, but also mediate the coastal response to open-ocean forcing. Vertical crustal motions are an additional factor causing sea level to vary with respect to the ground, in a proportion often larger than current climate-related sea level changes. Due to all involved processes, coastal sea level can significantly vary from one region to another. Here, we first discuss how coastal sea level is measured by in situ and space-based observing systems. Next, we review all processes causing sea level variations at the coast, from high-frequency extreme events to long-term multidecadal changes. We finally discuss different types of models developed for process understanding.
The limited resolution of surface seismic data remains a significant bottleneck in delineating fine-scale geological features, such as thin interlayers and small-scale faults. While DAS-VSP provides high-resolution imaging, its limited spatial coverage restricts its utility in large-scale exploration. Conversely, surface seismic data offer broader illumination but suffer from substantial resolution loss. Furthermore, existing deep learning approaches often fail to effectively bridge these modalities, as they rely heavily on fully paired training data or synthetic priors, leading to poor generalization and “model mismatch” under real-world conditions. To address these challenges, we propose a novel semi-supervised resolution enhancement framework that integrates joint borehole and surface exploration via a domain adaptation paradigm. A key component is the unsupervised Degradation Transformation Module (DTM), which models the resolution gap by mapping real-world data from an unknown degradation domain to a known domain defined by simulated priors. Unlike conventional methods, this transformation is learned using a Generative Adversarial Network (GAN) on unpaired data, allowing the network to capture complex degradation patterns, such as frequency attenuation and distortion, without requiring strict sample-level alignment. Once transformed into the known domain, a Transformer-CNN hybrid U-shaped Degradation Inverting Model (UDIM) performs the inversion to reconstruct high-resolution seismic profiles. By uniquely decoupling degradation modeling from the reconstruction process, the framework avoids overfitting to synthetic priors and ensures robustness in scenarios of data scarcity. Experiments on two field datasets validate the superior performance in seismic events restoration, fault continuity, and fidelity compared to state-of-the-art baselines. This study demonstrates the potential of borehole-guided domain transformation as a scalable, high-fidelity strategy for seismic imaging and resolution enhancement.
Solar–terrestrial physics, the science of interactions between the Sun and the Earth’s environment, is increasingly envisioned as a complex system consisting of regions of very different nature but strongly coupled to each other via multiple physical processes. Over time, a tremendous amount of observational data have been collected on the various solar–terrestrial physics subsystems: the Sun, the solar wind, Earth’s magnetosphere, ionosphere, and neutral atmosphere. Despite the quantity of data, measurements of the near-Earth and solar environments remain sparse with respect to the size of the solar–terrestrial system, implying that multiple datasets often need to be combined to gain insights into the physics at play. Besides, increasingly sophisticated numerical models have been built to elucidate the physics of those subsystems, and ongoing efforts aim at improving the interfacing of such models to get a system-level understanding. The solar–terrestrial physics community is facing the challenge of bringing together its various subcommunities whose combined data and expertise are needed to advance the science. While databases for observations and models exist, they are often catering for only part of the community and may not always follow the same standards and practice in terms of data access. Moreover, database and instrument maintenance requires continuity in funding and human power, posing an additional challenge. In this paper, we review some of the existing initiatives tackling the data challenge as well as emerging new data processing methods. We also discuss some of the current issues related to the production and management of observational data and propose ideas—such as mutualising resources in the form of ‘supersites’ and ensuring continuity in the measurements to overcome those challenges.
Deconvolution is a critical technique for improving seismic data resolution. It eliminates the filtering effects of the seismic wavelet to obtain reflection coefficients. However, deconvolution is a highly ill-posed problem, making accurate recovery of reflection coefficients challenging. Traditional methods obtain a stable solution to ill-posed problems by incorporating explicit model-based priors based on simplified assumptions. However, they often struggle to produce satisfactory results in complex geological environments or with noisy data. Although deep learning methods can effectively enhance deconvolution results by learning implicit priors from training data, they also suffer from certain limitations, such as generalization problems. Diffusion models offer a novel approach to acquiring prior information, known as generative priors. They directly model the approximate probability distribution of reflection coefficients by training a denoiser and learning to generate reflection coefficients from random noise. This denoiser provides the diffusion model with inherent robustness to random noise. We propose a 2D seismic deconvolution diffusion model (SeisDPS) that combines generative priors with model-based priors to achieve more accurate and high-fidelity reflection coefficients. However, existing diffusion-based deconvolution methods typically focus on processing 2D data, and even a few recent 3D approaches are computationally expensive. Then, building upon SeisDPS, we propose an algorithm for 3D seismic deconvolution (SeisDPS-3D). This 3D algorithm applies SeisDPS to one direction (e.g., inline profiles using SeisDPS) while applying unconditional sampling to the other (e.g., crossline profiles for continuous sampling correction) and vice versa. This unconditional sampling uses the prior information learned by the 2D diffusion model as 3D lateral constraints in orthogonal directions, thereby effectively improving the spatial continuity of the 3D deconvolution results. Under the same conditions, compared to the 2D diffusion model using slice processing, our method only increases the memory consumption by 5
Seismic exploration plays a pivotal role in subsurface characterization and has been greatly promoted by artificial intelligence techniques such as deep learning, yet its reliance on high-quality labeled data poses a significant challenge for conventional supervised learning methods. In recent years, label-efficient deep learning has emerged as a powerful solution to this bottleneck by leveraging unsupervised, self-supervised, semi-supervised, and weakly supervised paradigms. This survey provides a comprehensive overview of these learning paradigms and their growing impact on key stages of the seismic data processing workflow, including preprocessing, imaging and inversion, and geological interpretation. We first outline the theoretical foundations and representative architectures of each paradigm, highlighting their distinct supervision strategies and learning objectives. We then analyze recent advancements in applying label-efficient methods to tasks such as seismic denoising, interpolation, full waveform inversion, facies classification, fault detection, and salt body delineation. By systematically comparing methodologies across application scenarios, we identify their respective advantages, limitations, and domain-specific adaptations. Finally, we discuss the main challenges hindering large-scale deployment, including the lack of standardized benchmarks, difficulty in integrating geophysical constraints, and the interpretability gap, and we also suggest promising directions for future research. This review aims to serve as a comprehensive reference for geoscientists and machine learning practitioners seeking to harness label-efficient deep learning for intelligent and scalable seismic exploration.
Full waveform inversion (FWI) is an advanced velocity modeling method. FWI using active seismic data has high-resolution potential but often faces challenges dominated by cycle-skipping due to the insufficient low-frequency components. The ambient noise contains rich low-frequency components, but the inversion resolution is generally low. Joint FWI utilizing the complementary advantages of active and passive seismic data enables more accurate recovery of subsurface velocity structures. However, the conventional serial joint strategy is strongly dependent on the accuracy of passive source FWI, and complex interference from ambient noise introduces artifacts into inversion results, leading to instability of the serial strategy. To mitigate this problem, we develop a dynamically weighted parallel joint FWI. Firstly, an attenuated spatiotemporal window function combined with multidimensional deconvolution (MDD) seismic interferometry is designed to retrieve broadband body-wave Green’s functions from ambient noise. Subsequently, a weight function dynamically varying with iterations is established to integrate active and passive source data into a unified objective function framework, and the contributions of multi-source data are simultaneously utilized to construct a joint gradient for velocity reconstruction. Furthermore, we investigate the influence of parameter variations controlling the weight function on joint inversion performance, suggesting two parameter selection strategies, and verify their applicability and relative robustness. Numerical examples demonstrate that the proposed parallel joint FWI can effectively fuse the respective advantages of active seismic data and ambient noise, and robustly achieve high-resolution velocity modeling.
The atmosphere within Earth’s Geospace System, herein taken to be the region from the upper mesosphere (65 km) to the exobase ( 600 km), is cooling and contracting due to increasing carbon dioxide in its lower regions. These changes will affect many aspects of future orbiting satellite operations in Geospace. Recent assessments place the value of the orbiting-satellite-driven “space economy” at US 1.8 × 1012 within a decade. To assess Geospace change in a quantitative way that can facilitate a sustainable space economy, a Geospace System Observatory (GSO) is essential. The GSO would generate Geospace Data Records (GDRs) from which the rate and long-term extent of Geospace change can be determined with sufficient statistical significance to expedite confident international decisions and agreements regarding satellite and other operations in Geospace. In this paper, we present specific attributes of measurements intended to serve as GDRs. Foremost among these is high absolute accuracy that is driven by the need to limit trend error to a fraction of the anticipated trend magnitude over multiple instruments spanning decades. High accuracy also provides the statistical significance to enable confident decisions and shortens the time required to confidently detect trends. Other GDR aspects (record length; overlap/continuity; SI traceability) are also discussed. Examples of a variety of trend errors from current and past satellite sensors are given as a guide to GSO design requirements. A method for developing Level 1 design requirements for the GSO, used for developing systematic satellite observations for the troposphere, concludes the paper.
Over the past decades, numerous observations and model studies have provided substantial evidence that energetic particle precipitation (EPP) affects the chemistry and dynamics of the stratosphere. Concurrently, the significance of stratospheric dynamics, particularly in winter short-range and seasonal forecasts, has been highlighted. However, there has been little effort to integrate the knowledge from these two research fields. This review aims to bridge the gap between the Space Physics and Climate research communities. It will elucidate current knowledge on EPP and its impact on the chemistry and dynamics of the mesosphere and stratosphere, highlighting recent research. Additionally, it will present scientific findings demonstrating that EPP induced changes in the mesosphere and stratosphere can migrate downwards into the troposphere and reach the surface. Particularly during the eastward phase of the Quasi-Biennial Oscillation and/or close to a Sudden Stratospheric Warming (SSW), EPP can significantly impact stratospheric dynamics projected onto the North Atlantic Oscillation or Northern Annular Mode. The review highlights EPP as a potential moderator of SSWs in terms of their occurrence, timing, and strength, which are all crucial parameters for short-range and seasonal forecasts for the northern hemispheric winter. Moreover, it presents research demonstrating that the EPP chemical-dynamical coupling is becoming stronger in an atmosphere influenced by climate change. Bridging the gap between space physics and climate research is essential, as the natural variability of the atmosphere underpins the climate signal. Better prediction of SSWs and their effects on northern winter weather is crucial in preparing for extreme weather events and supporting economic activities. This interdisciplinary approach can enhance our overall understanding of the Earth’s atmosphere and its complex processes.
The importance of uncertainty estimates for Essential Climate Variable (ECV) data records is well recognized. Most ECV observing systems now estimate and report uncertainties as part of the measurement and retrieval procedure instead of leaving uncertainty characterization to a diagnostic "ex-post" validation/evaluation procedure. This paper focuses on the validation or evaluation of these prognostic “ex-ante" uncertainties provided with satellite ECV climate data records. In recent years, established validation protocols (primarily focused on the validation of the measurements themselves) have been extended to provide feedback on the prognostic uncertainties. Moreover, dedicated methods for uncertainty validation have been developed and applied across the Earth observation community. In this review, we classify, describe, and illustrate these approaches under three categories: (1) those relying on comparison to independent in-situ or other ground-based reference data, (2) those exploiting self-co-locations, i.e., the comparison between consecutive measurements by the same instrument under appropriate conditions, and (3) those integrating a multitude of correlative (satellite) data sets. The principles behind each method are described with a common nomenclature, facilitating a clear identification of strengths and weaknesses, and the illustrations are chosen to cover the three Global Climate Observing System (GCOS) ECV domains (atmosphere, ocean, terrestrial) toward the aim of facilitating cross-domain knowledge transfer. Finally, the most common and critical gaps and challenges are listed to support the scoping of future work.
This study aims to improve the uncertainty estimates of soil moisture datasets produced by merging various satellite products via inverse-variance weighting. In this scheme, the weight of each sensor depends on its specific uncertainty derived from Triple Collocation Analysis (TCA). However, the TCA-derived uncertainties are themselves uncertain due to finite sample sizes, introducing a second-order uncertainty we denote the ‘uncertainty of the uncertainty’. Here, we estimate it empirically by bootstrapping and find that it follows a power-law relationship as a function of the number of collocated observations, whose exponent is comparable to the analytical solution for simple error models. Furthermore, we propose an extended scheme that includes the resulting uncertainty of the weights in the uncertainty estimate of the merged dataset. The proposed scheme is tested on soil moisture retrievals from three different satellite sensors, the active Advanced Scatterometer (ASCAT), the passive Soil Moisture Active Passive (SMAP), and the passive Soil Moisture And Ocean Salinity (SMOS) sensors. Comparing the improved uncertainty estimates to skill metrics calculated against the global reanalysis product ERA5-Land confirms that they indeed better describe (spatial) uncertainty variations of the merged soil moisture product against the reference dataset. The findings of this study underscore the necessity of advancing uncertainty quantification methods in satellite-retrieved climate data sets.
When quantifying changes over time in the natural environment, the stability of the observations used should be considered. Stability conceptually refers to how accurately true geophysical changes and trends are reflected in observational data. We argue the need for a better approach to defining and quantifying stability consistently across climate data records. We propose that the appropriate stability metric is the stability uncertainty for specified spatial and temporal scales. We formally define stability uncertainty by analogy with metrological measurement uncertainty. Informally, stability uncertainty informs data analysts about the plausible magnitude of a non-geophysical contribution to trend values arising solely from the observing system. Neglecting the stability uncertainty leads to overconfident assessment of the significance of geophysical trends inferred from observations. We recommend that adopting this metric would greatly improve the clarity and practical impact of the Global Climate Observing System (GCOS) statements of requirement for stability of essential climate variable (ECV) products. Moreover, GCOS stability requirements would then become a useful resource for users of ECV products when evaluating and interpreting trends in observations, helping them avoid unjustified claims for the significance of computed trends; a synthetic illustration of such usage is provided.
Different types of gravity anomalies are engaged in geophysical and geodetic tasks. Whether they are used for regional or global applications, they require efficient calculations. All variants are based on the so-called free-air gravity anomalies. Mean free-air gravity anomalies on an equidistant grid are needed for gravity field modeling. Three possible ways of compiling mean free-air gravity anomalies are discussed in detail. One method is via simple Bouguer gravity anomalies, the second, more time-consuming method is via complete Bouguer gravity anomalies, and the third method is via topographic-isostatic reductions, which is a tedious task. In flat areas, the differences between using any of the three methods should not be significant. However, in mountainous regions, each dependency can negatively affect the interpolation process of gravity anomalies. The reduced gravity anomalies should be as smooth as possible in order to minimize the interpolation error which is inherent in the interpolation of the information in the arbitrarily distributed gravity observation points to obtain block average signals. This study investigates the effects of Bouguer and topographic-isostatic reductions on the accuracy of the mean gravity anomalies and the resulting gravimetric geoid model. The numerical results indicate that complete Bouguer approximations improve the accuracy of the geoid model by a few millimeters. Therefore, this method should be used to predict mean gravity anomalies, especially in mountainous regions, in few of the 1 cm geoid determination.
Full-waveform inversion estimates subsurface properties by minimizing the misfit between observed and modeled data. However, conventional deterministic approaches are highly sensitive to noise, dependent on the starting model and prone to converging to local minima of the cost function. Bayesian approaches offer a viable alternative, enhancing solution space exploration and providing uncertainty quantification. This study compares three Bayesian inversion methods: Stochastic Newton Markov Chain Monte Carlo, Ensemble Smoother with Multiple Data Assimilation and Annealed Stein Variational Gradient Descent. To mitigate ill-conditioning, reduce computational cost and lower problem dimensionality, we incorporate Discrete Cosine Transform compression of model and data spaces. These methods are evaluated under three scenarios: optimal conditions as a baseline, a less informative prior to assess initialization sensitivity and a more realistic setting with increased noise, missing traces and wrong assumptions about the source wavelet and noise level. These tests are conducted on a synthetic example based on a portion of the Marmousi model. Results indicate that the ensemble-based approach is the most computationally efficient and is robust to initialization, but strongly underestimates uncertainties. The Markov Chain method, while computationally demanding, is robust to initialization and effectively captures uncertainties. The variational inference approach, with intermediate computational cost, achieves good performance in complex scenarios, but it is more sensitive to initialization due to its deterministic nature. This study provides insights into the advantages and limitations of these three Bayesian approaches to acoustic full-waveform inversion and highlights the benefits of the Discrete Cosine Transform compression.
Gravitational forces are the major forces acting on near-Earth orbiting (e.g., altimetry) satellites. We perform a review of Earth’s mean time-variable gravity (TVG) field models developed in the past 23 years (2000–2023). This includes the models developed using CHAMP, GRACE, GRACE-FO, GOCE, SLR (Satellite Laser Ranging), and DORIS measurements. Some of these models contain just secular terms, while more recent models include also periodic (annual and semi-annual) variations of the Earth’s gravity. We show the impact of these models on precise orbit determination (POD) of selected altimetry satellites, namely TOPEX/Poseidon, Jason-1, Jason-2, and Jason-3 at the time interval from 1992 to 2023. The impact of these models is assessed for different orbit parameters as well as the root-mean-square (RMS) and mean values of SLR observation residuals and orbit differences. Furthermore, the impact of these models on altimetry (single- and multi-satellite) sea surface height crossover differences, radial errors, geographically correlated mean errors, and their trends is analyzed. We have found that the CNES RL05MF model derived using data of 1985–2022 performs best among the models tested in this study, particularly for the Jason-3 time span (2016–2023). Using this model reduces the RMS values of SLR observation residuals from 2.56 cm (for pre-CHAMP model GRIM5-C1) to 1.48 cm for this satellite. The RMS values of orbit differences in the radial direction fit within 0.7–0.8 cm for most recent TVG models, while using old GRIM5-C1 would result in 1.9 cm differences. It is important to reprocess regularly Earth’s TVG data covering the longest time span to minimize extrapolation errors of the models.
Over the past few decades, calculating a cm-precise geoid has been a major pursuit of geodesists. Numerous geoid modelling methods exist because geoid modelling theory was necessarily developed with some assumptions, which are handled differently by respective research groups. In the literature, numerous papers discuss in detail the theory of (aspects of) any one computational method. There are also numerous papers with empirical comparisons between different geoid modelling methods, where differences in excess of 1 cm are typically found. Almost all previous studies simply provide numerical comparisons of final geoid models, but more work is required to find out what causes the discrepancies. This study reviews the similarities and dissimilarities among three different geoid modelling methods: the approach followed at Curtin University of Technology, the Stokes-Helmert method, and the method of Least Squares Modification of Stokes formula with additive corrections. These methods may provide varying solutions due to, including but not limited to, choices of parameters and freely available data (Global Geopotential Models, Digital Elevation Models), kernel modifications, handling of the dataset (gridding, merging, interpolation etc.), etc. However, only the four following aspects are covered in this paper, i.e. the handling of 1) topographic masses, 2) atmospheric masses, 3) the ellipsoidal shape of the Earth, and 4) downward continuation. The major motivation behind this study is that with the pursuit of the cm-precise geoid, different geoid modelling methods should agree with one another within a given threshold because methods differ primarily in handling the discussed four aspects. Therefore, this study reviews these four aspects and compares them to identify the possible causes of discrepancies between the geoid modelling methods. Further, since no numerical comparisons are available on handling these corrections individually in different methods, this paper compares the mathematical formulations and suggests strategies as a roadmap to quantify the identified discrepancies among the methods.
This paper is a collaborative effort that originated at the International Space Science Institute Workshop on “Physical Links between Weather and Climate in Space and the Lower Atmosphere” held on January 22-26, 2024. Our goals are to survey the role of tides in facilitating the coupling of the lower and upper atmosphere and identify pathways forward that address challenges to our current understanding. To that end, we provide a brief review of the physics of atmospheric tides and the sources of their day-to-day and seasonal variability during quiet geomagnetic conditions. We identify the mechanisms that couple vertically propagating atmospheric tides to variations in thermosphere–ionosphere wind, composition, and plasma. Each process is punctuated with examples showcasing state-of-the-art observations or models, and requirements for scientific progress. A recurrent theme is a thermospheric measurement gap region between 100 and 200 km that precludes direct observations of tidal vertical coupling processes and their day-to-day variability.
