Saurabh Kaushik, Lalit Maurya, Elizabeth Tellman
et al.
The automated mapping of debris covered glaciers remains challenging due to spectral similarity between supraglacial debris (on-glaciers) and periglacial debris (off-glaciers). Deep learning offers promising capabilities, yet the lack of high-quality publicly available datasets and limited exploration of optimal model architecture constrain progress in this domain. To address this, we introduce the Global Supraglacial Debris Cover Dataset (GSDD), consisting of 1876 images (∼49,000.00 km2) collected globally from diverse glacierized regions, including High Mountain Asia, Andes, Western Canada, Alaska, and Swiss Alps, to incorporate the heterogeneity of glacial features and environments. This multisource remote sensing dataset includes 10 spectral bands—Blue, Green, Red, Near-Infrared, Shortwave Infrared (SWIR1 & SWIR2), Normalized Difference Rock Index (NDRI), Slope, Elevation, and Velocity—providing critical information to distinguish glacier debris. To evaluate the efficacy of deep learning models for mapping glacier debris, we compare Prithvi Geo-Foundational Model (GFM) combined with multiple decoders, CNN-based models (UNet, Attention U-Net, and DeepLabv3+), a Vision Transformer-based model (TransNorm), and variant of the Prithvi GFM (i.e., UViT). Our results show Prithvi GFM with UperNet decoder outperformed all, achieving mIoU = 0.80 and F1-score = 0.91, surpassing DeepLabv3+ (0.71 mIoU), Attention U-Net (0.73), U-Net (0.72), TransNorm (0.71), and UViT (0.70). Our results demonstrate significant methodological advances in accurately mapping glacier termini, along with the identification of glacier snouts. Feature analysis identified the optimal band combination (B-G-NIR-SWIR-Slope-Elevation) for debris mapping. The GSDD dataset enables direct comparison, development, and evaluation of deep learning models, supporting advancement in fast and reliable automated glacier mapping.
The risk of soil compaction by agricultural machinery threatens the structure and productivity of tilled soils. However, a quantitative understanding of how specific compaction loads alter the three-dimensional (3D) macropore architecture of tilled soil is still limited. This study employed X-ray computed tomography (CT) to quantitatively characterize the evolution of the 3D macropore network in clay soil under a series of controlled compaction pressures (0, 30, 60, 90, and 120 kPa). Our results revealed a non-monotonic response of macropore number to compaction, which initially increased due to the fragmentation of large pores before declining, peaking at 90 kPa. Most critically, we identified 90 kPa as a critical threshold, beyond which macroporosity and the volume of elongated beneficial pores underwent drastic reductions of 64.8% and 46.6%, respectively. Compaction significantly reduced pore connectivity and surface area, with larger macropores (>1000 μm) proving most vulnerable. The study establishes a quantitative link between applied pressure and specific pore-scale damage, providing a scientific basis for designing agricultural machinery with ground pressures below this critical threshold to preserve soil structure and function after tillage.
The FCC-ee phase of a Future Circular Collider is generating great interest due to its versatility, allowing the study of various electroweak thresholds, $Z$, $WW$, $ZH$, and $t \bar{t}$. Electroweak precision physics is complemented by flavour physics measurements based on the unprecedented statistics attainable at the $Z$ pole, and benefiting from the low-background experimental environment (similar to Belle II), and from the production of the full spectrum of hadron species together with large boosts (similar to LHCb). A wide range of measurements is possible, spanning a rich variety of physics cases in both quark and lepton flavour physics sectors. Other electroweak thresholds can also be considered in this endeavour. A commensurate effort from the theory community will be needed to interpret future measurements. I present an overview of the broad potential of the FCC-ee flavour physics program.
This paper presents HERO (Hierarchical Testing with Rabbit Optimization), a novel black-box adversarial testing framework for evaluating the robustness of deep learning-based Prognostics and Health Management systems in Industrial Cyber-Physical Systems. Leveraging Artificial Rabbit Optimization, HERO generates physically constrained adversarial examples that align with real-world data distributions via global and local perspective. Its generalizability ensures applicability across diverse ICPS scenarios. This study specifically focuses on the Proton Exchange Membrane Fuel Cell system, chosen for its highly dynamic operational conditions, complex degradation mechanisms, and increasing integration into ICPS as a sustainable and efficient energy solution. Experimental results highlight HERO's ability to uncover vulnerabilities in even state-of-the-art PHM models, underscoring the critical need for enhanced robustness in real-world applications. By addressing these challenges, HERO demonstrates its potential to advance more resilient PHM systems across a wide range of ICPS domains.
Julio Arrechea, Stefano Liberati, Hooman Neshat
et al.
Black hole spacetimes that exhibit integrable singularities have gained considerable interest as alternatives to both regular and singular black holes. Unlike most known regular black hole solutions, these models evade the formation of an inner horizon, thereby circumventing the well-known instability issues associated with such structures. Moreover, it has been suggested that the finite tidal forces near integrable singularities, may allow for a traversable extension beyond them. In this work, we present a set of arguments -- both theoretical, concerning test-field perturbations and the accumulation of matter at the singularity, and practical, related to the behavior of physical probes and extended objects -- with the aim of assessing the validity of the proposed integrability condition, and the feasibility of traversing such singularities. Our analysis highlights key subtleties that challenge the viability of said extensions as alternatives to regular black holes, and underscores the need for a more rigorous investigation of their physical implications.
Estimating the 3D poses of hands and objects from a single RGB image is a fundamental yet challenging problem, with broad applications in augmented reality and human-computer interaction. Existing methods largely rely on visual cues alone, often producing results that violate physical constraints such as interpenetration or non-contact. Recent efforts to incorporate physics reasoning typically depend on post-optimization or non-differentiable physics engines, which compromise visual consistency and end-to-end trainability. To overcome these limitations, we propose a novel framework that jointly integrates visual and physical cues for hand-object pose estimation. This integration is achieved through two key ideas: 1) joint visual-physical cue learning: The model is trained to extract 2D visual cues and 3D physical cues, thereby enabling more comprehensive representation learning for hand-object interactions; 2) candidate pose aggregation: A novel refinement process that aggregates multiple diffusion-generated candidate poses by leveraging both visual and physical predictions, yielding a final estimate that is visually consistent and physically plausible. Extensive experiments demonstrate that our method significantly outperforms existing state-of-the-art approaches in both pose accuracy and physical plausibility.
Topography determines run-off direction, redistributes groundwater, and affects land surface solar radiation loads and the associated evaporative forcing, consequently, topography can modulate the impact of drought and heat waves on ecosystems. This topographic modulation effect, which typically occurs at the local scale, is often overlooked when assessing ecosystem drought responses using moderate-to-coarse spatial resolution satellite observations, such as the Moderate-resolution Imaging Spectroradiometer (MODIS) imagery. Sentinel-2 and Landsat imagery with finer resolution are suitable for monitoring changes at the local scale, however, studies relying on single vegetation metrics may fail to get a holistic picture of vegetation drought responses, particularly for forests that have complex physiological mechanisms. Here, we performed a comprehensive assessment of the topographic effects on coniferous forest responses to the severe 2018 drought in Scandinavia, using 6 vegetation seasonal metrics during 2017–2021 from the Sentienl-2 High-Resolution Vegetation Phenology and Productivity (HR-VPP) products. We found significant differences (p < 0.05) between sunny and shady aspects, between higher and lower elevations, and between steep and gentle slopes, regarding the maximum impact time, forest drought resistance, and resilience. Specifically, the sunny aspects and steep slopes were related to higher risks of delayed impacts and low resistance and resilience, and elevation and slope were more powerful in regulating the phenology shift and greening rate loss. We also identified different sensitivity in greenness and productivity to topographic effect and greater sensitivity of spring phenology to topographic differences as compared to autumn phenology. The study demonstrates vegetation drought responses represented by multiple seasonal metrics, reveals the prevalent topographic effects at the local scale, and quantifies the magnitudes of the effects with regional statistics.
Study region: The midstream of the Hei River Basin (HRB). Study focus: The study investigated the changes in hydrological elements at different time scales in the basin from 1982 to 2020. We used the Budyko coupled balance equation to quantitatively assess the contributions of climate change and human activity-induced land surface changes to runoff during different periods. In addition, the driving factors of flow consumption within the basin were analyzed using wavelet coherence and generalized additive model (GAM). New hydrological insights for the region: Prior to 2001, there was a significant decrease in annual-scale runoff in the midstream. After that, a partial recovery was observed due to the influence of the Ecological Water Transfer Project (EWTP), while the NDVI trend exhibited an opposite pattern. The study indicates that human-induced modification of the underlying surface is the primary driving factor for the decrease in runoff, with land surface factors contributing 84.69% and 65.27% to the runoff reduction during two periods of change, respectively. Further research on the driving forces of runoff consumption reveals that vegetation cover and climate factors jointly regulate the variations in runoff consumption in the midstream, with multifactor effects surpassing those of individual factors, and NDVI emerges as the predominant controlling factor for runoff consumption. Moreover, all anthropogenic factors exhibit a high explanatory power for runoff consumption.
We explore the usage of large language models (LLM) in human-in-the-loop human-in-the-plant cyber-physical systems (CPS) to translate a high-level prompt into a personalized plan of actions, and subsequently convert that plan into a grounded inference of sequential decision-making automated by a real-world CPS controller to achieve a control goal. We show that it is relatively straightforward to contextualize an LLM so it can generate domain-specific plans. However, these plans may be infeasible for the physical system to execute or the plan may be unsafe for human users. To address this, we propose CPS-LLM, an LLM retrained using an instruction tuning framework, which ensures that generated plans not only align with the physical system dynamics of the CPS but are also safe for human users. The CPS-LLM consists of two innovative components: a) a liquid time constant neural network-based physical dynamics coefficient estimator that can derive coefficients of dynamical models with some unmeasured state variables; b) the model coefficients are then used to train an LLM with prompts embodied with traces from the dynamical system and the corresponding model coefficients. We show that when the CPS-LLM is integrated with a contextualized chatbot such as BARD it can generate feasible and safe plans to manage external events such as meals for automated insulin delivery systems used by Type 1 Diabetes subjects.
Abstract Many chemical processes depend non‐linearly on temperature. Gravity‐wave‐induced temperature perturbations have been shown to affect atmospheric chemistry, but accounting for this process in chemistry‐climate models has been a challenge because many gravity waves have scales smaller than the typical model resolution. Here, we present a method to account for subgrid‐scale orographic gravity‐wave‐induced temperature perturbations on the global scale for the Whole Atmosphere Community Climate Model. Temperature perturbation amplitudes T^ consistent with the model's subgrid‐scale gravity wave parameterization are derived and then used as a sinusoidal temperature perturbation in the model's chemistry solver. Because of limitations in the parameterization, we explore scaling of T^ between 0.6 and 1 based on comparisons to altitude‐dependent T^ distributions of satellite and reanalysis data, where we discuss uncertainties. We probe the impact on the chemistry from the grid‐point to global scales, and show that the parameterization is able to represent mountain wave events as reported by previous literature. The gravity waves for example, lead to increased surface area densities of stratospheric aerosols. This increases chlorine activation, with impacts on the associated chemical composition. We obtain large local changes in some chemical species (e.g., active chlorine, NOx, N2O5) which are likely to be important for comparisons to airborne or satellite observations, but the changes to ozone loss are more modest. This approach enables the chemistry‐climate modeling community to account for subgrid‐scale gravity wave temperature perturbations interactively, consistent with the internal parameterizations and are expected to yield more realistic interactions and better representation of the chemistry.
Serjoscha W. Evers, Kimberley E. J. Chapelle, Walter G. Joyce
Abstract Trionychid (softshell) turtles have a peculiar bauplan, which includes shell reductions and cranial elongation. Despite a rich fossil record dating back to the Early Cretaceous, the evolutionary origin of the trionychid bauplan is poorly understood, as even old fossils show great anatomical similarities to extant species. Documenting structural detail of fossil trionychids may help resolve the evolutionary history of the group. Here, we study the cranial and mandibular anatomy of Plastomenus thomasii using µCT scanning. Plastomenus thomasii belongs to the Plastomenidae, a long-lived (Santonian–Eocene) clade with uncertain affinities among trionychid subclades. The skulls of known plastomenids are characterized by unusual features otherwise not known among trionychids, such as extremely elongated, spatulate mandibular symphyses. We use anatomical observations for updated phylogenetic analyses using both parsimony and Bayesian methods. There is strong support across methods for stem-cyclanorbine affinities for plastomenids. The inclusion of stratigraphic data in our Bayesian analysis indicates that a range of Cretaceous Asian fossils including Perochelys lamadongensis may be stem-trionychids, suggesting that many features of trionychid anatomy evolved prior to the appearance of the crown group. Divergence time estimates from Bayesian tip-dating for the origin of crown Trionychia (134.0 Ma) and Pan-Trionychidae (123.8 Ma) constrain the evolutionary time span during which the trionychid bauplan has evolved to a range of < 11 million years. Bayesian rate estimation implies high morphological rates during early softshell turtle evolution. If correct, plastomenids partially fill the stratigraphic gap which results from shallow divergence times of crown cyclanorbines during the late Eocene.
Physics-informed neural networks (PINNs) have gained prominence for their capability to tackle supervised learning tasks that conform to physical laws, notably nonlinear partial differential equations (PDEs). This paper presents "PINNs-TF2", a Python package built on the TensorFlow V2 framework. It not only accelerates PINNs implementation but also simplifies user interactions by abstracting complex PDE challenges. We underscore the pivotal role of compilers in PINNs, highlighting their ability to boost performance by up to 119x. Across eight diverse examples, our package, integrated with XLA compilers, demonstrated its flexibility and achieved an average speed-up of 18.12 times over TensorFlow V1. Moreover, a real-world case study is implemented to underscore the compilers' potential to handle many trainable parameters and large batch sizes. For community engagement and future enhancements, our package's source code is openly available at: https://github.com/rezaakb/pinns-tf2.
Foundation Industries (FIs) constitute glass, metals, cement, ceramics, bulk chemicals, paper, steel, etc. and provide crucial, foundational materials for a diverse set of economically relevant industries: automobiles, machinery, construction, household appliances, chemicals, etc. Reheating furnaces within the manufacturing chain of FIs are energy-intensive. Accurate and real-time prediction of underlying temperatures in reheating furnaces has the potential to reduce the overall heating time, thereby controlling the energy consumption for achieving the Net-Zero goals in FIs. In this paper, we cast this prediction as a regression task and explore neural networks due to their inherent capability of being effective and efficient, given adequate data. However, due to the infeasibility of achieving good-quality real data in scenarios like reheating furnaces, classical Hottel's zone method based computational model has been used to generate data for model training. To further enhance the Out-Of-Distribution generalization capability of the trained model, we propose a Physics-Informed Neural Network (PINN) by incorporating prior physical knowledge using a set of novel Energy-Balance regularizers.
Meagan Sundstrom, Andy Schang, Ashley B. Heim
et al.
Engaging in interactions with peers is important for student learning. Many studies have quantified patterns of student interactions in in-person physics courses using social network analysis, finding different network structures between instructional contexts (lecture and lab) and styles (active and traditional). Such studies also find inconsistent results as to whether and how student-level variables (e.g., grades and demographics) relate to the formation of interaction networks. In this cross-sectional research study, we investigate these relationships further by examining lecture and lab interaction networks in four different remote physics courses spanning various instructional styles and student populations. We apply statistical methods from social network analysis -- exponential random graph models -- to measure the relationship between network formation and multiple variables: students' discussion and lab section enrollment, final course grades, gender, and race/ethnicity. Similar to previous studies of in-person courses, we find that remote lecture interaction networks contain large clusters connecting many students, while remote lab interaction networks contain smaller clusters of a few students. Our statistical analysis suggests that these distinct network structures arise from a combination of both instruction-level and student-level variables, including the learning goals of each instructional context, whether assignments are completed in groups or individually, and the distribution of gender and major of students enrolled in a course. We further discuss how these and other variables help to understand the formation of interaction networks in both remote and in-person physics courses.
Eduardo Martínez, Javier Montaño-Domínguez, Héctor Novales-Sánchez
et al.
Current experimental data guarantees the presence of physics beyond the Standard Model in the neutrino sector. The responsible physical description might show itself through virtual effects on low-energy observables. In particular, massive neutrinos are able to produce contributions to the triple gauge coupling $WWγ$. The present paper deals with the calculation, estimation and analysis of one-loop contributions from Majorana neutrinos to the Lorentz-covariant $WWγ$ parametrization. Our calculations show that CP-odd effects vanish exactly, whereas CP-even contributions, $Δκ$ and $ΔQ$, remain. According to our estimations, the effects from heavy neutrinos with masses in the range of hundreds of GeVs dominate over those from light neutrinos. This investigation shows that contributions from heavy Majorana neutrinos to the anomaly $Δκ$ could be as large as $\sim{\cal O}(10^{-3})$, one order of magnitude below the Standard-Model contribution. We find that the International Linear Collider, sensitive to triple gauge couplings participating in $WW$ production, might measure these effects in electron-positron collisions at a center-of-mass energy of $800\,{\rm GeV}$, as long as heavy-neutrino masses are $\gtrsim300\,{\rm GeV}$ and below $\sim1500\,{\rm GeV}$.
Christopher Oze, Joshua B. Smaill, Catherine M. Reid
et al.
Plant nutrients such as potassium (K) may be limited in soil systems and additions (i.e., fertilizer) are commonly required. Glaucony is a widely distributed and abundant marine-derived clay mineral present in soils worldwide which may serve as a source of potassium. The South Island of New Zealand contains numerous deposits of glaucony-rich rocks and related soils providing an opportunity to explore how glaucony might be a beneficial source of potassium. Here, the geochemistry of glaucony and its suitability as a mineral source of soil K from four deposits in New Zealand was examined using spatially resolved chemical analyses and dissolution experiments. Geochemical and morphological analyses revealed that glaucony from all deposits were K-enriched and were of the evolved (6%−8% K<sub>2</sub>O) to highly evolved type (>8% K<sub>2</sub>O). Glaucony derived from growth inside pellets contain elevated K and Fe concentrations compared to bioclast-hosted glaucony. Solubility analysis showed that K was released from glaucony at rates higher than any other metal present in the mineral. Additionally, decreasing the pH and introducing an oxidizing agent (i.e., birnessite which is ubiquitous in soil environments) appeared to accelerate K release. Trace metals including Cr, Zn, Cu, and Ni were present in the solid phase analysis; however, further investigation with a focus on Cr revealed that these elements were released into solution at low concentrations and may present a source of soil micronutrients. These results suggest that glaucony may offer a source of slow releasing K into soils, and so could be used as a locally sourced environmentally sustainable K resource for agriculture, whether in New Zealand or worldwide.