arXiv subjects

Simon Jeffery

Publications and source records attributed to Simon Jeffery.

9 recordsLinked to original sources

arXiv2026

LUCAS-MEGA: A Large-Scale Multimodal Dataset for Representation Learning in Soil-Environment Systems

Understanding soil is fundamental to agriculture, carbon cycling, and environmental sustainability, yet progress is limited by fragmented and heterogeneous datasets that constrain modeling to small-scale predictive settings rather than high-dimensional representation learning. We introduce LUCAS-MEGA, a large-scale multimodal dataset constructed through systematic data fusion of European soil-environment observations, with the LUCAS survey as its backbone. The fused dataset comprises over 70,000 samples and more than 1,000 features spanning physical, chemical, environmental, biological, and visual attributes, aggregated from 68 source datasets. To enable integration at scale, we develop SoilFuser, a multi-agent, human-in-the-loop data fusion pipeline that standardizes heterogeneous data formats and measurement protocols, resolves inconsistencies and invalid entries (e.g., unit inconsistencies, codebook mismatches, and erroneous values), incorporates natural language annotations, and harmonizes multimodal attributes and metadata into a unified, machine learning-ready feature space. The resulting dataset captures key characteristics of real-world soil observations, including multimodality, uneven feature coverage, and heterogeneous uncertainty. To demonstrate the usability of LUCAS-MEGA for data-driven modeling, we pretrain a multimodal tabular transformer (SoilFormer) using a self-supervised objective based on feature masking, achieving stable training, strong predictive performance, and representations that support uncertainty-aware prediction. We further show that the learned representations recover relationships consistent with established soil processes. LUCAS-MEGA is released with open access and is accompanied by composable, agent-friendly APIs that support structured querying and data-driven workflows.

cs.LG↗

arXiv2025

Agroseismology: unraveling the impact of farming practices on soil hydrodynamics

Farmed landscapes provide a natural laboratory to test how management reshapes near-surface hydrodynamics. Combining distributed acoustic sensing with physics-based hydromechanical modeling, we tracked minute-resolution, meter-scale changes across experimental fields with controlled tillage and compaction histories. We find that dynamic capillary effects, rate-dependent suction stresses during wetting and drying, govern transient stiffness and moisture redistribution in disturbed soils, producing sharp post-rain velocity drops from near-surface saturation and large hysteretic velocity rebounds driven by evapotranspiration. By pairing a seismic rainfall proxy with a drainage closure, we invert velocity changes to estimate evapotranspiration, revealing how disturbance alters flux partitioning and storage. These results establish agroseismology as a non-invasive, extendable tool to uncover soil hydromechanics, explain why conventional farming intensifies variability, and provide new constraints for Earth system models, land management, and hazard resilience.

physics.geo-ph↗

arXiv2025

KG-FGNN: Knowledge-guided GNN Foundation Model for Fertilisation-oriented Soil GHG Flux Prediction

Precision soil greenhouse gas (GHG) flux prediction is essential in agricultural systems for assessing environmental impacts, developing emission mitigation strategies and promoting sustainable agriculture. Due to the lack of advanced sensor and network technologies on majority of farms, there are challenges in obtaining comprehensive and diverse agricultural data. As a result, the scarcity of agricultural data seriously obstructs the application of machine learning approaches in precision soil GHG flux prediction. This research proposes a knowledge-guided graph neural network framework that addresses the above challenges by integrating knowledge embedded in an agricultural process-based model and graph neural network techniques. Specifically, we utilise the agricultural process-based model to simulate and generate multi-dimensional agricultural datasets for 47 countries that cover a wide range of agricultural variables. To extract key agricultural features and integrate correlations among agricultural features in the prediction process, we propose a machine learning framework that integrates the autoencoder and multi-target multi-graph based graph neural networks, which utilises the autoencoder to selectively extract significant agricultural features from the agricultural process-based model simulation data and the graph neural network to integrate correlations among agricultural features for accurately predict fertilisation-oriented soil GHG fluxes. Comprehensive experiments were conducted with both the agricultural simulation dataset and real-world agricultural dataset to evaluate the proposed approach in comparison with well-known baseline and state-of-the-art regression methods. The results demonstrate that our proposed approach provides superior accuracy and stability in fertilisation-oriented soil GHG prediction.

cs.LG↗

arXiv2024

EC 19529-4430: SALT identifies the most carbon- and metal-poor extreme helium star

EC 19529-4430 was identified as a helium-rich star in the Edinburgh-Cape Survey of faint-blue objects and subsequently resolved as a metal-poor extreme helium (EHe) star in the SALT survey of chemically-peculiar hot subdwarfs. This paper presents a fine analysis of the SALT high-resolution spectrum. EC 19529-4430 has $T_{\rm eff} = 20\,700 \pm250$\,K, $\log g /{\rm cm\,s^{-2}} = 3.49\pm0.03$, and an overall metallicity some 1.3 dex below solar; surface hydrogen is $\approx 0.5\%$ by number. The surface CNO ratio 1:100:8 implies that the surface consists principally of CNO-processed helium and makes EC 19529-4430 the coolest known carbon-poor and nitrogen-rich EHe star. Metal-rich analogues include V652 Her and GALEX J184559.8-413827. Kinematically, its retrograde orbit indicates membership of the galactic halo. No pulsations were detected in TESS photometry and there is no evidence for a binary companion. EC 19529-4430 most likely formed from the merging of two helium white dwarfs, which themselves formed as a binary system some 11 Gyr ago.

astro-ph.SR↗

arXiv2020

The SALT survey of helium-rich hot subdwarfs: methods, classification, and coarse analysis

A medium- and high-resolution spectroscopic survey of helium-rich hot subdwarfs is being carried out using the Southern African Large Telescope (SALT). Objectives include the discovery of exotic hot subdwarfs and of sequences connecting chemically-peculiar subdwarfs of different types. The first phase consists of medium-resolution spectroscopy of over 100 stars selected from low-resolution surveys. This paper describes the selection criteria, and the observing, classification and analysis methods. It presents 107 spectral classifications on the MK-like Drilling system and 106 coarse analyses ($T_{\rm eff}, \log g, \log y$) based on a hybrid grid of zero-metal non-LTE and line-blanketed LTE model atmospheres. For 75 stars, atmospheric parameters have been derived for the first time. The sample may be divided into 6 distinct groups including the classical `helium-rich' sdO stars with spectral types (Sp) sdO6.5 - sdB1 (74) comprising carbon-rich (35) and carbon-weak (39) stars, very hot He-sdO's with Sp $\lesssim$ sdO6 (13), extreme helium stars with luminosity class $\lesssim 5$ (5), intermediate helium-rich subdwarfs with helium class 25 -- 35 (8), and intermediate helium-rich subdwarfs with helium class $10 - 25$ (6). The last covers a narrow spectral range (sdB0 -- sdB1) including two known and four candidate heavy-metal subdwarfs. Within other groups are several stars of individual interest, including an extremely metal-poor helium star, candidate double-helium subdwarf binaries, and a candidate low-gravity He-sdO star.

astro-ph.SR↗

arXiv2020

Merged white dwarfs and nucleosynthesis

Orbital decay mechanisms argue that double white dwarf mergers are inevitable, but extremely rare. Whilst some mergers result in explosions, the survivors re-ignite helium and burn brightly for tens of thousands or millions of years. Candidate survivors include extreme helium stars, R CrB variables and various classes of helium-rich subluminous star. Nuclear waste on the survivors' surfaces provides evidence of the stars' nuclear history prior to and their nucleosynthesis during the merger. Extensive and deep spectroscopic surveys offer rich prospects for future discoveries.

astro-ph.SR↗

arXiv2018

Post-common envelope binary stars: Radiative Levitation and Blue Large- Amplitude Pulsators

Following the recent discovery of a new class of pulsating star, the blue large-amplitude pulsators (BLAPs), pulsation stability analysis was carried out on evolutionary models of post-common envelope (CE) ejection stars of 0.3 and 0.46 solar masses. These models subsequently evolve to become a low-mass helium white dwarf and a core helium-burning extreme horizontal-branch star respectively. We investigate the effects of atomic diffusion, particularly radiative levitation, on the pulsation behaviour of the models. We find that when the models have effective temperatures comparable to those of BLAPs, the inclusion of radiative levitation allows sufficient enhancement of heavy metals to produce opacity-driven fundamental mode pulsations with periods similar to those observed in BLAPs.

astro-ph.SR↗

arXiv2014

Discovery of extremely lead-rich subdwarfs: does heavy metal signal the formation of subdwarf B stars?

Hot subdwarfs represent a group of low-mass helium-burning stars formed through binary-star interactions and include some of the most chemically-peculiar stars in the Galaxy. Stellar evolution theory suggests that they should have helium-rich atmospheres but, because radiation causes hydrogen to diffuse upwards, a majority are extremely helium poor. Questions posed include: when does the atmosphere become chemically stratified and at what rate? The existence of several helium-rich subdwarfs suggests further questions; are there distinct subgroups of hot subdwarf, or do hot subdwarfs change their surface composition in the course of evolution? Recent analyses have revealed remarkable surface chemistries amongst the helium-rich subgroup. In this paper, we analyse high-resolution spectra of nine intermediate helium-rich hot subdwarfs. We report the discovery that two stars, HE 2359-2844 and HE 1256-2738, show an atmospheric abundance of lead which is nearly ten thousand times that seen in the Sun. This is measured from optical Pb IV absorption lines never previously seen in any star. The lead abundance is ten to 100 times that measured in normal hot subdwarf atmospheres from ultraviolet spectroscopy. HE 2359-2844 also shows zirconium and yttrium abundances similar to those in the zirconium star LS IV-14 116. The new discoveries are interpreted in terms of heavily stratified atmospheres and the general picture of a surface chemistry in transition from a new-born helium-rich subdwarf to a normal helium-poor subdwarf.

astro-ph.SR↗

arXiv2000

Radial velocities of pulsating subdwarf B stars: KPD 2109+4401 and PB 8783

High-speed spectroscopy of two pulsating subdwarf B stars, KPD 2109+4401 and PB 8783, is presented. Radial motions are detected with the same frequencies as reported from photometric observations and with amplitudes of ~2 km/sec in two or more independent modes. These represent the first direct observations of surface motion due to multimode non-radial oscillations in subdwarf B stars. In the case of the sdB+F binary PB 8783, the velocities of both components are resolved; high-frequency oscillations are found only in the sdB star and not the F star. There also appears to be evidence for mutual motion of the binary components. If confirmed, it implies that the F-type companion is >~1.2 times more massive than the sdB star, while the amplitude of the F star acceleration over 4 hours would constrain the orbital period to lie between 0.5 and 3.2d.

astro-ph↗