Reservoir computing (RC), a brain-inspired neuromorphic algorithm, offers simplicity and efficiency for processing spatiotemporal signals. However, conventional RC systems face limitations in handling diverse temporal scales and spatial complexities due to invariant temporal dynamics. This study introduces a temporally reconfigurable RC system utilizing ultrathin, flexible, all-solid-state electrolyte-gated thin-film transistors (UFLEX TFTs) with high performance: an on/off ratio of ≈107, endurance beyond 2.5 × 104 pulses, and low variability. UFLEX TFTs, based on molybdenum disulfide (MoS2) channels and organic-inorganic hybrid AlOx dielectrics, enable modulation of temporal dynamics via simple electrical signals. The system maintains mechanical flexibility and robust performance after bending tests. By extracting features across varied temporal and spatial scales, it achieves classification accuracies of 90.3% for CIFAR-10 object images and 81.8% for NIH chest X-ray images. This work lays a foundation for flexible neuromorphic hardware systems capable of efficient, high-performance spatiotemporal signal processing.

Dynamic scale position embedding for cross-modal representation learning.

Spatiotemporal patterns and driving factors of dissolved oxygen dynamics in a highly turbid subtropical macrotidal river estuary.

Global review of jellyfish monitoring programs to design fit-for-purpose approaches.

How does the reference period influence meteorological drought analysis and monitoring? A case study in Northeast Brazil using a century of data.

Multi-parameter prediction of seawater quality based on dynamic spatio-temporal relationship network.

Aerosols influence forest ecosystems through changes in radiation and climate affecting plant physiology and structure. Conversely, forests also contribute to aerosol formation. They emit primary aerosol particles and volatile organic compounds, which promote secondary organic aerosol formation in the atmosphere. This forest-aerosol coupling is highly dynamic, influenced by temperature, radiation, humidity, and trace gases. Wildfires add further complexity via smoke plumes altering radiation and ecosystem functioning, tropospheric ozone levels and stratospheric chemistry. Aerosols modify the quantity, directionality, and composition of solar radiation. The type of diffuse light produced by aerosol particles is however strongly dissimilar to the one produced under clouds, and the relevance of the traditional diffuse/direct binary paradigm is discussed. Therefore, potential benefits from increased diffuse radiation are contingent on aerosol load, canopy structure, and prevailing environmental conditions. Beyond photosynthetic responses, aerosols alter forest water-use efficiency and microclimate, yet their long-term effects on plant development, architecture, and community composition remain poorly understood. This review highlights significant knowledge gaps and recent advances in understanding aerosol-forest interactions across temporal and spatial scales. We underline the urgent need for improved experiments with realistic diffuse shading, extensive in situ observations, mechanistic model intercomparison, and global validation to guide future research and policy.

Climate change and land-use/land-cover (LULC) dynamics jointly reshape watershed hydrology and water quality, yet their relative contributions remain difficult to isolate across regions, indicators, and methods. This systematic review synthesizes 28 peer-reviewed studies (2000–2025) that explicitly attribute or partition climate and LULC effects on streamflow, water yield, evapotranspiration, baseflow, and multiple water-quality indicators (e.g., nutrients, sediments, dissolved organic matter, salinity/alkalinity, and contaminant mixtures). Studies were grouped into four synthesis themes: (i) conceptualizations and study designs, (ii) process-based and hybrid modeling frameworks, (iii) statistical and decomposition approaches, and (iv) cross-context patterns and water-quality attribution. Across the evidence base, attribution outcomes are strongly conditioned by methodological choices—especially baseline definition, construction of climate-only and LULC-only counterfactuals, spatial and temporal scale, and the metric used to express contributions (e.g., scenario contrasts, sensitivities, or variance explained). Long-term water-balance responses are often attributed primarily to climate forcing, while water-quality outcomes are more frequently attributed to LULC and direct anthropogenic pressures, with climate acting as a key modulator of transport pathways and exposure. We conclude that robust climate–LULC attribution requires explicit counterfactual design, integrated use of process-based and data-driven frameworks, explicit representation of interactions, and routine uncertainty analysis to support context-sensitive watershed management and climate adaptation.

Microorganisms drive essential ecosystem functions by mediating carbon, nitrogen, sulfur, and phosphorus transformations that regulate productivity and shape climate feedbacks. Rapid methodological advances now allow precise linkage of microbial identity, in situ activity, and ecosystem processes across spatial and temporal scales. High-resolution approaches-including long-read metagenomics and Hi-C-generate near-complete metagenome-assembled genomes (MAGs) from diverse environments, enabling reconstruction of microbial and viral-host interaction networks. Activity-resolved tools such as quantitative stable isotope probing (qSIP) and bioorthogonal non-canonical amino acid tagging (BONCAT), combined with fluorescence-activated cell sorting (FACS), yield taxon-specific growth and substrate assimilation rates within hours. Single-cell isotope techniques, including Raman-SIP and nanoSIMS, deliver nanometer-scale metabolic insights. Spatial meta-omics platforms, such as MetaFISH and MALDI-MSI, map metabolites alongside microbial identities with micrometer-level precision. Meanwhile, autonomous sequencing systems, including environmental sample processors and nanopore adaptive sampling, enable real-time (<24h) ecological surveillance. Integrating these multimodal datasets into trait-based frameworks has reduced uncertainty in carbon flux predictions by nearly 20%. This review synthesizes these innovations, outlines optimized analytical pipelines, and proposes a framework for embedding eco-omics into predictive ecosystem and climate models, supporting evidence-driven management aligned with Climate Action and Life on Land.

Abstract Cloud microphysics—the collection of processes that govern the small‐scale formation, evolution, and interactions of liquid droplets and ice crystals in clouds and precipitation—remains a major source of uncertainty in weather and climate models. Although too small in scale to be explicitly resolved in any large‐eddy simulation, weather, or climate model, the representation of cloud microphysical processes has significant impact at the climate scale. Current microphysical schemes are limited by both parametric uncertainty, linked to uncertainty in physical parameter values, and structural uncertainty, arising from incomplete physical understanding of the processes at play or approximations made for computational efficiency. Recent advances in the application of machine learning (ML) to the physical sciences show significant potential for minimizing these limitations by leveraging high‐fidelity simulations and observations. Here we outline the challenges that must be addressed to apply ML toward cloud microphysics scheme development. This perspectives paper synthesizes recent progress in using data‐driven methods, including ML, to improve cloud microphysics parameterizations and highlights opportunities to address key uncertainties. We discuss the roles of aleatoric (irreducible, or statistical) and epistemic (reducible, or systematic) errors in contributing to microphysics parameterization uncertainty. ML can leverage observations to improve microphysical schemes via bottom‐up and top‐down constraints. Methods such as differentiable programming and ML‐enhanced sampling strategies and the creation of large scale benchmark data sets promise to bridge the gap between observations and models and to improve the consistency of cloud microphysical representation across temporal and spatial scales.

Abstract We perform a comprehensive superposed epoch analysis of more than 200 corotating interaction regions (CIRs) using WIND spacecraft observations at 1 au. The stream interfaces are identified by minimum variance analysis, and turbulence properties are evaluated using wavelet transforms over a wide range of temporal scales. The analysis of normalized cross helicity ( σ c ) and normalized residual energy ( σ r ) reveals distinct turbulence behaviors across frequencies. The spectral indices of both magnetic and velocity fluctuations smoothly transition from steeper in the slow wind to shallower in the fast wind, while a localized steepening of the velocity spectra near the interface indicates enhanced dissipation due to compression. Across broad frequency bands, σ c shows a clear dip at the stream interface—signifying increased inward Alfvén wave energy—whereas σ r displays a peak–valley–peak structure mainly driven by large-scale velocity shear. In lower-frequency ranges, velocity shear artificially enhances velocity fluctuation energy, producing strong peaks in σ r , while higher-frequency ranges show a smooth increase of σ r from slow wind to fast wind. Nearly half of the analyzed CIRs are accompanied by a heliospheric current sheet (HCS), with many HCSs closely aligned with the stream interface, suggesting an intrinsic link between the two structures. Our findings offer valuable clues for reconciling discrepancies among earlier observational and simulation studies, and provide new insight into how compression and velocity shear modulate solar wind turbulence near CIRs.

Terrestrial ecosystems regulate climate by absorbing about one-third of anthropogenic CO2 emissions. Monitoring carbon, water, and energy fluxes is essential for understanding ecosystem responses to climate change. However, existing flux datasets lack sufficient spatial resolution and consistency needed for fragmented landscapes like UK agricultural areas. This study presents the Unified FLUXes (UFLUX) ensemble, a globally consistent dataset of gross primary productivity, evapotranspiration, and sensible heat fluxes derived from eddy covariance data, satellite observations, and machine learning. UFLUX comprises ~ 60 ensemble members across multiple spatial and temporal scales: global (monthly, 0.25°), Europe (daily, 0.25°; biannual, 100 m), and UK (daily, 100 m). Validation against eddy covariance (EC) measurements shows UFLUX captures over 80% of flux variability, with low mean absolute errors, reproducing climate responses and interannual patterns in line with existing literature, though uncertainties in net carbon flux remain. UFLUX holds promise for supporting cross-scale climate policymaking and actions, providing valuable insights for land management and carbon sequestration efforts aimed at a carbon-neutral future.

Accurate load forecasting is essential for energy-efficient scheduling in cold storage facilities, where cooling demand is shaped by strong periodicity, nonlinear temporal dynamics, and irregular operational disturbances. Traditional statistical and machine-learning models struggle with these multi-scale variations, and existing deep learning approaches often rely on fixed receptive fields or fail to extract adaptive periodic structures. This study introduces MA-CFAN, a multi-scale and adaptive multi-period forecasting framework that integrates temporal decomposition, dynamic frequency-period identification, and a newly designed Compression-Fusion Attention Block (CFABlock) for cross-period representation learning. The architecture leverages FFT-derived adaptive periods to capture seasonal-trend components and employs compression-fusion attention to enhance feature discrimination across temporal scales. Furthermore, this work provides the first systematic evaluation of state-of-the-art forecasting models, including Informer, Autoformer, iTransformer, TimesNet, DLinear, and TimeMixer, to the domain of cold storage load prediction. Experiments on real operational data from a logistics center in Jinan, China, demonstrate that MA-CFAN consistently outperforms all baselines, reducing average MSE and MAE by up to 19.3% and 14.8%, respectively.

We used the Monin–Obukhov similarity theory (MOST) flux-variance relationship to estimate greenhouse gas (GHG) fluxes from high-precision mole fraction measurements at 3 instrumented urban communication towers over several years, demonstrating the ability of this method to detect and quantify changes in emissions. Depending on data availability, we used carbon dioxide (CO2) and carbon monoxide (CO) measurements and/or tracer ratios to estimate fluxes at 1 urban site (Site 3) and 1 suburban site (Site 7) in Indianapolis, IN, USA, and 1 urban site (COM) in Los Angeles, CA, USA. We also compared the estimated fluxes of CO2 from fossil fuel sources (CO2ff) at Sites 3 and 7 and the total CO2 fluxes at Site 3 to 20 m, hourly resolution subdomains of the high-resolution CO2 emissions inventory, Hestia, for the year 2020, introducing a new way to evaluate emissions inventories at small spatial and temporal scales. Using the flux-variance relationship, we detected and quantified abrupt decreases in CO and CO2 fluxes at Site 3 and COM in April 2020, coinciding with the stay-at-home order due to COVID-19 pandemic, as well as abrupt decreases in CO and CO2 fluxes at Site 3 in July 2018 coinciding with a highway closure next to the site. The Hestia emissions inventory detected a decrease in emissions in April 2020 at Sites 3 and 7, but this decrease differed in magnitude from those detected in the atmospheric estimates. Seasonal trends in emissions are similar between Hestia and the atmospheric estimates at Site 7. We use differences in seasonal and spatial trends between the flux estimation methods to identify potential sources of uncertainty in both the atmospheric and inventory methods. The results from this study show that the flux-variance estimation method is a useful tool to monitor local-scale emissions and evaluate high-resolution emissions inventories.

Neural signals such as EEG, ECoG, and intracortical recordings offer a valuable window into brain dynamics but remain difficult to decode due to high dimensionality, nonstationarity, and substantial interindividual variability. Traditional machine learning and deep learning models often show limited generalizability and insufficient interpretability in these settings. Foundation models (FMs)-large-scale architectures pretrained on diverse datasets-have recently emerged as a promising paradigm for building robust, transferable, and physiologically grounded neural representations. Among these modalities, EEG currently serves as the most practical and representative platform for FM development due to its large-scale open datasets, standardized protocols, and broad clinical applicability, while the same conceptual framework remains generalizable to other neural recording types. This review synthesizes emerging FM approaches for neural decoding and critically examines representative EEG-based architectures. We highlight three essential design principles: physiology-aware representation learning that captures oscillatory and dynamic structure, structure-aware architectures that incorporate spatial and anatomical priors, and interpretability mechanisms that ensure neuroscientific and clinical validity. Although models such as the Patched Brain Transformer, CBraMod, and BrainGPT demonstrate encouraging adaptability, many still inherit objectives from non-neural domains and underutilize spatial priors such as electrode topology or functional connectivity. While this review focuses on EEG as the most data-rich and scalable testbed, the same framework can extend to ECoG and intracortical recordings to support unified neural representations across spatial and temporal scales. Fully realizing the potential of neural FMs will require biologically informed objectives, structure-aware architectures, interpretable representations, and standardized data ecosystems.

Rechargeable batteries are increasingly constrained by performance and safety limitations originating from anode degradation, including lithium dendrite-induced short circuits, bulk phase transformations, and interfacial SEI evolution. These processes are highly dynamic, strongly coupled, and span multiple spatial and temporal scales, complicating the establishment of direct structure-property correlations and impeding the rational design of next-generation anodes. Transmission electron microscopy (TEM), particularly in situ modalities, overcomes these limitations by combining sub-ångström resolution with multimodal capabilities-including real-space imaging, diffraction, and spectroscopy-to directly visualize morphology, structural, and chemical in real time. This review summarizes mechanistic insights obtained through analytical TEM across three hierarchical domains: (1) electrode surfaces, revealing deposition/stripping kinetics and dendrite nucleation; (2) bulk phases, tracking phase transformations, volumetric changes, and ion transport; and (3) electrode-electrolyte interfaces, elucidating SEI formation, evolution. Methodological challenges, such as electron-beam-induced artifacts, lithium's low atomic number, and liquid-phase confinement, are critically assessed, alongside emerging solutions including cryogenic TEM, low-dose imaging, and advanced liquid-cell designs. By integrating materials science with state-of-the-art electron microscopy, this review establishes a framework for leveraging in situ TEM to accelerate the development of durable, high-performance, and safe battery anodes.

ABSTRACT Coarse-grained (CG) molecular dynamics simulation has become an increasingly powerful approach for the rational design of drug delivery systems, providing access to spatial and temporal scales beyond the reach of atomistic simulations. This review synthesises recent progress in applying CG methodologies to diverse delivery platforms, including micelles, liposomes, dendrimers, polymeric nanoparticles, hydrogels, metal-organic frameworks (MOFs), and amorphous solid dispersions (ASDs). We examine how CG simulations elucidate self-assembly mechanisms, nanostructural organisation, drug encapsulation and release behaviour, and carrier stability under physiological conditions. Particular attention is given to the Martini and related CG force fields for predicting key descriptors such as aggregation number, solvent-accessible surface area (SASA), release kinetics, and morphological transitions. By integrating computational insights with experimental evidence, CG approaches have shown remarkable potential to accelerate formulation development, optimise carrier performance, and enable predictive screening of drug delivery materials. Collectively, this review underscores the translational relevance of CG simulations and their expanding role in advancing pharmaceutical nanotechnology.

Abstract The wave number (WN) structures of temperature from TIMED/SABER and electron density from COSMIC‐2 GIS data are extracted for the period 2020–2021 within ±45° latitudes. For the first time, a new version of the Climatological Tidal Model of the Thermosphere (CTMT.v2) is used to analyze the vertical‐temporal‐latitudinal tidal structures of temperature and density. CTMT.v2 uses solar flux dependent Hough Mode Extensions (HMEs), includes a more extensive collection of TIMED Doppler Interferometer (TIDI) data, compiles SABER V2.08, updates ion drag and dissipation, and provides tidal components for individual years. The vertical profile of CTMT.v2 tides from below are examined to investigate evolutions, variations, coupling, and impacts on structures of the atmospheric and ionospheric variables in the ionosphere, thermosphere, and mesosphere (ITM) regions. The main results are summarized as follows: (a) the antisymmetric structures of the WN1 move closer to the equator when the equatorial westward‐propagating semidiurnal components are strong above 140 km, (b) the antisymmetric component of the WN2 structure in the northern hemisphere is stronger than that in the southern hemisphere at ionospheric heights (above 105 km), (c) the WN3 structure shows intermittent equatorial symmetric structures at 105 km, but cannot form a clear hemispheric antisymmetric structure at other altitudes, (d) the stronger the WN4 structures in the E‐region, the well‐separated the crests of equatorial ionization anomaly (EIA) in the F‐region. This study highlights the need for more space‐based observations in the ∼100–400 km region and the development of models to advance the understanding of interconnections between terrestrial and space weather processes across different spatial and temporal scales.

Abstract Turbulence is a ubiquitous process that transfers energy across many spatial and temporal scales, thereby influencing particle transport and heating. Recent progress has improved our understanding of the anisotropy of turbulence with respect to the mean magnetic field; however, its exact form and implications for magnetic topology and energy transfer remain unclear. In this study, we investigate the nature of magnetic anisotropy in compressible magnetohydrodynamic turbulence within low- β solar wind using measurements from the Cluster spacecraft. By decomposing small-amplitude fluctuations into Alfvén and compressible modes, we reveal that magnetic anisotropy is largely mode dependent: Alfvénic fluctuations are broadly distributed in propagation angle, whereas compressible fluctuations are concentrated near the quasi-parallel (slab) direction, a feature closely linked to collisionless damping of compressible modes. For β → 0, compressible modes become dominant within the slab component at smaller scales. These findings advance our understanding of magnetic anisotropy in solar wind turbulence and offer a new perspective on the three-dimensional turbulence cascade, with broad implications for particle transport, acceleration, and magnetic reconnection.

Cancer dormancy refers to an asymptomatic stage in cancer progression that contains residual disease. Cancer cells can disseminate from early tumors even before they are detectable, from advanced tumors, and from other metastases. Thus, cancer dormancy is a collective phenomenon, composed of single dormant cells that stopped dividing, tumor mass dormancy where cell proliferation is balanced by cell death, and active micrometastases. Dormancy evolves with complex spatiotemporal dynamics across length scales (from cell-intrinsic to cell-extrinsic interactions and microenvironmental regulation up to the body-wide systemic level) and across timescales (from single dormant cells to dormant tumor masses and active micrometastases), each responding differently to fluctuating microenvironments. Here we review biological in vivo and clinical observations of breast cancer dormancy across scales in length and time. Next, we outline 3D bioengineered models in which these different spatial and temporal scales are considered. Finally, we discuss challenges and opportunities of incorporating patient-derived cells. Collective cell behavior is an important aspect in cancer progression and, as such, modeling dormancy across scales in length and time could open new avenues to help us understand and predict the transition to active metastatic growth.