ABSTRACT This article presents a unified ozonesonde dataset, COSM (National Research Council of Italy (CNR) Ozone Sounding Merged), integrating observations from the SHADOZ (Southern Hemisphere ADditional OZonesondes), NDACC (Network for the Detection of Atmospheric Composition Change) and WOUDC (World Ozone and Ultraviolet Radiation Data Centre) archives. The dataset spans the period 1978–2024, with actual observations covering this entire range, and contains approximately 101,500 vertical ozone profiles collected at 124 stations worldwide. The COSM dataset was constructed through a harmonization process and the application of systematic quality control procedures designed to detect and correct issues such as unit inconsistencies, implausible values and incomplete metadata, while accounting for differences in instrumentation, reporting formats and vertical resolution across networks. The COSM dataset provides a structured, quality‐assessed and well‐documented archive of global ozonesonde observations. Compared with the original archives, the COSM dataset improves spatial and temporal coverage, facilitating consistent use in long‐term studies of ozone variability and trends. A representativeness assessment, based on satellite‐derived ozone profiles, confirms that stations with the longest and most continuous records provide meaningful spatial coverage, particularly in tropical and mid‐latitude regions. In addition to ozone, the COSM dataset also includes typical thermodynamic variables such as temperature, pressure and relative humidity. To our knowledge, this is the first unified and openly available collection of ozonesonde records combining SHADOZ, NDACC and WOUDC. The dataset is relevant for a wide range of applications, including climate and chemistry model evaluation, satellite product validation, and data assimilation into reanalysis. By consolidating heterogeneous archives into a single resource, it provides the scientific community with a comprehensive, quality‐assessed and accessible dataset, which is openly distributed via Zenodo.
ABSTRACT The Mesosphere and Lower Thermosphere establish an essential connection between Earth's lower atmosphere and upper atmosphere because wind patterns essential for atmospheric movement also influence energy distribution and chemical movements during geomagnetic storms that disrupt satellite operations and space weather prediction systems. The existing reconstruction methods that use single‐instrument data and physics‐based models and statistical methods show two main flaws because they cannot capture all the spatial details and face varying levels of uncertainty at different altitudes, and they cannot model the complex weather patterns during storm events. The study presents an innovative intelligent generative reconstruction framework which combines TIMED Doppler Interferometer wind data with SABER thermal emission data through a Generative Adversarial Network structure that utilizes an attention‐based spatial–channel feature fusion method which the Adam algorithm helps to improve. The attention module dynamically weights wind–temperature interactions to enhance spatial continuity and altitude‐aware learning, while the Generative Adversarial Network generator reconstructs physically consistent three‐dimensional wind vectors (U, V, W) and the discriminator enforces physical consistency. The research results based on 1390 fused samples, which included 973 training samples, 208 validation samples and 209 testing samples, achieved excellent prediction results. The study achieved R2 values of 0.9889, 0.9891 and 0.9911 for the train, validation and test sets, respectively, while providing normalized RMSE values of 0.2321, 0.2343 and 0.2069 and MAE values of 0.1073, 0.1198 and 0.1111. The R2 value between 0.98 and 0.99 stays intact during altitude‐binned evaluation across the 60–110‐km range, while the storm‐phase analysis shows that geomagnetic disturbances lead to increased RMSE, yet the system maintains its fundamental reconstruction stability. The proposed framework delivers an expandable solution that utilizes data to create ongoing high‐resolution worldwide Mesosphere and Lower Thermosphere wind reconstruction, which enables advanced space weather monitoring and upper atmospheric research.
ABSTRACT Loess is a widely distributed and collapsible soil on China's Loess Plateau, and due to its poor dynamic performance under cyclic loads, it poses significant challenges to the stability of infrastructure. However, systematic datasets linking calcium oxalate crystal‐induced modification, microstructural characteristics, and dynamic mechanical responses remain limited. The dataset includes compacted loess specimens prepared at oxalic acid concentrations of 0, 0.5, 1.5 and 2.0 mol/L and target moisture contents of 14%, 17% and 20%. Dynamic damping‐ratio curves were obtained using a hollow cylindrical torsional shear apparatus, while SEM observations and PCAS analysis were used to quantify pore‐size distribution, pore morphology, pore orientation and related microstructural descriptors. Microstructural characterisation was performed using JSM‐6700F scanning electron microscopy combined with the Pore and Crack Analysis System. The image‐based analysis provided quantitative information on pore‐size distribution, pore orientation, pore morphology and the bonding characteristics between crystals and soil particles. This dataset includes the basic physical properties of loess, the dynamic damping ratio curves versus shear strain for calcium oxalate crystal‐modified loess and untreated loess, scanning electron microscopy images of microstructures before and after dynamic shear, quantitative pore structure parameters derived from Pore and Crack Analysis System analysis, and Pearson correlation matrices linking oxalic acid concentration, microstructural parameters and dynamic damping ratios. This dataset provides reusable data for Earth system science and civil engineering research. It can be used to validate numerical models of soil modification, compare eco‐friendly stabilisation methods for collapsible loess and support further assessment of infrastructure performance in loess regions.
ABSTRACT Spatial characterisation of building morphology and roughness parameters is critical for modelling urban aerodynamic, thermal and radiative exchange processes and their variability within the urban roughness sublayer. New parameters included in v2.0 of the MAPSECC: London database (Multi‐scale harmonisation Across Physical and Socio‐Economic Characteristics of a City region) capture the spatial variability of building roughness and morphology both horizontally across neighbourhoods of the city (500 m resolution), as well as vertically (1 m resolution). The latter allows modelling of height variations of roughness‐layer state variables including radiative exchange. New data features are: building plan and frontal area indices, total impervious fraction, ratio of exposed building envelope area to footprint area, momentum roughness length and displacement height, street‐canyon aspect ratio and vertical profiles of building plan area index, building perimeter length and effective building diameter. The new processing follows the protocol and conventions established with database v1.0 (with > 100 parameters), assuring that consistency is maintained across all building‐related parameters and statistics.
ABSTRACT Flooding is one of the costliest hazards facing Aotearoa New Zealand (ANZ). We present nationally consistent flood maps organised into four datasets corresponding to a range of current and future climate scenarios in Aotearoa New Zealand, along with the methodology used to create them. This was developed through the Endeavour project Mā te haumaru ō te wai: flood resilience for Aotearoa New Zealand (https://niwa.co.nz/hazards/ma‐te‐haumaru‐o‐nga‐puna‐wai‐o‐rakaihautu‐ka‐ora‐mo‐ake‐tonu), funded by the Ministry of Business, Innovation & Employment (MBIE). These flood maps were generated by a semi‐automatic workflow linking a chain of models. For each catchment, rainfall corresponding to a 1% AEP (Annual Exceedance Probability) design storm event was routed downstream using hydrological modelling in the upper catchment and injected into a hydrodynamic model to produce fluvial‐pluvial inundations maps. Tide was imposed on open sea boundaries and levees were enforced. An adaptable grid down to 4 m resolution was used and hazard layers of maximum water depth, maximum water elevation and maximum (depth times velocity) were generated. Flood hazard was first generated for actual climate conditions (or one degree warmer than pre‐industrial global climate conditions). Using relations between global temperature increase and local extreme rainfall intensity, future climate flood maps were generated for scenarios corresponding to an increase of 1°C, 2°C and 3°C compared to actual current climate conditions (i.e., two, three and four degrees warmer than pre‐industrial global temperature). These national maps can be used to reveal actual and future flood hazards for the country, especially in rural areas where flood hazard has never been assessed. They can also illustrate how flood‐prone areas may evolve with climate change. They present a first consistent national flood assessment needed to inform policy. These maps were also used for a first consistent national flood risk assessment and are accessible on the Earth Sciences New Zealand (ESNZ) datahub under an attribution non‐commercial licence.
ABSTRACT In the summer of 2021, observations of balloon‐borne electric field within thunderstorms were conducted in Nanchang City, Jiangxi Province, China. On 24 August, two electric field profiles were obtained during the mature stage of a local thunderstorm in Nanchang City, Jiangxi Province, China. The initial sounding (S1) and subsequent sounding (S2) were initiated with a 6 min interval. However, S1 entered the weak echo region from the leading edge of the thunderstorm and entered the 30 dBZ region at an altitude of 6 km, while the second sounding (S2) entered from the cloud base and entered the strong echo region at approximately 0°C. At altitudes more than 6 km, the observations were synchronized (with a horizontal distance of 3–4 km). The results showed that two sounding balloons traversed a total of 8 and 11 charge regions for S1 and S2, respectively, and the predominant charge regions conformed to the tripole charge structure. The synchronized observations above 6 km indicated that the distribution of charge regions within the thunderstorm indeed exhibits horizontal inhomogeneity. In addition, an anomalous negative charge region was identified in the upper portion of the thunderstorm, with a depth ranging from 2.3 to 2.7 km, which is larger than the normal negative screening layer. It is notable that a pair of charges with negative upper and positive lower charges is observed at 8.5 and 10 km, which is difficult to explain reasonably using non‐inductive charging mechanisms. This indicates the complexity of the formation of charge region within clouds.
ABSTRACT An objective analysis dataset of the Venus atmosphere, ALERA‐V version 1.0, has been released. This objective analysis provides the best estimate of the state of the Venus atmosphere produced by combining observations and a forecast model weighted according to their respective levels of uncertainty. The accumulation of frequent observations by the Venus Climate Orbiter ‘Akatsuki’ has, for the first time, enabled an objective analysis of the Venus atmosphere. ALERA‐V stands for AFES‐LETKF experimental ensemble objective (re)analysis of the Venus atmosphere, and is generated by ALEDAS‐V, the AFES‐LETKF data assimilation system for the Venus atmosphere. ALEDAS‐V uses AFES‐Venus (Atmospheric General Circulation Model for the Earth Simulator for Venus) as a forecast model and the LETKF (Local Ensemble Transform Kalman Filter) for data assimilation. The observations of the zonal and meridional winds obtained by a cloud tracking technique from images taken by Akatsuki's UVI (Ultraviolet Imager) are assimilated to produce ALERA‐V version 1.0. The dataset consists of atmospheric variables defined on a grid of 128 longitude and 64 latitude points, and 60 vertical levels, with output every 6 Earth hours from September to December 2018, including the intensive observation period of November. ALERA‐V is expected to be useful for both scientific and engineering research, such as understanding the dynamical mechanisms of various atmospheric phenomena and planning for future satellite missions, with appropriate consideration of its quality and limitations.
ABSTRACT Burned‐area information is essential for documenting fire activity and supporting environmental management in dynamic ecosystems such as the Pantanal wetland. This study presents AQM‐HLS, an 11‐year burned‐area dataset (2014–2024) derived from multisensor surface‐reflectance imagery from Landsat‐8/9 and Sentinel‐2. The dataset is generated using multitemporal NBR‐based compositing and an automated sampling procedure that integrates VIIRS active‐fire detections with morphological filtering to produce consistent training samples for Random Forest classification. The classification achieved high performance, with overall accuracy above 99.6% and Dice and CSI values exceeding 0.99. Spectral diagnostics indicate a clear separation between burned and unburned surfaces, particularly in the NIR and SWIR regions. The resulting 30‐m resolution annual and monthly maps capture the spatial and temporal patterns of fire occurrence across the Pantanal and provide a consistent time series suitable for climate, ecological and risk‐management applications. The dataset is openly available and can support the intercomparison of burned‐area products with lower spatial resolution, contribute to the development and improvement of algorithms based on medium‐ and low‐resolution sensors, support cross‐sensor calibration across different orbital platforms and inform regional studies on fire dynamics.
ABSTRACT A clean, well‐organized and comprehensive dataset developed in accordance with the FAIR principles (Findable, Accessible, Interoperable and Reusable), provides a solid foundation for a new research initiative and supports open science. In this context, also legacy, deep boreholes data represent a valuable and unrepeatable source of geological and geophysical information to recover. This article presents the digitalization of legacy well documentation originally available only as scanned images, resulting in a dataset of 30 files, derived by an accurate digitalization workflow carried out on exploration boreholes documentation. The latter was collected across the offshore area between Pesaro‐Fano in the Adriatic Sea, a region that has experienced significant seismic activity, including the 9 November 2022 Mw 5.5 Fano‐Pesaro earthquake sequence. Such data were originally gathered as scanned images extracted from vintage raster files (PDF format), publicly available on the Italian ViDEPI Project website (www.videpi.com). Despite their findability and accessibility are straightforward, their interoperability and reusability are severely limited, due to variable image quality, non‐editable contents and obsolete stratigraphic nomenclature. Such wells are named from northwest to southeast as follows: Boheme 01, Tamara 01, Pesaro Mare 03, Pesaro Mare 04, Malachite 01, Cornelia 01 and Elga 01. The digitalization was carried out at high fidelity with respect to the original data, reassessing the relevant well stratigraphy by taking into account all the handwritten comments found inside the original images. Only minor reinterpretations based on lithology, depositional environment and age, were made. Some local formation names were updated, aligned and thus correlated with modern and officially recognized regional stratigraphic units. The digitalization also includes spontaneous potential, resistivity and sonic logs of the Tamara 01 and Boheme 01 wells. Such digitized data were successively stored in open and standard formats (CSV and LAS). Nowadays, these datasets are freely accessible and carry substantial significance as a foundation of earthquake studies, for enhancing geological and geophysical models, depicting the stratigraphic, structural and geophysical characteristics of the study area. Notable potential use includes, among others, seismotectonic studies, gas and CO2 storage, basinal analysis, stratigraphic and paleogeographic investigations within the region. Digitizing such legacy data reduces the risk of data loss, improves modernization and accessibility, supports interoperability and collaboration. It also enables easier quality control and facilitates future reuse, especially when integrated with other data sources such as seismic surveys, production data and reservoir models.
ABSTRACT Detection of terrestrial ecosystems involves the identification and classification of land cover types and vegetation patterns using remote sensing and geospatial analysis. It provides assistance in environmental monitoring, the assessment of biodiversity, and sustainable land management, among others. The study given proposes a hybrid Deep Learning (DL) system that simultaneously incorporates ResNet‐50 and LSTM for accurate classification of Land Use and Land Cover (LULC) in satellite images. The EuroSAT dataset, consisting of 27,000 labelled samples across 10 land‐use classes, was used to train and evaluate the proposed model. This large, diverse dataset enables robust model development for accurate land‐cover classification in satellite imagery. The sequence of activities involves advanced image pre‐processing to enhance edge preservation and reduce noise, such as bilateral, guided, and median filtering. The next instance is feature extraction performed with HOG, which retains important edge and shape properties. The extracted Histogram of Oriented Gradients (HOG) features are given into the ResNet‐50 backbone, where spatial representations at a higher level are derived. Then, those ResNet‐50 based high‐level features are fed into the LSTM layer to model temporal and sequential dependencies and then provide additional contextual information for spatial transitions. The last classification is done using a fully connected layer with a Softmax activation function that provides the class probabilities. The model gained higher performance with a precision of 0.9903, an accuracy of 0.9904, and recall of 0.9903, which showed its ability to differentiate efficiently between land cover types that are more similar. By integrating spatial and sequential modelling, the architecture proposed gives a very high degree of reliability and robustness in classification with regard to a variety of geographic and environmental conditions, rendering them well out for environmental monitoring, urban planning, and geospatial intelligence.
ABSTRACT Nuclear winter refers to the suite of physical and biological consequences that may follow nuclear conflict, particularly the cooling and darkening of Earth's surface due to black carbon soot in the upper atmosphere. While the associated changes in temperature, precipitation, and food system productivity have been the subject of climate modelling for decades, the outputs of models used to project these effects are stored in large files with formats unfamiliar to the broader research community. This paper introduces a standardized, user‐friendly repository of simulated nuclear conflict climate impact data designed to lower barriers for non‐specialist researchers. The data product provides simplified, spreadsheet‐ready datasets derived from established Earth System Model simulations and includes variables relevant to human and environmental impacts: temperature, precipitation, ultraviolet radiation, crop yields, fish catch, and sea ice thickness for a range of nuclear conflict scenarios. This repository aims to facilitate interdisciplinary research into the long‐term consequences of nuclear detonations to support policy development.
ABSTRACT NCAR's Earth Observing Laboratory (EOL) In situ Sensing Facility (ISF) deployed all of its instrumentation in the Sundowner Wind EXperiment (SWEX) in April–May 2022 to characterize downslope winds along the southern slopes of the Santa Ynez Mountains in Santa Barbara (SB) County, California. The goal was to understand boundary layer processes and mechanisms that favour sundowners. ISF provided ground‐based surface and remote sensing observations, sounding systems, operations expertise, data quality control and archiving. This paper describes the measurement system employed and important aspects of the data quality control and management process. The ISF observing system can serve as a prototype to build an integrated network of profiling and flux measuring instruments to better understand the complexity and inhomogeneities within the lower troposphere.
ABSTRACT We present an update of the Irish Hydrometric Reference Network (IHRN) of river gauging stations from across the Republic of Ireland that have been deemed suitable for assessing climate‐driven changes in high, mean and low flows. Selection criteria, analysis of metadata and historical flows, and stakeholder feedback are applied to identify 51 stations for inclusion in the network. Missing daily data were infilled using a conceptual hydrological model and an Artificial Neural Network. As well as providing a dataset for monitoring and detecting the impact of changing climatic conditions on Irish catchments, the updated IHRN offers utility for assessing extremes of flood and drought and for modelling future flow regimes via climate change impact assessments.
ABSTRACT Earthquake (EQ) forecasting is critical for early warning systems. We propose an EQ Magnitude‐Time (EQMT) Forecasting Model that integrates real‐world inter‐fault relations, multiple data sources, and advanced modelling techniques to forecast EQ magnitudes and classify occurrence times. The model converts fault lines into graph representations and extracts node embeddings to capture spatial inter‐fault dependencies. We evaluated EQMT on seismic events in the northern Aegean region, where the North Anatolian Fault intersects the Aegean extensional system. The proposed model attains a mean absolute error of 0.33 for magnitude forecasting and an F1 score of 81.39% for occurrence‐time classification. Furthermore, to enhance the interpretability of both magnitude and occurrence time forecasting tasks, we applied SHapley Additive exPlanations (SHAP)‐based explainable Artificial Intelligence (XAI) analyses, which provide insights into the contribution of individual seismic and fault‐related features to the model's forecast. SHAP‐based explainability analysis reveals that observation‐duration features and graph embeddings are the primary contributors to model predictions. These results demonstrate the potential of fault‐aware, graph‐based approaches for operational EQ forecasting.
ABSTRACT Caves represent challenging systems for monitoring temperature dynamics. While caves are often portrayed as textbook examples of systems with high thermal inertia and minimal annual temperature variability (typically less than 1°C in their deepest sections), the reality is more complex. Site‐specific climatic anomalies, particularly in shallow caves, introduce significant variation. To better understand these localized conditions, long‐term temperature records from multiple caves with diverse characteristics within a single region are needed. This paper presents a dataset comprising air temperature records from 19 (mostly shallow) caves in the Piedmont region (Western Italian Alps), collected in 2012 and from 2019 to 2025. Temperature sensors were deployed at two distinct locations in each cave—one at the entrance and one in internal sectors—capturing thermal signals and buffering effects. This yielded a dataset suitable for both climatological modelling and ecological applications. Calculation of data uncertainty was performed through post‐calibration corrections. The complete dataset is available in figshare (https://doi.org/10.6084/m9.figshare.29108090.v1). This resource is intended to support empirical testing of thermal models, improve the accuracy of climate studies in subterranean ecosystems, and facilitate comparative analyses across cave types. The development of a high‐quality dataset with strong ecological relevance provides a basis for interdisciplinary research at the intersection of climate science, speleology, and subterranean biology. The dataset is particularly suited for investigating ecological thresholds in thermally constrained cave fauna, being interoperable with faunal datasets available from the same area, as well as for evaluating cave microclimates as early indicators of surface climate anomalies.
ABSTRACT Top‐of‐canopy solar‐induced chlorophyll fluorescence (SIF) is strongly affected by canopy structure and illumination conditions, which limits the use of current satellite‐based SIF products for detecting vegetation stress. To address this, we present a new dataset that combines TROPOMI SIF observations with MODIS photosynthetically active radiation (PAR) data (MCD18C2/A2) to derive a fluorescence quantum efficiency (ΦF) product across continental Europe, the UK and Ireland and northern Africa from 2018 to 2025. The dataset has a spatial resolution of 0.05° and daily, decadal and monthly temporal coverage. Building on a multi‐sensor ΦF formulation from the literature, a previous analysis in Germany demonstrated that ΦF is sensitive to agricultural drought, revealing stress‐related signals with a two‐day lag in both agricultural and forest ecosystems, signals that were not detectable using top‐of‐canopy SIF alone or traditional vegetation indices. The dataset includes all input bands, allowing users to adapt or refine the calculation of ΦF. It also provides quality layers, including phase angle, solar zenith angle (SZA), cloud fraction and the retrieval error of the top‐of‐canopy SIF product. This dataset is designed to facilitate further research on the relationship between ΦF and vegetation stress across European biomes, notably in the context of drought detection.
ABSTRACT The low closure temperature of the dating system in low‐temperature thermochronology allows shallow crustal movements to be recorded. In recent decades, this approach has been used extensively in studies of orogenic belt exhumation, sedimentary basin evolution, and topographic and geomorphologic evolution. A substantial portion of the data for East Asia is documented in both Chinese and English literature, so establishing a unified database greatly enhances data reuse, which is crucial for studying regional geological evolution. This paper compiles data from four low‐temperature thermochronology methods—zircon fission track, zircon (U‐Th)/He, apatite fission track and apatite (U‐Th)/He—for East Asia. Based on these data, a comprehensive low‐temperature thermochronology database for East Asia has been established. It includes details of sample source (source literature, lithology, geological age), location (longitude, latitude, elevation), experimental methods, measurement results and related measurement information. Currently, the database contains 8804 sample records and 10,751 dating results from 1010 literature sources. The database is designed to help researchers understand the current research landscape, avoid redundant data collection, and prevent reprocessing of the same samples.
ABSTRACT This product provides MERRA‐2 bias‐corrected global hourly surface total PM2.5 mass concentration with the exact horizontal spatial resolution as MERRA‐2, covering a temporal range from 2000 to 2024. It is derived using a machine learning (ML) approach with a convolutional neural network (CNN) method. It is specifically developed for the NASA Health and Air Quality Applied Sciences Team (HAQAST). The dataset consists of two parameters: MERRA2_CNN_Surface_PM25 and QFLAG. MERRA2_CNN_Surface_PM25, a 3‐dimensional variable (time, latitude, longitude), represents the surface PM2.5 concentrations in μg/m3. QFLAG denotes the quality of data at each grid point, where four indicates the highest quality and 1 indicates the lowest quality. It is recommended to use QFLAG values of 3 and 4 for quantitative analysis.
ABSTRACT Ship logbooks represent a critical source of historical meteorological data, providing valuable observations of barometric pressure, air temperature, sea surface temperature, wind force and direction, and other variables. Substantial quantities of these records are unavailable to climate science as they have not yet been transcribed. We present ‘Weather Rescue at Sea’, a citizen‐science project which transcribed millions of weather observations contained in 19th Century UK Royal Navy ship logbooks. We describe the logbook structure and weather observation‐taking instructions and discuss significant challenges with the translation of handwritten text into accurate data due to errors arising from ambiguous handwriting, historical terminology, and inconsistent metadata. We present the dataset and explore its spatio‐temporal characteristics. The corrected and quality‐assured datasets will enhance climate reanalyses and other historical reconstructions of the pre‐ and early industrial climate by providing more input meteorological data. Furthermore, we highlight emerging tools, such as AI‐driven transcription correction, and outline remaining challenges in fully leveraging these historical records to advance climate science.
ABSTRACT Sea level rise is an inevitable consequence and one of the most significant threats posed by climate change, increasing the risk of flooding in low‐lying areas along the German coast. Based on the IPCC 6th Assessment Report (AR6) projections, we aimed to deliver improved projections of relative sea‐level change for Northern Europe's coastal regions. These projections are available as spatial data up to 2150. While most drivers of sea level change operate on a continental or global scale, vertical land motion is a regional factor—particularly relevant in Northern Europe—resulting from glacial isostatic adjustment and local processes. By combining the IPCC projections of absolute sea level change with a new, high‐resolution vertical land motion model for Fennoscandia, an optimised set of projections for relative sea level change for the North Sea and Baltic Sea was developed. In this context, the spatial resolution changes from the 1 × 1 grid commonly used in AR6 datasets to the finer 1/6 × 1/12 grid of the regional land motion model. This results in local differences in sea level rise by 2100 up to −200 mm to +500 mm within the domain. This dataset represents a contribution to the DAS core service ‘climate and water’, the operational climate service operated by four federal authorities under the umbrella of the German Federal Ministry of Transport.
