AI weather foundation models now achieve forecast skill comparable to numerical weather prediction at far lower computational cost, yet their predictability for high-impact extremes across dynamical regimes remains uncertain. We evaluate Aurora using an event-based framework spanning tropical cyclones, freezes, heatwaves, atmospheric rivers, and extreme precipitation at lead times from 1 to 21 days. Aurora demonstrates strong short-range (1-7 day) skill across event types, including competitive tropical cyclone track accuracy and high spatial agreement for temperature and moisture extremes. However, a consistent subseasonal failure mode emerges: while large-scale circulation patterns remain moderately skillful at 14-21 day leads, threshold-based extreme intensity collapses as fields regress toward climatology. This divergence indicates that Aurora retains synoptic-scale dynamical structure but loses surface-impact amplitude beyond 7-10 days. The practical predictability horizon for deterministic AI extreme-event forecasting therefore remains constrained by intrinsic atmospheric dynamics.
Groundwater level estimation is a crucial step in water resource management and planning. In this study, various artificial intelligence models were developed to predict groundwater levels using meteorological data and groundwater levels from previous days. Support Vector Machines (SVM), M5 Tree, and Multiple Linear Regression (MLR) models were used to predict groundwater levels, and the model results were compared with each other. Performance metrics, including Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Correlation Coefficient (R) served as comparative benchmarks. Model results showed that M5 Tree model produced the best results
High-dimensional vector autoregressive (VAR) models have numerous applications in fields such as econometrics, biology, climatology, among others. While prior research has mainly focused on linear VAR models, these approaches can be restrictive in practice. To address this, we introduce a high-dimensional non-parametric sparse additive model, providing a more flexible framework. Our method employs basis expansions to construct high-dimensional nonlinear VAR models. We derive convergence rates and model selection consistency for least squared estimators, considering dependence measures of the processes, error moment conditions, sparsity, and basis expansions. Our theory significantly extends prior linear VAR models by incorporating both non-Gaussianity and non-linearity. As a key contribution, we derive sharp Bernstein-type inequalities for tail probabilities in both non-sub-Gaussian linear and nonlinear VAR processes, which match the classical Bernstein inequality for independent random variables. Additionally, we present numerical experiments that support our theoretical findings and demonstrate the advantages of the nonlinear VAR model for a gene expression time series dataset.
Deepty Jain, Smriti Bhatnagar, Vanshika Rathi
et al.
Abstract Air pollution will likely increase as cities continue to intensify urban activities and expand their infrastructure. Delhi is one of the most polluted cities in the world. Yet, there is a lack of understanding of how the built environment (BE) strategies can address local air quality levels for the city. Additionally, the existing studies use the land use regression (LUR) technique, assuming independence between BE variables. We assessed the impact of BE variables measured at various spatial scales on Delhi's air pollutants (PM2.5, PM10, CO, NO2 and O3). This study used the Principal Component Regression (PCR) approach to account for the multicollinearity between BE variables. As per the analysis, PCR provided better estimates for PM10, PM2.5, and CO concentrations. LUR was found better for modelling NO2 and O3. The findings show that as built-up percentage and the metro station density increases the PM10 and CO levels are also likely to increase, while increasing green percentage is likely to result in decreasing pollutant concentrations. We also identify BE variables that affect a particular pollutant. Percentage institutional within 700 m buffer radii affects PM10, distance to CBD affects CO levels, and distance to the bus depot is affects both CO and NO2 levels. The PCR helped measure the joint effect of BE variables on pollutant concentrations in Delhi. Simultaneously modelling multiple air pollutants can help develop a better urban development strategy for addressing air pollution.
The identification of tipping points is essential for prediction of collapses or other sudden changes in complex systems. Applications include studies of ecology, thermodynamics, climatology, and epidemiology. However, detecting early signs of proximity to a tipping is made challenging by complexity and non-linearity. Strongly interacting Rydberg atom gases offer model systems that offer both complexity and non-linearity, including phase transition and critical slowing down. Here, via an external probe we observe prior warning of the proximity of a phase transition of Rydberg thermal gases. This warning signal is manifested as a deviation from linear growth of the variance with increasing probe intensity. We also observed the dynamics of the critical slowing down behavior versus different time scales, and atomic densities, thus providing insights into the study of a Rydberg atom system's critical behavior. Our experiment suggests that the full critical slowing down dynamics of strongly-interacting Rydberg atoms can be probed systematically, thus providing a benchmark with which to identify critical phenomena in quantum many-body systems.
Large-scale streaming data are common in modern machine learning applications and have led to the development of online learning algorithms. Many fields, such as supply chain management, weather and meteorology, energy markets, and finance, have pivoted towards using probabilistic forecasts. This results in the need not only for accurate learning of the expected value but also for learning the conditional heteroskedasticity and conditional moments. Against this backdrop, we present a methodology for online estimation of regularized, linear distributional models. The proposed algorithm is based on a combination of recent developments for the online estimation of LASSO models and the well-known GAMLSS framework. We provide a case study on day-ahead electricity price forecasting, in which we show the competitive performance of the incremental estimation combined with strongly reduced computational effort. Our algorithms are implemented in a computationally efficient Python package ondil.
Dmitry V. Divine, Svetlana Divina, Ole Edvard Bjørge
et al.
Abstract Maritime historical documentary sources of weather and state of sea surface including sea ice can aid in filling a known climate knowledge gap for the Southern Ocean and Antarctica for the first half of the 20th century. This study presents a data set of marine climate, sea ice and icebergs recovered from a collection of logbooks from mainly Norwegian whaling factory ships that operated in the Southern Ocean during 1929–1940. The data set comprises some 8000 weather and 4000 sea ice/open sea records from austral summers of the study period. This paper further discusses the structure and content of most common Norwegian maritime documentary sources of the period along with the practices of logging information relevant for the study, such as time keeping, positioning and making weather observations. An emphasis was made on recovery of notes on sea ice and icebergs and their interpretation in terms of WMO categories of sea ice concentration. Data, including ship‐related metadata from all individual documents are homogenized and structured to a common machine‐readable format that simplifies its ingestion into relevant climate data depositories.
Kartika Pandey, Sumit Kumar Mishra, Bhanu Pratap Singh
et al.
Atmospheric particles can undergo aging as they are transported over long distances and mix with particles from other sources. This can lead to the accumulation of pollutants and the formation of complex aerosol mixtures with diverse chemical and physical properties. To investigate the process of aging in ambient atmosphere, 24h sampling of PM2.5 aerosol particles on Quartz microfiber filter with a tin substrate was carried out from November 2020 to March 2021 at CSIR-National Physical Laboratory, New Delhi (28°38'10″ N and 77°10′17'' E), using fine particle sampler. Based on the observations of weather and meteorological parameters, a few episodic cases have been selected, and samples were analyzed at bulk and individual particle level. The objective of the present study is to investigate the aging characteristics of aerosols, enabling us to understand the mixing of aerosols (at both bulk and individual particle levels) and the variation in fresh and deformed (aged with other species) graphitic content in the episodic cases. The Raman Spectroscopy technique employed measures the intensity of graphitic (G band; around 1580 cm−1) and disordered graphitic (D band; around 1320 cm−1) content of aerosols. Individual particle microscopic observations reveal the occurrence of open chain fractals of black carbon in variable monomer sizes, sometimes agglomerated with metals like Cu, Cr, Ca etc., along with the presence of S- rich and organic aerosols while the Raman Spectrum (bulk sample analysis) highlights graphitic and disordered (when graphite interacts with other chemical species) graphitic intensities. Comparing the intensities of heavy haze and moderate haze with non-haze days (for comparison purpose, March 23, 2021 with the lowest PM2.5 concentration ∼ 62 μg/m3, has been considered as a non-haze day), it was observed that the intensities recorded on haze days were 45 to 200 times higher for the G band and 43 to 93 times higher for the D band; while for moderate haze days, the intensities were 4 to 61 times higher for the G band and 2 to 29 times higher for the D band. These findings suggest chemical processing of BC during haze days.
Sergiy Snizhko, Sergiy Snizhko, Iulii Didovets
et al.
This article focuses on water security in Mykolaiv, a city of 0.5 million inhabitants in southern Ukraine, in the situation of scarcity of usable water resources caused by climate change and military operations. This problem arose after the Dnipro-Mykolaiv water pipeline was destroyed in April 2022 as a result of military operations and the supply of drinking water to the city was cut off. To ensure that the city’s population has constant access to sufficient water of acceptable quality, a search for alternative water sources and a climate risk assessment were carried out for the new municipal water supply system from the Southern Bug River. The possible change in flow and its intra-annual distribution under the influence of climate change was modeled using the WaterGAP2 hydrological model and climate projections under the SSP1-2.6 and SSP5-P8.5 scenarios. It was found that under the SSP1-2.6 scenario, the reduction in river flow will be insignificant (up to a maximum of 14% in the far future) and there will be no restrictions on the city’s water supply from this section of the river in the near (2021-2050) and far (2051-2080) period. The maximum water withdrawal for municipal water supply and the minimum environmental flow will reach their maximum value only in August (56% of the projected flow), which is not critical. Under the SSP5-8.5 scenario, in the long-term perspective of 2051-2080, the largest decrease in runoff will occur from May to October, and the water withdrawal will increase to 40-79% of the projected flow. The use of the research results not only in water management, but also in municipal administration, and their dissemination in territorial communities will contribute to the successful adaptation of socio-economic and environmental processes in the region and can bring successful benefits not only to the economy, but also to communities.
Volatile organic compounds (VOCs) emitted from fugitive sources are crucial for environmental and health risk assessments. However, monitoring these emissions at ground level, according to traditional technical specifications, has made it challenging to identify polluted air masses and collect purposeful samples. In this study, we focused on utilizing an unmanned aerial vehicle system to obtain air samples around a petrochemical industrial park. We conducted a quantitative analysis of 108 VOC species and compared the results between aerial and ground-level samples. The findings indicated a higher presence of reactive compounds in the aerial samples. The sample pairs exhibited relatively homogeneous compositions of hydrocarbons with fewer than eight carbon atoms, suggesting a well-mixed condition for light compounds. Conversely, the aerial samples exclusively exhibited high mixing ratios of C8–C15 compounds, including branched paraffins and aldehydes. Based on the quantified VOCs, we evaluated the ozone formation potential (OFP) and secondary organic aerosol formation potential (SOAP). The results highlighted aldehydes, alkenes, and aromatics, particularly propanal, 2-butene, m/p-xylene, and benzaldehyde, as priority control compounds. Additionally, the semiquantitative concentrations of these non-quantitative C8–C15 species ranged from 1 to 15 ppbv, with a total content exceeding 150 ppbv, it indicated the significant contribution to ambient secondary pollution. These results provide valuable insights into the identification of potential emission sources and the assessment of environmental repercussions attributed to these intermediate-volatile organic compounds from fugitive emissions around petrochemical plant.
A new index for high-impact weather forecasting is introduced and assessed in comparison with the well-established extreme forecast index (EFI). Two other ensemble summary statistics are also included in this comparison study: the shift-of-tail and a standardised ensemble mean anomaly. All these forecasts are based on the same ingredients: the ensemble forecast run at the European Centre for Medium-Range Weather Forecasts and the corresponding model climatology derived from a set of reforecasts. The new index emerges from recent developments in forecast verification of extreme events: it is derived as a consistent forecast with the diagonal score, a weighted version of the continuous ranked probability score targetting high-impact events. In this study, we emphasise the importance of forecast discretisation for communication purposes and decision-making. A forecast is actionable in the situation where a user can decide to take action when a threshold is exceeded by the forecast. Forecast verification is performed to assess both the potential skill of the different indices as well as their specific skill as actionable forecasts. Among the investigated actionable forecasts, the new proposed index demonstrates the strongest discrimination power, in particular at longer lead times, paving the way for seamless predictions of high-impact weather across time ranges.
Regression analysis is a key area of interest in the field of data analysis and machine learning which is devoted to exploring the dependencies between variables, often using vectors. The emergence of high dimensional data in technologies such as neuroimaging, computer vision, climatology and social networks, has brought challenges to traditional data representation methods. Tensors, as high dimensional extensions of vectors, are considered as natural representations of high dimensional data. In this book, the authors provide a systematic study and analysis of tensor-based regression models and their applications in recent years. It groups and illustrates the existing tensor-based regression methods and covers the basics, core ideas, and theoretical characteristics of most tensor-based regression methods. In addition, readers can learn how to use existing tensor-based regression methods to solve specific regression tasks with multiway data, what datasets can be selected, and what software packages are available to start related work as soon as possible. Tensor Regression is the first thorough overview of the fundamentals, motivations, popular algorithms, strategies for efficient implementation, related applications, available datasets, and software resources for tensor-based regression analysis. It is essential reading for all students, researchers and practitioners of working on high dimensional data.
Intense solar activity can lead to an acceleration of solar energetic particles and accordingly increase in the complex radiation field at commercial aviation flight altitudes. We considered here the strongest ever reported event, namely that of 774 AD registered on the basis of cosmogenic-isotope measurements, and computed the ambient dose at aviation altitude(s). Since the spectrum of solar protons during the 774 AD event cannot be directly obtained, as a first step, we derived the spectra of the solar protons during the ground level enhancement (GLE) #5 on 23 February 1956, the strongest event observed by direct measurements, which was subsequently scaled to the size of the 774 AD event and eventually used as input to the corresponding radiation model. The GLE #5 was considered a conservative approach because it revealed the hardest-ever derived energy spectrum. The global map of the ambient dose was computed under realistic data-based reconstruction of the geomagnetic field during the 774 AD epoch, based on paleomagnetic measurements. A realistic approach on the basis of a GLE #45 on 24 October 1989 was also considered, that is by scaling an event with softer spectra and lower particle fluxes compared to the GLE #5. The altitude dependence of the event-integrated dose at altitudes from 30 kft to 50 kft (9.1–15.2 km) was also computed for both scenarios. Our study of the radiation effects during the extreme event of 774 AD gives the necessary basis to be used as a reference to assess the worst-case scenario for a specific threat, that is radiation dose at flight altitudes.
Using the China National Knowledge Infrastructure (CNKI) and Web of Science (WoS) databases, 487 articles that used remote sensing methods to study the intensity of surface urban heat islands (SUHIs) over the past 20 years were obtained using keyword searches. A multidimensional analysis was conducted on these articles from the perspectives of the research methods used, spatiotemporal distribution characteristics of the research area, research development trends, and main challenges. The research found that (1) the growth trend of the various SUHI research methods over the years was similar to the overall trend in the number of publications, which has rapidly increased since 2009. (2) Among the SUHI research methods, temperature dichotomy is the most widely used worldwide; however, defining urban and rural areas is a main challenge. The Gaussian surface and local climate zoning methods have gradually emerged in recent years; however, owing to the limitations of the different urban development levels and scales, these methods require further improvement. (3) There are certain differences in the application of SUHI research methods between China and other countries.
Laura J. Brown, Billie M. Turner, Victoria Cavero
et al.
Introduction: Women, especially those living in low-and-middle-income countries experience increased exposure to and impacts of environmental threats. Peru is especially susceptible, with high levels of pollutants associated with extractive industries, and climatic-related disasters exacerbated by climate change. International policies and movements are increasingly calling for a gendered approach to environmental health. We aimed to understand the current Peruvian research, advocacy, and policy landscape at the environment-gender-health nexus. Methods: We held 18 in-depth semi-structured interviews with key informants from the Peruvian Government, academia, and non-governmental organizations to explore how a gender-sensitive approach and interdisciplinary environmental health collaborations are delivered. We used thematic analysis to compare gender approaches, priorities, and barriers/facilitators to delivering projects within this nexus. Results: We remotely interviewed 6 representatives of each sector between July 2020 and March 2021. Interviewees mentioned the detrimental role of weak institutions, multilevel corruption, and the lack of interdisciplinarity and intersectorality across environmental health programs and research. They described several barriers to successful collaboration across organizations and sectors, including funding scandals related to extractive economies, high staff turnover impairing long-term program implementation, and machismo culture in organizations and communities. Women's empowerment was described as important for successful program delivery, especially in female-led associations. Some interviewees emphasized the invisibilization of vulnerable groups, such as girls, teenagers, pregnant women, victims of gender-based violence, and LGBTQI+ people. Conclusions: These qualitative findings highlight the multiple and inter-related contextual issues faced by environmentally threatened communities in Peru, and how macrostructural barriers contribute to a paucity of sustainable, gender-oriented, environmental health projects.
Public aspects of medicine, Meteorology. Climatology
Debris-covered ice is widespread in mountain regions with debris an important control on surface ice melt and glacier retreat. Quantifying debris cover extent and its evolution through time over large regions remains a challenge. This study develops two Normalized Difference Supraglacial Debris Indices for mapping debris-covered ice based on thermal and near Infrared Landsat 8 bands. They were calibrated with field data. Validation suggests that they have a high level of accuracy. They are then applied to Landsat data for 2016 to produce the first detailed glacier inventory of the Afghanistan Hindu Kush Himalaya that includes debris cover. 3408 glaciers were identified which, for those ⩾0.01 km2 in area, gives an ice cover of 2,222 ± 11 km2 and a debris cover of 619 ± 40 km2. Principal components analysis was used to identify the most influential drivers of debris-covered ice extent. Lower proportions of debris cover were associated with glaciers with a higher elevation range, that were larger, longer and wider. These relations were statistically clearer when the dataset was broken down into climate and geological zones. A glaciers continue to shrink, the proportion of debris cover will become higher, making it more important to map debris cover reliably.
The horizontal drift round-trip sounding observation is a new sounding technology developed by China Meteorological Administration. By releasing one sounding balloon with this technology, three sections of observations can be obtained, including two sections of vertical tropospheric sounding(ascending and descending) with an interval of about 6 hours and a 4-hour horizontal sounding within the stratosphere. This technology effectively makes up for the insufficiency of conventional soundings, improving the time and space resolution of radiosonde data at a lower cost. The detection system adopts Beidou radiosonde, which significantly improves the accuracy of sounding and wind measurements. In addition, the drift section of horizontal drift round-trip sounding observation fills the gap of the stratospheric temperature detection technology in China. Therefore, horizontal drift round-trip sounding data can be used to verify the temperature profile and stratosphere temperature data of meteorological satellite.Fengyun series meteorological satellites are widely used in China, supporting the meteorological forecast in the Eastern Hemisphere. Among Fengyun satellites in use, FY-3D has the longest years of service. To test the accuracy of FY-3D satellite temperature products, an algorithm is designed according to the characteristics of the horizontal drift round-trip sounding data and satellite data, and the temporal and spatial thresholds are calculated. Based on this algorithm, FY-3D satellite retrieved atmospheric temperature data are verified using the horizontal drift round-trip sounding data in the middle and lower reaches of the Yangtze from March to September in 2021. It can be concluded from the inspection results that the temperature data of FY-3D satellite has a high accuracy, with an average absolute deviation of about 1.34℃ from the data of ascending section and 1.9℃ from the data of descending section. Above 100 hPa and below 850 hPa, the temperature errors of satellite data are 0.59℃ and 0.33℃ larger, respectively. The average absolute deviation of the stratosphere is about 3.92℃, which is slightly larger than the ascending section and descending section. Compared with the sounding profile, the satellite temperature profile has lower vertical resolution and smoother trend, so it cannot show more details of atmospheric vertical variation.
Thomas Maullin-Sapey, Armin Schwartzman, Thomas E. Nichols
The analysis of excursion sets in imaging data is essential to a wide range of scientific disciplines such as neuroimaging, climatology and cosmology. Despite growing literature, there is little published concerning the comparison of processes that have been sampled across the same spatial region but which reflect different study conditions. Given a set of asymptotically Gaussian random fields, each corresponding to a sample acquired for a different study condition, this work aims to provide confidence statements about the intersection, or union, of the excursion sets across all fields. Such spatial regions are of natural interest as they directly correspond to the questions "all random fields exceed a predetermined threshold?", or "Where does at least one random field exceed a predetermined threshold?". To assess the degree of spatial variability present, we develop a method that provides, with a desired confidence, subsets and supersets of spatial regions defined by logical conjunctions (i.e. set intersections) or disjunctions (i.e. set unions), without any assumption on the dependence between the different fields. The method is verified by extensive simulations and demonstrated using a task-fMRI dataset to identify brain regions with activation common to four variants of a working memory task.
Studies on droplet transmission are needed to understand the infection mechanism of SARS-CoV-2. This research investigated the effects of coughing intensity and wind direction on respiratory droplets transportation using the Euler–Lagrange method. The results revealed that both coughing intensity and wind conditions considerably influence the transmission of small and medium droplets but had little effect on large droplets. A stronger coughing intensity resulted in small and medium droplets traveling farther in a calm wind and spreading widely and rapidly in a windy environment. The droplets do not travel far in the absence of ambient wind, even with stronger coughing. Medium droplets spread in clusters, and small droplets drifted out of the domain in the band area in different wind conditions except for 60° and 90° wind directions, in which cases, the droplets were blown directly downstream. In 0° wind direction, many droplets were deposited on the human body. The fast and upward movement of particles in 60° and 90° directions could cause infection risk with short exposure. In 180° wind direction, droplets spread widely and traveled slowly because of the reverse flow downstream, prolonged exposure can result in a high risk of infection.