Aaron W. Zeeb, John T. Allen, Matthew Van Den Broeke
Left-moving (LM) supercells, characterized by anticyclonically rotating updrafts in the Northern Hemisphere, are significant due to their propensity to produce large hail. Although less common than right-moving supercells, they present notable forecasting challenges and societal impacts. However, despite these impacts, the environments of LM supercells are poorly understood compared to their right-moving counterparts. To address this gap, this research focuses on enhancing the understanding of LM supercells by examining the environmental conditions conducive to their development. A manually compiled and quality-controlled dataset of over 850 LM supercell cases across North America is used to provide a robust sample. Near-storm environments are characterized through the use of RAP/RUC inflow proximity sounding profiles. Leveraging storm properties, including mesoanticyclone strength, hail size, wind speed, and duration, we investigate whether environments can differentiate between these varying strengths and categories, thereby enhancing forecaster awareness. Results show that LMs typically form in environments supportive of right movers, with a key difference being that LMs likely only realize the shape of the hodograph above their LCLs. Lapse rates, CAPE, and LCL height are the best predictors of LM strength and hail potential. LMs with wind reports have drier boundary layer moisture, steeper 0--3 km lapse rates, larger CAPE, and higher LCL heights, leading to increased evaporational cooling. Longer-lived LMs often have weaker CAPE and stronger shear as compared to shorter-lived LMs. These results establish a unique parameter space climatology of LM supercells, thus providing essential forecasting insight and reducing the research gap for these storms.
Giovanni De Cillis, Alberto Carrassi, Julien Brajard
et al.
This study explores a physics-data driven hybrid approach for sea-ice column physics models, in which a machine learning (ML) component acts as a state-dependent parameterization of forecast errors. We examine how perturbations in snow thermodynamics and sea-ice radiative properties affect forecast errors, and train dedicated neural networks (NNs) for each model configuration. The performance of the hybrid models is evaluated for long lead-time forecasts and compared against a benchmark system based on climatological forecast-error estimates. The NN-based hybrids prove to be stable, robust to initial condition and atmospheric forcing errors, and consistently outperform their climatology-based counterpart. To derive guiding principles for efficiently handling possible physical model updates, we perform transfer learning experiments to test whether pretrained NNs optimized for one model configuration can be successfully adapted to another. Results indicate that direct evaluation of pretrained networks on the target task provides useful insights into their adaptability, recommending transfer learning whenever performance exceeds a trivial baseline. Finally, a feature-importance analysis shows that atmospheric forcing inputs have negligible influence on NN predictive skill, while ice-layer enthalpies play a key role in achieving satisfactory performance.
Jeffrey van der Voort, Martin Verlaan, Hanne Kekkonen
Currently, more and more machine learning (ML) surrogates are being developed for computationally expensive physical models. In this work we investigate the use of a Multi-Fidelity Ensemble Kalman Filter (MF-EnKF) in which the low-fidelity model is such a machine learning surrogate model, instead of a traditional low-resolution or reduced-order model. The idea behind this is to use an ensemble of a few expensive full model runs, together with an ensemble of many cheap but less accurate ML model runs. In this way we hope to reach increased accuracy within the same computational budget. We investigate the performance by testing the approach on two common test problems, namely the Lorenz-2005 model and the Quasi-Geostrophic model. By keeping the original physical model in place, we obtain a higher accuracy than when we completely replace it by the ML model. Furthermore, the MF-EnKF reaches improved accuracy within the same computational budget. The ML surrogate has similar or improved accuracy compared to the low-resolution one, but it can provide a larger speed-up. Our method contributes to increasing the effective ensemble size in the EnKF, which improves the estimation of the initial condition and hence accuracy of the predictions in fields such as meteorology and oceanography.
Panagiotis Theodoropoulos, Augustinos D. Saravanos, Evangelos A. Theodorou
et al.
Understanding complex systems by inferring trajectories from sparse sample snapshots is a fundamental challenge in a wide range of domains, e.g., single-cell biology, meteorology, and economics. Despite advancements in Bridge and Flow matching frameworks, current methodologies rely on pairwise interpolation between adjacent snapshots. This hinders their ability to capture long-range temporal dependencies and potentially affects the coherence of the inferred trajectories. To address these issues, we introduce \textbf{Momentum Multi-Marginal Schrödinger Bridge Matching (3MSBM)}, a novel matching framework that learns smooth measure-valued splines for stochastic systems that satisfy multiple positional constraints. This is achieved by lifting the dynamics to phase space and generalizing stochastic bridges to be conditioned on several points, forming a multi-marginal conditional stochastic optimal control problem. The underlying dynamics are then learned by minimizing a variational objective, having fixed the path induced by the multi-marginal conditional bridge. As a matching approach, 3MSBM learns transport maps that preserve intermediate marginals throughout training, significantly improving convergence and scalability. Extensive experimentation in a series of real-world applications validates the superior performance of 3MSBM compared to existing methods in capturing complex dynamics with temporal dependencies, opening new avenues for training matching frameworks in multi-marginal settings.
Time series forecasting plays a significant role in finance, energy, meteorology, and IoT applications. Recent studies have leveraged the generalization capabilities of large language models (LLMs) to adapt to time series forecasting, achieving promising performance. However, existing studies focus on token-level modal alignment, instead of bridging the intrinsic modality gap between linguistic knowledge structures and time series data patterns, greatly limiting the semantic representation. To address this issue, we propose a novel Semantic-Enhanced LLM (SE-LLM) that explores the inherent periodicity and anomalous characteristics of time series to embed into the semantic space to enhance the token embedding. This process enhances the interpretability of tokens for LLMs, thereby activating the potential of LLMs for temporal sequence analysis. Moreover, existing Transformer-based LLMs excel at capturing long-range dependencies but are weak at modeling short-term anomalies in time-series data. Hence, we propose a plugin module embedded within self-attention that models long-term and short-term dependencies to effectively adapt LLMs to time-series analysis. Our approach freezes the LLM and reduces the sequence dimensionality of tokens, greatly reducing computational consumption. Experiments demonstrate the superiority performance of our SE-LLM against the state-of-the-art (SOTA) methods.
In a growing renewable based energy system, accurate and reliable wind power forecasts are crucial for grid stability, balancing supply and demand and market risk management. Even though short-term weather forecasts have been thoroughly used to provide up to 3 days ahead renewable power predictions, forecasts involving prediction horizons longer than a week still need investigations. Despite the recent progress in subseasonal-to-seasonal weather probabilistic forecasting, their use for wind power prediction usually involves both temporal and spatial aggregation to achieve reasonable skill. In this study, we present a lead time and numerical weather model agnostic forecasting pipeline which enables to transform ECMWF subseasonal-to-seasonal weather forecasts into wind power forecasts for France for lead times ranging from 1 day to 46 days at daily resolution. By leveraging a post-processing step of the resulting power ensembles we show that these forecasts improve the climatological baseline by 15% to 5% for the Continuous Ranked Probability Score and 20% to 5% for ensemble Mean Squared Error up to 16 days in advance, before converging towards the climatological skill. This improvement in skill is jointly obtained with near perfect calibration of the forecasts for every lead time. The results suggest that electricity market players could benefit from the extended forecast range up to two weeks to improve their decision making on renewable supply
Aditya Pratap Singh, Shrey Shah, Ramanakumar Sankar
et al.
Insights into Jupiter's atmospheric dynamics are vital for understanding planetary meteorology and exoplanetary gas giant atmospheres. To study these dynamics, we require high-resolution, photometrically calibrated observations. Over the last 9 years, the Juno spacecraft's optical camera, JunoCam, has generated a unique dataset with high spatial resolution, wide coverage during perijove passes, and a long baseline. However, JunoCam lacks absolute photometric calibration, hindering quantitative analysis of the Jovian atmosphere. Using observations from the Hubble Space Telescope (HST) as a proxy for a calibrated sensor, we present a novel method for performing unpaired image-to-image translation (I2I) between JunoCam and HST, focusing on addressing the resolution discrepancy between the two sensors. Our structure-preserving I2I method, SP-I2I, incorporates explicit frequency-space constraints designed to preserve high-frequency features ensuring the retention of fine, small-scale spatial structures - essential for studying Jupiter's atmosphere. We demonstrate that state-of-the-art unpaired image-to-image translation methods are inadequate to address this problem, and, importantly, we show the broader impact of our proposed solution on relevant remote sensing data for the pansharpening task.
Studying the response of a climate system to perturbations has practical significance. Standard methods in computing the trajectory-wise deviation caused by perturbations may suffer from the chaotic nature that makes the model error dominate the true response after a short lead time. Statistical response, which computes the return described by the statistics, provides a systematic way of reaching robust outcomes with an appropriate quantification of the uncertainty and extreme events. In this paper, information theory is applied to compute the statistical response and find the most sensitive perturbation direction of different El Niño-Southern Oscillation (ENSO) events to initial value and model parameter perturbations. Depending on the initial phase and the time horizon, different state variables contribute to the most sensitive perturbation direction. While initial perturbations in sea surface temperature (SST) and thermocline depth usually lead to the most significant response of SST at short- and long-range, respectively, initial adjustment of the zonal advection can be crucial to trigger strong statistical responses at medium-range around 5 to 7 months, especially at the transient phases between El Niño and La Niña. It is also shown that the response in the variance triggered by external random forcing perturbations, such as the wind bursts, often dominates the mean response, making the resulting most sensitive direction very different from the trajectory-wise methods. Finally, despite the strong non-Gaussian climatology distributions, using Gaussian approximations in the information theory is efficient and accurate for computing the statistical response, allowing the method to be applied to sophisticated operational systems.
Jacob Rodríguez-Rivero, David López-García, Fermín Segovia
et al.
The RESISTO project, a pioneer innovation initiative in Europe, endeavors to enhance the resilience of electrical networks against extreme weather events and associated risks. Emphasizing intelligence and flexibility within distribution networks, RESISTO aims to address climatic and physical incidents comprehensively, fostering resilience across planning, response, recovery, and adaptation phases. Leveraging advanced technologies including AI, IoT sensors, and aerial robots, RESISTO integrates prediction, detection, and mitigation strategies to optimize network operation. This article summarizes the main technical aspects of the proposed solutions to meet the aforementioned objectives, including the development of a climate risk detection platform, an IoT-based monitoring and anomaly detection network, and a fleet of intelligent aerial robots. Each contributing to the project's overarching objectives of enhancing network resilience and operational efficiency.
This paper explores integrating Automatic Speech Recognition (ASR) into natural language query systems to improve weather forecasting efficiency for Korean meteorologists. We address challenges in developing ASR systems for the Korean weather domain, specifically specialized vocabulary and Korean linguistic intricacies. To tackle these issues, we constructed an evaluation dataset of spoken queries recorded by native Korean speakers. Using this dataset, we assessed various configurations of a multilingual ASR model family, identifying performance limitations related to domain-specific terminology. We then implemented a simple text-to-speech-based data augmentation method, which improved the recognition of specialized terms while maintaining general-domain performance. Our contributions include creating a domain-specific dataset, comprehensive ASR model evaluations, and an effective augmentation technique. We believe our work provides a foundation for future advancements in ASR for the Korean weather forecasting domain.
Rahul Ghosh, Arvind Renganathan, Zac McEachran
et al.
We present a framework for modeling multi-scale processes, and study its performance in the context of streamflow forecasting in hydrology. Specifically, we propose a novel hierarchical recurrent neural architecture that factorizes the system dynamics at multiple temporal scales and captures their interactions. This framework consists of an inverse and a forward model. The inverse model is used to empirically resolve the system's temporal modes from data (physical model simulations, observed data, or a combination of them from the past), and these states are then used in the forward model to predict streamflow. Experiments on several catchments from the National Weather Service North Central River Forecast Center show that FHNN outperforms standard baselines, including physics-based models and transformer-based approaches. The model demonstrates particular effectiveness in catchments with low runoff ratios and colder climates. We further validate FHNN on the CAMELS (Catchment Attributes and MEteorology for Large-sample Studies), which is a widely used continental-scale hydrology benchmark dataset, confirming consistent performance improvements for 1-7 day streamflow forecasts across diverse hydrological conditions. Additionally, we show that FHNN can maintain accuracy even with limited training data through effective pre-training strategies and training global models.
Yu-Qian Huang, Lian-Sheng Zhang, Ji-Xing Yang
et al.
Growing evidence has linked extreme temperature with neuropsychiatric disorders under climate warming with frequent extreme heat events over the past decades, while cognitive performance in relation to heat exposure remains largely unstudied, particularly in populations at high vulnerability to climate risks (e.g., China). Based on five survey waves of a nationwide dynamic cohort (2011–2020), we analyzed 47,825 cognitive test records from 14,729 respondents aged 45+ years across 126 Chinese cities. Global cognitive performance and its two dimensions (episodic memory and mental status) was measured using standardized questionnaires. Temperature exposure prior to cognitive tests was assessed using both average temperatures and heat days exceeding predefined temperature thresholds. Linear mixed-effects models were utilized to examine the relationship between high temperature exposure and cognitive function. This study revealed consistent evidence for heat-related declines in global cognitive performance and episodic memory across multiple exposure-window analyses, while robust associations were observed solely during prolonged exposure periods (more than 90 d) for mental status. For each 1-°C rise in annual mean temperature within 1 year prior to investigation, cognitive scores declined by 0.058 (95% CI: −0.079, −0.037) points for global performance, 0.033 (95% CI: −0.048, −0.018) points for episodic memory, and 0.025 (95% CI: −0.038, −0.013) points for mental status, respectively. Similar findings were seen in analyses using heat exposure days defined by multiple temperature percentiles, linking per 10-d increase in heat duration to reduced global cognitive scores ranging from −0.142 (95% CI: −0.214, −0.070) to −0.168 (95% CI: −0.254, −0.082). Despite varied evidence by heat exposure metrics and cognitive dimensions, stratified analyses suggested possibly higher susceptibility among females, less-educated, and urban-dwelling residents to heat-related cognitive impairment. These results provided suggestive evidence for the role of exposure to heat in triggering cognitive impairment in middle-aged and older individuals. This finding may be crucial in developing public health strategies for managing climate change risks of neurobehavioral disorders in a healthy aging society.
Meteorology. Climatology, Social sciences (General)
Previous research has shown that people use a benchmark hurricane as part of their preparation and evacuation decision-making process. While hurricanes are a common occurrence along the Gulf Coast, research on personal memories of past storms is lacking. Particularly, how well do people remember the track and geophysical hazards (wind speed, storm surge, and total rainfall) of past storms? The accurate or inaccurate recollection and perception of previous storm details can influence personal responses to future storms, such as the decision to evacuate or take other life-saving actions. Survey responses of residents in Alabama and Mississippi were studied to determine if people were accurately able to recall a notable storm’s name when seeing an image of the storm’s track. Those who were able to identify the storm by its track were also asked if they could remember the storm’s maximum reported rainfall, maximum sustained winds, and storm surge at landfall. Results showed that there were statistically significant differences between the levels of accurate recall for different storms, with Hurricanes Katrina and Michael having the most correct responses. Regardless of the storm, most people struggled to remember geophysical hazards. The results of this study are important as they can inform broadcast meteorologists and emergency managers on forecast elements of the storm to better emphasize in future communication in comparison to the actual values from historical benchmark storms.
Gas cannon is a new type of equipment used for rainfall enhancement operating which comprehensively utilizes the influence of shock waves, sound waves, and catalysts to interfere with and catalyze local weather. At present, the use of gas cannons to conduct artificial weather operations in China is still in the experimental stage. Based on multi-source observations from dual-polarization weather radar, rain gauges and other equipment, the rainfall enhancement effect and the possible physical mechanism are comprehensively analyzed for 81 gas cannon operation cases in Anhui Province from 2021 to 2023. Observations of typical cases show that the effect of rainfall enhancement is better when the gas cannon is operated prior to the onset of rainfall, accompanied by an increase in the horizontal reflectivity factor ZH and the differential reflectivity ZDR, and the decrease in the co-polarization correlation coefficient ρhv. However, the effectiveness is poor when the operation is after the start of rainfall. It is observed that the cloud undergoes significant changes primarily in the sub-zero layer following the use of warm cloud catalyst, and the cloud changes rapidly but effects are short-lived. On the other hand, when a cold cloud catalyst is used, the cloud undergoes obvious changes in both the warm cloud region and the cold cloud region with a greater effecting range and longer duration of effects. This may be attributed to the impact of the cold cloud catalyst on the ice phase microphysical processes within the cloud. The increase in radar velocity spectrum width (SW) during the operation of a gas cannon may be caused by the increase in air vortex. Statistical results of hourly rainfall enhancement show that the number of cases of significant rainfall enhancement from the gas cannon is slightly higher than that of significant rainfall reduction. Among the three different types of operation timing, the rainfall enhancement effect is best for Type 2 (rainfall operation at the beginning). The significance of rainfall enhancement is negatively correlated with the duration of the operation, while the duration of the operation is negatively correlated with the increment of ZDR. Excessive sowing can lead to a reduction in rainfall. The significance of rainfall enhancement is negatively correlated with the amount of rainfall in the affected area prior to the operation. After the beginning of rainfall, the operational effectiveness of the gas cannon is poor. The rainfall enhancement is positively correlated with ZH, as well as with middle and low-level wind speed and wind shear. However, the enhancement of rainfall is negatively correlated with high-level wind speed. The high wind speed in the middle and high levels is not conducive to enhancing the rainfall through gas cannon operation. These results provide physical evidence for the effect of a gas cannon on cloud microphysical structure and rainfall.
As the basic units of soil structure, soil aggregate is essential for maintaining soil stability. Intensified freeze–thaw cycles have deeply affected the size distribution and stability of aggregate under global warming. To date, it is still lacking about the effects of freeze–thaw cycles on aggregate in the permafrost regions of the Qinghai-Tibetan Plateau (QTP). Therefore, we investigated the effects of diurnal and seasonal freeze–thaw processes on soil aggregate. Our results showed that the durations of thawing and freezing periods in the 0–10 cm layer were longer than in the 10–20 cm layer, while the opposite results were observed during completely thawed and frozen periods. Freeze–thaw strength was greater in the 0–10 cm layer than that in the 10–20 cm layer. The diurnal freeze–thaw cycles have no significant effect on the size distribution and stability of aggregate. However, < 0.25 mm fraction dominated wet sieving aggregate with the highest proportion during thawing period, while the < 1 mm fraction reached the highest during completely frozen period in the 10–20 cm layer ( P < 0.05). Likewise, the mean weight diameter and water-stable aggregate were decreased during thawing period compared with the other periods, which were influenced by soil microbial biomass carbon and belowground biomass. Hence, the seasonal freeze–thaw processes destroyed macro-aggregate (> 0.25 mm) and reduced aggregate stability. Our study has scientific guidance for evaluating the effects of freeze–thaw cycles on soil steucture and provides a theoretical basis for further exploration on soil and water conservation in the permafrost regions of the QTP.
In the area of a petrochemical industrial site, ten monitoring stations are established to determine the airborne pollutants that are emitted, which include criteria air pollutants and 54 species of ozone formation precursors of volatile organic compounds (VOCs). The hourly pollutants are increased by human activities, such as traffic flow after 7:00 a.m., and ozone becomes more abundant as solar radiation increases in intensity. Monthly air pollutants are present in low concentrations during the rainy season from May to September and in high concentrations from October to April. Results show that VOC concentrations are low in the summer (average concentration 5.7–5.9 ppb) and more than double in the winter (11–12 ppb), with 52–63% alkanes, 18–24% aromatics, 11–22% alkenes and 4.7–7.1% alkynes. Ethane, toluene, propane, n-butane, ethylene and acetylene are the major VOCs, with an annual average concentration exceeding 0.50 ppb. In 2016–2020, the VOC concentration is decreased from 10.1 to 7.73 ppb, corresponding to the ozone formation potential (OFP) decrease from 84 to 61 μg-O<sub>3</sub> m<sup>−3</sup>, with toluene, m,p-xylene, ethylene and propene being the most abundant species. The primary VOC sources are petrochemical industry sites, fuel combustion, vehicle exhaust emissions and evaporation, solvent application, industrial facilities and emission from farming vegetation.
The atmospheres of rocky exoplanets are close to being characterized by astronomical observations, in part due to the commissioning of the James Webb Space Telescope. These observations compel us to understand exoplanetary atmospheres, in the voyage to find habitable planets. With this aim, we investigate the effect that CO$_2$ partial pressure (pCO$_2$) has on exoplanets' climate variability, by analyzing results from ExoCAM model simulations of the tidally locked TRAPPIST-1e exoplanet, an Earth-like aqua-planet and Earth itself. First, we relate the differences between the planets to their elementary parameters. Then, we compare the sensitivity of the Earth analogue and TRAPPIST-1e's surface temperature and precipitation to pCO$_2$. Our simulations suggest that the climatology and extremes of TRAPPIST-1e's temperature are $\sim$1.5 times more sensitive to pCO$_2$ relative to Earth. The precipitation sensitivity strongly depends on the specific region analyzed. Indeed, the precipitation near mid-latitude and equatorial sub-stellar regions of TRAPPIST-1e is more sensitive to pCO$_2$, and the precipitation sensitivity is $\sim$2 times larger in TRAPPIST-1e. A dynamical systems perspective, which provides information about how the atmosphere evolves in phase-space, provides additional insights. Notably, an increase in pCO$_2$, results in an increase in atmospheric persistence on both planets, and the persistence of TRAPPIST-1e is more sensitive to pCO$_2$ than Earth. We conclude that the climate of TRAPPIST-1e may be more sensitive to pCO$_2$, particularly on its dayside. This study documents a new pathway for understanding the effect that varying planetary parameters have on the climate variability of potentially habitable exoplanets and on Earth.
Israel Leyva-Mayorga, Marc M. Gost, Marco Moretti
et al.
In real-time and high-resolution Earth observation imagery, Low Earth Orbit (LEO) satellites capture images that are subsequently transmitted to ground to create an updated map of an area of interest. Such maps provide valuable information for meteorology or environmental monitoring, but can also be employed in near-real time operation for disaster detection, identification, and management. However, the amount of data generated by these applications can easily exceed the communication capabilities of LEO satellites, leading to congestion and packet dropping. To avoid these problems, the Inter-Satellite Links (ISLs) can be used to distribute the data among the satellites for processing. In this paper, we address an energy minimization problem based on a general satellite mobile edge computing (SMEC) framework for real-time and very-high resolution Earth observation. Our results illustrate that the optimal allocation of data and selection of the compression parameters increase the amount of images that the system can support by a factor of 12 when compared to directly downloading the data. Further, energy savings greater than 11% were observed in a real-life scenario of imaging a volcanic island, while a sensitivity analysis of the image acquisition process demonstrates that potential energy savings can be as high as 92%.