Accurately predicting long-term rainfall is challenging. Global climate indices, such as the El Niño-Southern Oscillation, are standard input features for machine learning. However, a significant gap persists regarding local-scale indices capable of improving predictive accuracy in specific regions of Thailand. This paper introduces a novel North-East monsoon climate index calculated from sea surface temperature to reflect the climatology of the boreal winter monsoon. To optimise the calculated areas used for this index, a Deep Q-Network reinforcement learning agent explores and selects the most effective rectangles based on their correlation with seasonal rainfall. Rainfall stations were classified into 12 distinct clusters to distinguish rainfall patterns between southern and upper Thailand. Experimental results show that incorporating the optimised index into Long Short-Term Memory models significantly improves long-term monthly rainfall prediction skill in most cluster areas. This approach effectively reduces the Root Mean Square Error for 12-month-ahead forecasts.
Caron Pablo, Bourdarie Sébastien, Carron Jérôme
et al.
The effects of the space radiation on spacecraft materials and devices are significant design considerations for space missions. In order to meet these challenges, the radiation environment must be understood. Measuring energetic particles in the radiation belts is therefore crucial, and this is why ICARE (Influence sur les Composants Avancés des Radiations de l’Espace) radiation monitors have been developed over several decades. Two ICARE_NG2 (for the second version of the New Generation of the instrument) radiation monitors have recently been launched on HotBird 13F and 13G satellites at the end of 2022, reaching the geostationary orbit in 2023 and providing measurements of electrons of the outer belt. The methods used to derive these electron fluxes are detailed and the results compared with NGRM (Next Generation Radiation Monitor) and specification models. The observed electron dynamics are strongly correlated with solar wind data, and a fully analytical model is developed. This model provides an accurate representation of the measurements, using a limited number of parameters.
Severe Typhoon Danas over the northern part of the South China Sea and seas near Taiwan in early July 2025 had an erratic path that had not been observed before, according to historical data in the region. Its formation, movement, and intensification posed significant challenges to the timely tropical cyclone (TC) warning services. This paper documents the observational aspect and forecasting aspect of this cyclone. There are key findings: (a) when Danas interacted with the Central Mountain Range of Taiwan, a “secondary cyclone” appeared over the northeastern part of Taiwan, which was observed by both weather radars and meteorological satellite winds, and was simulated to a certain extent by a mesoscale numerical weather prediction (NWP) model; (b) data-driven AI global models performed better than physics-based global NWP models in capturing the formation and the rather erratic track of Danas a couple of days earlier, although AI models generally underestimate the intensity forecasts; and (c) an atmosphere–ocean–wave coupled model was found to perform the best in capturing both the track changes of Danas (because of being driven by an AI global model) and its intensity changes (because of better physical representation of the oceanic impact on the intensity of this TC), whereas AI global models, though with various recent enhancements, still tended to underestimate the strength of Danas. This paper serves as a reference of this rather unusual TC for the weather forecasting services in the region.
API calls by large language models (LLMs) offer a cutting-edge approach for data analysis. However, their ability to effectively utilize tools via API calls remains underexplored in knowledge-intensive domains like meteorology. This paper introduces KG2data, a system that integrates knowledge graphs, LLMs, ReAct agents, and tool-use technologies to enable intelligent data acquisition and query handling in the meteorological field. Using a virtual API, we evaluate API call accuracy across three metrics: name recognition failure, hallucination failure, and call correctness. KG2data achieves superior performance (1.43%, 0%, 88.57%) compared to RAG2data (16%, 10%, 72.14%) and chat2data (7.14%, 8.57%, 71.43%). KG2data differs from typical LLM-based systems by addressing their limited access to domain-specific knowledge, which hampers performance on complex or terminology-rich queries. By using a knowledge graph as persistent memory, our system enhances content retrieval, complex query handling, domain-specific reasoning, semantic relationship resolution, and heterogeneous data integration. It also mitigates the high cost of fine-tuning LLMs, making the system more adaptable to evolving domain knowledge and API structures. In summary, KG2data provides a novel solution for intelligent, knowledge-based question answering and data analysis in domains with high knowledge demands.
Amid accelerated digitalization, not only is the scale of data processing and storage increasing, but so too is the associated infrastructure load on the climate. Current climate models and environmental protocols almost entirely overlook the impact of information and communication technologies on the thermal and energy balance of the biosphere. This paper proposes the theory of information and climate feedback (ICF) as a new nonlinear model describing the loop of digitalization, energy consumption, the thermal footprint, the climatic response, and the vulnerability of digital infrastructure. The system is formalized via differential equations with delays and parameters of sensitivity, greenness, and phase stability. A multiscenario numerical analysis, phase reconstructions, and thermal cartography were conducted. Critical regimes, including digital overheating, fluctuational instability, and infrastructural collapse in the absence of adaptive measures, were identified. The paper concludes with the proposal of an international agreement titled the Green Digital Accord and a set of metrics for sustainable digitalization. This work integrates climatology, information technologies, and the political economy of sustainability.
Daniel J. Alford-Lago, Christopher W. Curtis, Alexander T. Ihler
et al.
We present a novel method for forecasting key ionospheric parameters using transformer-based neural networks. The model provides accurate forecasts and uncertainty quantification of the F2-layer peak plasma frequency (foF2), the F2-layer peak density height (hmF2), and total electron content (TEC) for a given geographic location. It includes a number of exogenous variables, including F10.7cm solar flux and disturbance storm time (Dst). We demonstrate how transformers can be trained in a data assimilation-like fashion that uses these exogenous variables along with naive predictions from climatology to generate 24-hour forecasts with nonparametric uncertainty bounds. We call this method the Local Ionospheric Forecast Transformer (LIFT). We demonstrate that the trained model can generalize to new geographic locations and time periods not seen during training, and we compare its performance to that of the International Reference Ionosphere (IRI) using CCIR coefficients.
Guilherme Pumi, Danilo Hiroshi Matsuoka, Taiane Schaedler Prass
et al.
Time series in natural sciences, such as hydrology and climatology, and other environmental applications, often consist of continuous observations constrained to the unit interval (0,1). Traditional Gaussian-based models fail to capture these bounds, requiring more flexible approaches. This paper introduces the Matsuoka Autoregressive Moving Average (MARMA) model, extending the GARMA framework by assuming a Matsuoka-distributed random component taking values in (0,1) and an ARMA-like systematic structure allowing for random time-dependent covariates. Parameter estimation is performed via partial maximum likelihood (PMLE), for which we present the asymptotic theory. It enables statistical inference, including confidence intervals and model selection. To construct prediction intervals, we propose a novel bootstrap-based method that accounts for dependence structure uncertainty. A comprehensive Monte Carlo simulation study assesses the finite sample performance of the proposed methodologies, while an application to forecasting the useful water volume of the Guarapiranga Reservoir in Brazil showcases their practical usefulness.
Abstract In this study, we investigated the translation speed and intensity change characteristics for landfalling North American tropical cyclones (TCs) from 1971 to 2020. We calculated three variables—intensity change, mean translation speed, and translation speed change—prior to each TC landfall and investigated the climatology of these variables for seven coastal segments. We found that lower-latitude segments generally had greater positive intensity changes prior to landfall, and higher-latitude segments had greater translation speeds. Longitude primarily influenced translation speed changes, with landfalling TCs along the Atlantic coast of the United States notably accelerating prior to landfall. Temporal trends in each of these variables were inconsistent geographically, but most segments showed an increase in positive intensity changes over time, demonstrating the increasing likelihood of intensifying TCs before landfall in recent years. We defined extreme intensification and extreme weakening as the 90th and 10th percentile of all landfalling TC intensity changes, respectively. We found that extreme intensification and weakening have been increasing and decreasing in frequency, respectively. Results from this study can be used in a variety of future applications, including in operational forecasting and model production, and provide a baseline for climate attribution studies investigating extreme intensity change events. Significance Statement Landfalling tropical cyclones pose significant hazards to life and property in coastal regions across North America. This study develops a comprehensive climatology of intensity change and translation speed of tropical cyclones during the final 36 h prior to landfall. We found that incidences of extreme intensification and weakening of landfalling North American tropical cyclones have increased and decreased, respectively, since 1971. We also identify lower-latitude coastal segments tend to average faster translation speeds that higher-latitude segments, while segments on the east coast of the United States tended to average a greater acceleration. These results show the importance of looking at tropical cyclone characteristics regionally and provide a useful baseline to assess how tropical cyclone risk is changing in a warming climate.
Abstract Tianshan Mountains are the headwater regions for the central Asia rivers, providing water resources for ecological protection and economic development in semiarid regions. Due to scarce observations, the hydroclimatic characteristics of the Tianshan Mountains Precipitation (TMP) measured over highland (>1500 m) regions remain to be revealed. Here, we show the TMP belongs to a monsoon-like climate regime, with a distinct annual range and a high ratio of summer-to-yearly rainfall, and exhibits six abrupt changes, dividing the annual cycle into six precipitation sub-seasons. Over the past 60 years, the yearly TMP has significantly increased by 17.3%, with a dramatic increase in winter (135.7%). The TMP displays a significant 40-day climatological intra-seasonal oscillation (CISO) in summer. The TMP CISO’s wet phase results from the confrontation of the eastward propagating mid-tropospheric Balkhash Lake Low and the southward migrating Mongolian High. The sudden changes in the two climatological circulation systems trigger TMP’s changes, shaping the 40-day CISO. Emerging scientific issues are also discussed.
The original purpose of resilience design for traditional architectures is to be coordinated with and adapted to the natural environment. The natural environmental resilience of traditional dwellings refers to the ability of the dwellings to maintain the residents’ comfort, safety, and health in the face of natural environmental challenges including various disasters. In the process of designing traditional dwellings, the wisdom contributed to improving living environment conditions is of significant reference value. Therefore, this paper reviews the main literature on the resilience of traditional dwellings to the natural environment in Fujian of China. A topical review framework is proposed to cover the resilience performance of traditional dwellings in Fujian under various natural environmental conditions. Specifically, it is divided into two aspects: internal comfort resilience and external disaster-resistance resilience, based on which researchers in related fields can establish a clearer classification of resilience research in their future studies. In terms of internal comfort resilience, this paper focuses on relevant perspectives such as humidity, temperature, brightness, noise, etc. In terms of external disaster-resistance resilience, this paper summarizes the adaptability of traditional dwellings in the face of disasters triggered by natural hazards such as fires, floods, earthquakes, and typhoons. Based on this framework, this paper reviews the current research status, discusses the limitations and shortcomings in this research area, and proposes corresponding prospects for future research.
Abstract Downbursts are severe wind systems originating from thunderstorm clouds, and their strong horizontal outflows can pose serious hazards to natural and built environments. In the context of the activities of the European project THUNDERR—Detection, simulation, modelling and loading of thunderstorm outflows to design wind‐safer and cost‐efficient structures—a comprehensive database of full‐scale downburst measurements was built. All records were acquired by bi‐ or tri‐axial ultrasonic anemometers installed in the main ports of the High Tyrrhenian Sea, namely Genova, Livorno and La Spezia, within the European projects ‘Wind and Ports’ and ‘Wind, Ports and Sea’. The very limited space and time structure of downburst outflows makes the available records in nature inadequate for developing models that could be used in the atmospheric science and engineering communities. The database described herein represents a step forward in attempting to fill this gap. The downburst nature of all events contained in the dataset was verified through detailed meteorological analyses, including comparisons with radar and satellite images and lightning recordings. The wind speed records associated with the events detected by the anemometric network are made publicly available through the online repository Zenodo and can be reused for multiple purposes. The dataset is expected to convey an important impulse towards the physical characterization and modelling of downburst winds and their codification into design tools for the assessment of wind loading and its effects on structures and infrastructure. Furthermore, it could serve as a promising, essential tool for researchers and risk‐related insurance companies.
In this paper, we study interior estimates for solutions to linearized Monge-Ampère equations in divergence form with drift terms and the right-hand side containing the divergence of a bounded vector field. Equations of this type appear in the study of semigeostrophic equations in meteorology and the solvability of singular Abreu equations in the calculus of variations with a convexity constraint. We prove an interior Harnack inequality and Hölder estimates for solutions to equations of this type in two dimensions, and under an integrability assumption on the Hessian matrix of the Monge-Ampère potential in higher dimensions. Our results extend those of Le (Analysis of Monge-Ampère equations, Graduate Studies in Mathematics, vol.240, American Mathematical Society, 2024) to equations with drift terms.
Air quality estimation can provide air quality for target regions without air quality stations, which is useful for the public. Existing air quality estimation methods divide the study area into disjointed grid regions, and apply 2D convolution to model the spatial dependencies of adjacent grid regions based on the first law of geography, failing to model the spatial dependencies of distant grid regions. To this end, we propose a Dual-view Supergrid-aware Graph Neural Network (DSGNN) for regional air quality estimation, which can model the spatial dependencies of distant grid regions from dual views (i.e., satellite-derived aerosol optical depth (AOD) and meteorology). Specifically, images are utilized to represent the regional data (i.e., AOD data and meteorology data). The dual-view supergrid learning module is introduced to generate supergrids in a parameterized way. Based on the dual-view supergrids, the dual-view implicit correlation encoding module is introduced to learn the correlations between pairwise supergrids. In addition, the dual-view message passing network is introduced to implement the information interaction on the supergrid graphs and images. Extensive experiments on two real-world datasets demonstrate that DSGNN achieves the state-of-the-art performances on the air quality estimation task, outperforming the best baseline by an average of 19.64% in MAE.
Eva-Maria Walz, Peter Knippertz, Andreas H. Fink
et al.
Numerical weather prediction (NWP) models struggle to skillfully predict tropical precipitation occurrence and amount, calling for alternative approaches. For instance, it has been shown that fairly simple, purely data-driven logistic regression models for 24-hour precipitation occurrence outperform both climatological and NWP forecasts for the West African summer monsoon. More complex neural network based approaches, however, remain underdeveloped due to the non-Gaussian character of precipitation. In this study, we develop, apply and evaluate a novel two-stage approach, where we train a U-Net convolutional neural network (CNN) model on gridded rainfall data to obtain a deterministic forecast and then apply the recently developed, nonparametric Easy Uncertainty Quantification (EasyUQ) approach to convert it into a probabilistic forecast. We evaluate CNN+EasyUQ for one-day ahead 24-hour accumulated precipitation forecasts over northern tropical Africa for 2011--2019, with the Integrated Multi-satellitE Retrievals for GPM (IMERG) data serving as ground truth. In the most comprehensive assessment to date we compare CNN+EasyUQ to state-of-the-art physics-based and data-driven approaches such as a monthly probabilistic climatology, raw and postprocessed ensemble forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF), and traditional statistical approaches that use up to 25 predictor variables from IMERG and the ERA5 reanalysis.Generally, statistical approaches perform about en par with post-processed ECMWF ensemble forecasts. The CNN+EasyUQ approach, however, clearly outperforms all competitors for both occurrence and amount. Hybrid methods that merge CNN+EasyUQ and physics-based forecasts show slight further improvement. Thus, the CNN+EasyUQ approach can likely improve operational probabilistic forecasts of rainfall in the tropics, and potentially even beyond.
Athul Rasheeda Satheesh, Peter Knippertz, Andreas H. Fink
Numerical weather prediction (NWP) models often underperform compared to simpler climatology-based precipitation forecasts in northern tropical Africa, even after statistical postprocessing. AI-based forecasting models show promise but have avoided precipitation due to its complexity. Synoptic-scale forcings like African easterly waves and other tropical waves (TWs) are important for predictability in tropical Africa, yet their value for predicting daily rainfall remains unexplored. This study uses two machine-learning models--gamma regression and a convolutional neural network (CNN)--trained on TW predictors from satellite-based GPM IMERG data to predict daily rainfall during the July-September monsoon season. Predictor variables are derived from the local amplitude and phase information of seven TW from the target and up-and-downstream neighboring grids at 1-degree spatial resolution. The ML models are combined with Easy Uncertainty Quantification (EasyUQ) to generate calibrated probabilistic forecasts and are compared with three benchmarks: Extended Probabilistic Climatology (EPC15), ECMWF operational ensemble forecast (ENS), and a probabilistic forecast from the ENS control member using EasyUQ (CTRL EasyUQ). The study finds that downstream predictor variables offer the highest predictability, with downstream tropical depression (TD)-type wave-based predictors being most important. Other waves like mixed-Rossby gravity (MRG), Kelvin, and inertio-gravity waves also contribute significantly but show regional preferences. ENS forecasts exhibit poor skill due to miscalibration. CTRL EasyUQ shows improvement over ENS and marginal enhancement over EPC15. Both gamma regression and CNN forecasts significantly outperform benchmarks in tropical Africa. This study highlights the potential of ML models trained on TW-based predictors to improve daily precipitation forecasts in tropical Africa.