A statistical learning approach to Mediterranean cyclones

Chaos An Interdisciplinary Journal of Nonlinear Science, Journal Year: 2025, Volume and Issue: 35(5)

Published: May 1, 2025

Mediterranean cyclones are extreme meteorological events of which much less is known compared to their tropical, oceanic counterparts. The rising interest in such phenomena due impact on a region increasingly more affected by climate change, but precise characterization remains nontrivial task. In this work, we showcase how Bayesian algorithm (Latent Dirichlet Allocation) can classify relying wind velocity data, leading drastic dimensional reduction that allows the use supervised statistical learning techniques for detecting and tracking new cyclones.

Language: Английский

---

Calibration and uncertainty quantification for deep learning-based drought detection

International Journal of Applied Earth Observation and Geoinformation, Journal Year: 2025, Volume and Issue: 140, P. 104563 - 104563

Published: May 9, 2025

Language: Английский

---

Assessing the Role of Machine Learning in Climate Research Publications

Sustainability, Journal Year: 2024, Volume and Issue: 16(24), P. 11086 - 11086

Published: Dec. 18, 2024

Climate change is an aspect in our lives that presents urgent challenges requiring innovative approaches and collaborative efforts across diverse fields. Our research investigates the growth thematic structure of intersection between climate machine learning (ML). Employing a mixed-methods approach, we analyzed 7521 open-access publications from Web Science Core Collection (2004–2024), leveraging both R Python for data processing advanced statistical analysis. The results reveal striking 37.39% annual publications, indicating rapidly expanding increasingly significant role ML research. This accompanied by increased international collaborations, highlighting global effort to address this challenge. approach integrates bibliometrics, text mining (including word clouds, knowledge graphs with Node2Vec K-Means, factorial analysis, map, topic modeling via Latent Dirichlet Allocation (LDA)), visualization techniques uncover key trends themes. Thematic analysis using LDA revealed seven areas, reflecting multidisciplinary nature field: hydrology, agriculture, biodiversity, forestry, oceanography, forecasts, models. These findings contribute in-depth understanding evolving area inform future directions resource allocation strategies identifying established emerging themes along areas further investigation.

Language: Английский

---

Estimation of Beach Profile Response on Coastal Hydrodynamics Using LSTM-Based Encoder–Decoder Network

Journal of Marine Science and Engineering, Journal Year: 2024, Volume and Issue: 12(12), P. 2212 - 2212

Published: Dec. 2, 2024

Beach profiles are constantly changing due to external ocean forces. Estimating these changes is crucial understanding and addressing coastal erosion issues, such as shoreline advance retreat. To estimate beach profile changes, obtaining long-term, high-resolution spatiotemporal data essential. However, the limited availability of survey both on land underwater along coast, generating continuous, over extended periods a critical technological challenge. Therefore, we herein developed long short-term memory-based encoder–decoder network for effective representation learning responses temporal scales from weeks months hydrodynamics. The proposed approach was applied 12 transects seven beaches located in three different littoral systems east coast Korean Peninsula, where problems severe. performance method demonstrated improved results compared with recent study that performed same estimation task, an average root mean square error 0.50 m. Moreover, most exhibited reasonably accurate morphological shape estimated profile. instances exceed attributed extreme caused by storm waves typhoons.

Language: Английский

---

Archival and Dissemination of Knowledge on Global Warming and Protection of the Arctic

Published: Jan. 1, 2024

Language: Английский

---

Learn from Simulations, Adapt to Observations: Super-Resolution of Isoprene Emissions via Unpaired Domain Adaptation

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(21), P. 3963 - 3963

Published: Oct. 24, 2024

Plants emit biogenic volatile organic compounds (BVOCs), such as isoprene, significantly influencing atmospheric chemistry and climate. BVOC emissions estimated from bottom-up (BU) approaches (derived numerical simulations) usually exhibit denser more detailed spatial information compared to those through top-down (TD) satellite observations). Moreover, numerically simulated are typically easier obtain, even if they less reliable than acquisitions, which, being derived actual measurements, considered a trustworthy instrument for performing climate investigations. Given the coarseness relative lack of satellite-derived fine-grained could be exploited enhance them. However, observed differ regarding value range spatiotemporal resolution. In this work, we present novel deep learning (DL)-based approach increase resolution isoprene emissions, investigating adoption efficient domain adaptation (DA) techniques bridge gap between avoiding need retraining specific super-resolution (SR) algorithm on For this, propose methodology based cycle generative adversarial network (CycleGAN) architecture, which has been extensively used adapting natural images (like digital photographs) different domains. our depart standard CycleGAN framework, proposing additional loss terms that allow better DA emissions’ SR. We demonstrate proposed method’s effectiveness robustness in restoring patterns emissions. compare setups validate using emission inventories both Eventually, show strategy paves way towards robust SR solutions case mismatch training testing domains unknown data.

Language: Английский

---

Enhancing Sub-Seasonal Soil Moisture Forecasts through Land Initialization

Research Square (Research Square), Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 2, 2024

We assess the relative contributions of land, atmosphere, and oceanic initializations to forecast skill root zone soil moisture (SM) utilizing Community Earth System Model version 2 Sub-seasonal climate experiments (CESM2-SubX). Using eight sensitivity experiments, we disentangle individual impacts these three components their interactions on skill, quantified using anomaly correlation coefficient. The SubX experiment, in which land states are realistically initialized while atmosphere ocean remain climatological states, contributes 91 ± 3% total sub-seasonal across varying conditions during summer winter seasons. Most SM predictability stems from memory effect. Additionally, land-atmosphere coupling 50% land-driven predictability. A comparative analysis CESM2-SubX skills against two other models highlights potential for enhancing accuracy by improving representation precipitation feedback.

Language: Английский

---

Harnessing AI in Physical Therapy Modalities

Advances in healthcare information systems and administration book series, Journal Year: 2024, Volume and Issue: unknown, P. 269 - 278

Published: Nov. 22, 2024

The integration of AI in physical remedy is revolutionizing treatment modalities by unifying Eastern and Western approaches to recuperation. This composition examines the operation technologies, similar engine literacy real-time data analytics, enhancing practices. primarily focuses on biomechanical duties substantiation-grounded styles, while punctuate holistic ways that manipulate body-mind connection. By using AI, clinicians can enhance estimations, epitomize recuperation plans, objectively charge traditional curatives like acupuncture Tai Chi. Despite pledge expostulations sequestration, algorithm translucency, integrating different sources remain. underscores significance a clearheaded path combines puissance both optimize strategies.

Language: Английский

---