A Machine Learning-Based Parameterized Tropical Cyclone Precipitation Model DOI Creative Commons

Yi Lu,

Jie Yin, Peiyan Chen

et al.

International Journal of Disaster Risk Science, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 9, 2024

Abstract Current simulation models considerably underestimate local-scale, short-duration extreme precipitation induced by tropical cyclones (TCs). This problem needs to be addressed establish active response policies for TC-induced disasters. Taking Shanghai, a coastal megacity, as study area and based on the observations from 192 meteorological stations in city during 2005–2018, this optimized parameterized Tropical Cyclone Precipitation Model (TCPM) initially designed TCs at national scale (China) local or regional scales using machine learning (ML) methods, including random forest (RF), gradient boosting (XGBoost), ensemble (EL) algorithms. The TCPM-ML was applied multiple temporal hazard assessment. results show that: (1) not only improved TCPM performance simulating hourly precipitations, but also preserved physical meaning of results, contrary ML methods; (2) Machine algorithms enhanced ability reproduce observations, although precipitations remained slightly underestimated; (3) Best obtained with XGBoost EL Combining strengths both RF, algorithm yielded best overall performance. provides essential model support TC disaster risk assessment China.

Language: Английский

A Novel Iterative Self-Organizing Pixel Matrix Entanglement Classifier for Remote Sensing Imagery DOI
Guoqing Zhou, Lihuang Qian, Paolo Gamba

et al.

IEEE Transactions on Geoscience and Remote Sensing, Journal Year: 2024, Volume and Issue: 62, P. 1 - 21

Published: Jan. 1, 2024

Language: Английский

Citations

7
A Reliable Generative Adversarial Network Approach for Climate Downscaling and Weather Generation DOI Creative Commons
Neelesh Rampal, Peter B. Gibson, Steven C. Sherwood

et al.

Journal of Advances in Modeling Earth Systems, Journal Year: 2025, Volume and Issue: 17(1)

Published: Jan. 1, 2025

Abstract Anticipating climate impacts and risks in present or future climates requires predicting the statistics of high‐impact weather events at fine‐scales. Direct numerical simulations fine‐scale are computationally too expensive for many applications. While deterministic‐based (deep‐learning statistical) downscaling low‐resolution several orders magnitude faster than direct simulations, it suffers from limitations. These limitations include tendency to regress mean, which produces excessively smooth predictions underestimates extreme events. They also fail preserve statistical measures that key research. We use a conditional GAN (cGAN) architecture downscale daily precipitation as Regional Climate Model (RCM) emulator. The cGAN generates plausible residuals on top predictable expectation state produced by deterministic deep learning algorithm. skill cGANs is highly sensitive hyperparameter known weight adversarial loss (), where value required accurate results varies with season performance metric, casting doubt reliability usually implemented. However, applying simple intensity constraint function, possible obtain reliable across spanning two magnitude. CGANs considerably more skillful capturing climatological statistics, including distribution spatial characteristics With this modification, we expect be readily transferable other applications time periods, making them useful generator representing event climates.

Language: Английский

Citations

0
CARE-SST: Context-Aware reconstruction diffusion model for Sea surface temperature DOI
Minki Choo, Sihun Jung, Jungho Im

et al.

ISPRS Journal of Photogrammetry and Remote Sensing, Journal Year: 2025, Volume and Issue: 220, P. 454 - 472

Published: Jan. 9, 2025

Language: Английский

Citations

0
An Overview of Statistical Downscaling Methods: Techniques, Applications, and Advances DOI

David Labeurthre,

Anatole Reffet,

Anthony Schrapffer

et al.

Published: Jan. 1, 2025

Language: Английский

Citations

0
Suitability of CMIP6 Models Considering Statistical Downscaling Based on GloH2O and E-OBS Dataset in River Basin Districts of the Southeastern Baltic Sea Basin DOI Creative Commons
Vytautas Akstinas, Karolina Gurjazkaitė, Diana Meilutytė-Lukauskienė

et al.

Atmosphere, Journal Year: 2025, Volume and Issue: 16(2), P. 229 - 229

Published: Feb. 18, 2025

Climate projections based on global climate models (GCMs) are generally subject to large uncertainties, as the only reflect local in past a limited extent. Statistical downscaling is most cost-effective approach identify systematic biases of GCMs from and eliminate them projections. This study seeks evaluate effectiveness capturing climatic characteristics at river basin district scale by applying gridded statistical techniques using regional datasets. The historical observational datasets E-OBS GloH2O were selected downscale raw data 17 ~1° grid cells 0.25° resolution. dataset supported dense network meteorological stations Europe, while covering all continents. results show that suitability varies depending parameter. revealed advantages performance representing during period emphasized crucial role for good depiction. Such an provides possibility assess relative high-resolution reanalysis datasets, generating statistically downscaled best ranked GCMs. strategies used this can help appropriate assemble right ensemble specific studies.

Language: Английский

Citations

0
A novel generative adversarial network and downscaling scheme for GRACE/GRACE-FO products: Exemplified by the Yangtze and Nile River Basins DOI
Jielong Wang, Yunzhong Shen, Joseph L. Awange

et al.

The Science of The Total Environment, Journal Year: 2025, Volume and Issue: 969, P. 178874 - 178874

Published: Feb. 24, 2025

Language: Английский

Citations

0
MDG625: a daily high-resolution meteorological dataset derived by a geopotential-guided attention network in Asia (1940–2023) DOI Creative Commons
Zijiang Song,

Zhixiang Cheng,

Yuying Li

et al.

Earth system science data, Journal Year: 2025, Volume and Issue: 17(4), P. 1501 - 1514

Published: April 11, 2025

Abstract. The long-term and reliable meteorological reanalysis dataset with high spatial–temporal resolution is crucial for various hydrological applications, especially in regions or periods scarce situ observations limited open-access data. Based on the fifth-generation (ERA5, produced by European Centre Medium-Range Weather Forecasts, 0.25°×0.25°, since 1940) CLDAS (China Meteorological Administration Land Data Assimilation System, 0.0625°×0.0625°, 2008), we propose a novel downscaling method Geopotential-guided Attention Network (GeoAN), leveraging spatial of extended historical coverage ERA5, produce daily multi-variable (2 m temperature, surface pressure, 10 wind speed) MDG625. MDG625 (0.0625° Dataset derived GeoAN) covers most Asia from 0.125° S to 64.875° N 60.125 160.125° E, contains data starting 1940. Compared other methods, GeoAN shows better performance R2 speed reach 0.990, 0.998, 0.781, respectively). demonstrates superior continuity consistency both temporal perspectives. We anticipate that this dataset, MDG625, will aid climate studies contribute improving accuracy products 1940s. (Song et al., 2024) presented at https://doi.org/10.57760/sciencedb.17408, code can be found https://github.com/songzijiang/GeoAN (last access: 8 April 2025).

Language: Английский

Citations

0
CMIP6-driven 10 km super-resolution daily climate projections with PET estimates in China DOI Creative Commons

Fuyao Zhang,

Xiubin Li, Xue Wang

et al.

Scientific Data, Journal Year: 2025, Volume and Issue: 12(1)

Published: April 30, 2025

Global warming has intensified extreme weather events, posing challenges to regional climate and hydro-ecological systems. To address the low-resolution limitations of current multi-climate variables potential evapotranspiration (PET), this study develops a super-resolution fusion framework based on deep residual attention mechanisms, establishing China's 10-km resolution multi-model-multi-scenario high-resolution PET dataset (SRCPCN10). The Residual Channel Attention Network (RCAN) demonstrates exceptional downscaling performance for temperature, radiation, pressure (R2/KGE > 0.99), while precipitation exhibits significantly lower accuracy (R2 = 0.897) due spatial discontinuity. findings reveal distinct emission-gradient responses in future under SSP scenarios, with increases escalating alongside radiative forcing intensification. comparison annual mean differences between original CMIP6 downscaled data showed excellent agreement, most indices differing by less than 1%. This work overcomes traditional limitations, providing kilometer-scale multivariate watershed hydrological modeling, agricultural risk assessment, carbon-neutral pathway optimization, enhancing precision adaptation strategies.

Language: Английский

Citations

0
Spatial Downscaling of Satellite Sea Surface Wind with Soft-Sharing Multi-Task Learning DOI Creative Commons

Yinlei Yue,

Jia Liu, Yongjian Sun

et al.

Remote Sensing, Journal Year: 2025, Volume and Issue: 17(4), P. 587 - 587

Published: Feb. 8, 2025

Sea surface wind (SSW) plays a pivotal role in numerous research endeavors pertaining to meteorology and oceanography. SSW fields derived from remote sensing have been widely applied; however, regional local studies require higher-spatial-resolution identify refined details. Most of the existing based on deep learning constructed mappings low-resolution inputs high-resolution downscaled estimates. However, these methods failed capture relationships between multiple variables as revealed by physical processes. Therefore, this paper proposes spatial downscaling approach for satellite sea that employs soft-sharing multi-task learning. temperature water vapor are included auxiliary SSW, considering close correlation principles data availability. The is designed an task integrated into network with generative adversarial dual regression structures. proposed achieves flexible parameter sharing information exchange tasks through mechanism bridge modules. Comprehensive experiments were conducted WindSat products at 0.25° Remote Sensing Systems. experimental results validate outstanding capability methodology respect precision comparison buoy measurements reconstruction quality.

Language: Английский

Citations

0
Statistical spatial downscaling of significant wave height in a regional sea from the global ERA5 dataset DOI Creative Commons
Bing Yuan, Marcel Ricker, Wei Chen

et al.

Ocean Engineering, Journal Year: 2025, Volume and Issue: 329, P. 121100 - 121100

Published: April 6, 2025

Language: Английский

Citations

0