Deep learning-based spatial downscaling and its application for tropical cyclone detection in the western North Pacific

被引:0
作者
Chen, Anqi [1 ]
Yuan, Chaoxia [1 ,2 ]
机构
[1] Nanjing Univ Informat Sci & Technol, Key Lab Meteorol Disaster, Collaborat Innovat Ctr Forecast & Evaluat Meteorol, Minist Educ,Inst Climate & Applicat Res ICAR, Nanjing, Peoples R China
[2] Japan Agcy Marine Earth Sci & Technol, Applicat Lab, Yokohama, Japan
基金
中国国家自然科学基金;
关键词
deep learning-based downscaling method; generative adversarial networks; super-resolution; global climate model; tropical cyclone; RESOLUTION; CMIP5; MODELS;
D O I
10.3389/feart.2024.1345714
中图分类号
P [天文学、地球科学];
学科分类号
07 ;
摘要
Resolution of global climate models (GCMs) significantly influences their capacity to simulate extreme weather such as tropical cyclones (TCs). However, improving the GCM resolution is computationally expensive and time-consuming, making it challenging for many research organizations worldwide. Here, we develop a downscaling model, MSG-SE-GAN, based on the Generative Adversarial Networks (GAN) together with Multiscale Gradient (MSG) technique and a Squeeze-and-Excitation (SE) Net, to achieve 10-folded downscaling. GANs consist of a generator and a discriminator network that are trained adversarially, and are often used for generating new data that resembles a given dataset. MSG enables generation and discrimination of multi-scale images within a single model. Inclusion of an attention layer of SE captures better underlying spatial structure while preserving accuracy. The MSG-SE-GAN is stable and fast converging. It outperforms traditional bilinear interpolation and other deep-learning methods such as Super-Resolution Convolutional Neural Networks (SRCNN) and MSG-GAN in downscaling low-resolution meteorological data in assessment metrics and power spectral density. The MSG-SE-GAN has been used to downscale the TC-related variables in the western North Pacific in the low-resolution GCMs of HadGEM3-GC31 and EC-Earth3P, respectively. The downscaled data show highly similar TC activities to the direct outputs of the high-resolution HadGEM3-GC31 and EC-Earth3P, respectively. These results not only suggest the validity of the MSG-SE-GAN but also indicate its possible portability among low-resolution GCMs.
引用
收藏
页数:14
相关论文
共 56 条
[1]   Role of predictors in downscaling surface temperature to river basin in India for IPCC SRES scenarios using support vector machine [J].
Anandhi, Aavudai ;
Srinivas, V. V. ;
Kumar, D. Nagesh ;
Nanjundiah, Ravi S. .
INTERNATIONAL JOURNAL OF CLIMATOLOGY, 2009, 29 (04) :583-603
[2]   Configuration and intercomparison of deep learning neural models for statistical downscaling [J].
Bano-Medina, Jorge ;
Manzanas, Rodrigo ;
Manuel Gutierrez, Jose .
GEOSCIENTIFIC MODEL DEVELOPMENT, 2020, 13 (04) :2109-2124
[3]   On using principal components to represent stations in empirical-statistical downscaling [J].
Benestad, Rasmus E. ;
Chen, Deliang ;
Mezghani, Abdelkader ;
Fan, Lijun ;
Parding, Kajsa .
TELLUS SERIES A-DYNAMIC METEOROLOGY AND OCEANOGRAPHY, 2015, 67
[4]   Statistical downscaling of daily precipitation using support vector machines and multivariate analysis [J].
Chen, Shien-Tsung ;
Yu, Pao-Shan ;
Tang, Yi-Hsuan .
JOURNAL OF HYDROLOGY, 2010, 385 (1-4) :13-22
[5]   Gated-GAN: Adversarial Gated Networks for Multi-Collection Style Transfer [J].
Chen, Xinyuan ;
Xu, Chang ;
Yang, Xiaokang ;
Song, Li ;
Tao, Dacheng .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2019, 28 (02) :546-560
[6]  
Clark P., 2006, High resolution model progress and plans
[7]   The role of horizontal resolution in simulating drivers of the global hydrological cycle [J].
Demory, Marie-Estelle ;
Vidale, Pier Luigi ;
Roberts, Malcolm J. ;
Berrisford, Paul ;
Strachan, Jane ;
Schiemann, Reinhard ;
Mizielinski, Matthew S. .
CLIMATE DYNAMICS, 2014, 42 (7-8) :2201-2225
[8]   Biases in the Atlantic ITCZ in Seasonal-Interannual Variations for a Coarse- and a High-Resolution Coupled Climate Model [J].
Doi, Takeshi ;
Vecchi, Gabriel A. ;
Rosati, Anthony J. ;
Delworth, Thomas L. .
JOURNAL OF CLIMATE, 2012, 25 (16) :5494-5511
[9]   Image Super-Resolution Using Deep Convolutional Networks [J].
Dong, Chao ;
Loy, Chen Change ;
He, Kaiming ;
Tang, Xiaoou .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2016, 38 (02) :295-307
[10]   Downscaling CMIP5 climate models shows increased tropical cyclone activity over the 21st century [J].
Emanuel, Kerry A. .
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 2013, 110 (30) :12219-12224