MSG-GAN-SD: A Multi-Scale Gradients GAN for Statistical Downscaling of 2-Meter Temperature over the EURO-CORDEX Domain

Abstract

One of the most important open challenges in climate science is downscaling. It is a procedure that allows making predictions at local scales, starting from climatic field information available at large scale. Recent advances in deep learning provide new insights and modeling solutions to tackle downscaling-related tasks by automatically learning the coarse-to-fine grained resolution mapping. In particular, deep learning models designed for super-resolution problems in computer vision can be exploited because of the similarity between images and climatic fields maps. For this reason, a new architecture tailored for statistical downscaling (SD), named MSG-GAN-SD, has been developed, allowing interpretability and good stability during training, due to multi-scale gradient information. The proposed architecture, based on a Generative Adversarial Network (GAN), was applied to downscale ERA-Interim 2-m temperature fields, from 83.25 to 13.87 km resolution, covering the EURO-CORDEX domain within the 1979–2018 period. The training process involves seasonal and monthly dataset arrangements, in addition to different training strategies, leading to several models. Furthermore, a model selection framework is introduced in order to mathematically select the best models during the training. The selected models were then tested on the 2015–2018 period using several metrics to identify the best training strategy and dataset arrangement, which finally produced several evaluation maps. This work is the first attempt to use the MSG-GAN architecture for statistical downscaling. The achieved results demonstrate that the models trained on seasonal datasets performed better than those trained on monthly datasets. This study presents an accurate and cost-effective solution that is able to perform downscaling of 2 m temperature climatic maps.

Downscaling Seasonal Weather Forecasting with Generative Adversarial Networks

“…Such approaches could refine the network's capacity to produce more precise predictions. Despite these challenges, GANs hold considerable promise, offering advantages such as cost-effectiveness and decreased computational demands, once the network is trained for the defined area of interest [33]. Consequently, they are promising tools for achieving long-term, high-resolution predictions, potentially revolutionizing multiple sectors reliant on such detailed forecasting.…”

Super-Resolution Reconstruction of Humidity Fields based on Wasserstein Generative Adversarial Network with Gradient Penalty

“…Li Tao et al used GANs to downscale the ERA5 2m temperature product and reconstruct a higher-resolution temperature product [26]. Gabriele Accarino et al proposed a method called MSG-GAN-SD [27] based on Wasserstein Generative Adversarial Nets -Gradient Penalty(WGAN-GP). Through monthly and quarterly training as well as the use of different discriminator update strategies, they effectively performed statistical downscaling on 2m temperature fields.…”

Analyzing Driving Characteristics of Personal Mobility According to Urban Physical Environments - Focused on the Yongshin-dong, Dongdaemun-gu, Seoul

“…Numerous studies [30,16,9,1,25] have utilized deep learning for downscaling numerical model simulation data. ResLap [9] employed LapSRN [19], a model that proposes progressive upsampling as its backbone and calculated losses from each progressively downscaled predictions using a root mean square error.…”

“…ResLap [9] employed LapSRN [19], a model that proposes progressive upsampling as its backbone and calculated losses from each progressively downscaled predictions using a root mean square error. MSG-GAN-SD [1] used all progressively downscaled predictions for a GAN (Generative Adversarial Networks) discriminator, to leverage semantic information appearing at different scales. While both studies' models could generate outputs at various scales, they could not infer at…”