Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall

Hess, P.; Boers, Niklas

doi:10.1029/2021ms002765

Cited by 40 publications

(27 citation statements)

References 55 publications

(70 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In general, most of the analyzed DL were able to reproduce reasonably well the occurrence of precipitation events. However, we found that the U-Net outperformed the rest of the tested architectures by a large margin, which is in line with previous studies (Hess & Boers, 2022;Larraondo et al, 2019) that used a U-Net architecture to simulate precipitation. In general, the skill scores that measure the precision to classify heavy precipitation events (i.e., >95th percentile) were higher than those obtained for extreme precipitation events (i.e., >99th percentile), due to the unbalanced number of classes where the number of extremes is significantly reduced in the training data.…”

Section: Discussionsupporting

confidence: 92%

“…Recently, many studies have proposed using sophisticated ML methods to improve precipitation estimates in various contexts, such as precipitation nowcasting (Ayzel et al, 2019) and post-processing of NWP precipitation output (Hess & Boers, 2022). This section reviews the most relevant studies closely related to our objectives and methodology.…”

Section: Related Workmentioning

confidence: 99%

“…Recently, Hess and Boers (2022) showed that a U-Net-based network, using NWP ensemble simulations as input features, captures well heavy rainfall events. They applied DL as a post-processing step to correct biases in the NWP-predicted rainfall.…”

Section: Related Workmentioning

confidence: 99%

See 2 more Smart Citations

Intercomparison of deep learning architectures for the prediction of precipitation fields

Otero¹,

Horton²

2022

Preprint

View full text Add to dashboard Cite

In recent years, the use of deep learning methods has rapidly increased in many research fields. Similarly, they have become a powerful tool within the climate scientific community. Deep learning methods have been successfully applied for different tasks, such as identification of atmospheric patterns, weather extreme classification, or weather forecasting. However, due to the inherent complexity of the atmospheric processes, the ability of deep learning models to simulate natural processes, such as precipitation, is still challenging. Therefore, a thorough evaluation of their performance and robustness in predicting precipitation fields is still needed, especially for extreme precipitation events, which can be devastating in terms of infrastructure damage, economic losses, and even loss of life. In this study, we present a comprehensive evaluation of a set of deep learning architectures to realistically simulate precipitation, including heavy precipitation events (>95th percentile) and extreme events (>99th percentile) over the European domain. Moreover, we examine the optimal number of inputs based on the importance of the predictors derived from a layer-wise relevance propagation procedure. Among the architectures analyzed here, the U-Net network was found to be superior and outperformed the other networks to simulate precipitation events. Moreover, we found that a simplified version of the original U-Net with a single encoder-decoder level achieves similar skill scores as deeper versions for predicting precipitation extremes, significantly reducing overall complexity and computing resources.

show abstract

Section: Discussionsupporting

confidence: 92%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Intercomparison of deep learning architectures for the prediction of precipitation fields

Otero¹,

Horton²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…This changes the flavor of the problem substantially—for example, both papers prioritize standard metrics like RMSE, which is inappropriate on higher resolutions and shorter timescales (Rossa et al., 2008 ). Hess and Boers ( 2022 ) uses independent data sets, and has a particular focus on heavy rainfall events, but does not increase resolution. Finally, many authors have used convolutional neural networks for nowcasting: forecasting precipitation events over short lead times (typically 0–6 hr).…”

Section: Introductionmentioning

confidence: 99%

A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecasts

Harris

McRae

Chantry

et al. 2022

J Adv Model Earth Syst

View full text Add to dashboard Cite

show abstract

“…Therefore, UNet uses multiple convolutions and deconvolutions to fuse multiscale features. UNet has been used for correction of numerical forecast products and precipitation forecast (Grönquist, et al., 2021; Han et al., 2021; Hess & Boers, 2022), and it is representative enough as a benchmark model. GIPMN single and UNet used the predictors in both Table 1 (a and b) as inputs.…”

Section: Resultsmentioning

confidence: 99%

A Structured Graph Neural Network for Improving the Numerical Weather Prediction of Rainfall

Peng,

Li,

Chen

et al. 2023

JGR Atmospheres

View full text Add to dashboard Cite

The current challenges of numerical weather prediction (NWP) of rainfall mainly stem from the complex and multiscale nature of rainfall. In recent years, as observation capability improved worldwide, there has been an increased feasibility to use data‐driven models to enhance forecasting performance with rainfall observation. Compared to traditional statistical and machine learning models, deep‐learning models show considerable promise in capturing the spatial‐temporal features of weather processes from multiple predictors, but the convolution‐based feature extractor is suboptimal due to the linear nature of convolution kernels. In this study, a multilevel forecasting model is proposed to forecast each rainfall level, in which each submodel adopts a graph neural network for feature extraction. Spatial and temporal propagation functions based on grid structure are designed to explicitly represent feature fusion and propagation of multiple predictors across multiple scales. On model training, a weight setting strategy that balances the impact of samples with different rainfall values on the total training loss is proposed, and a soft classification label is designed to convert observed rainfall into the probability of rainfall above each threshold. The proposed model was trained and validated on NWP data provided by European Center for Medium‐Range Weather Forecast, and results show significant improvement over the NWP in terms of threat score (TS) and Heidke Skill Score (HSS) scores. Analysis of the forecast results for two typical rainfall processes also illustrates that the proposed method can predict rainfall with more reasonable location and intensity.

show abstract

Deep Learning for Improving Numerical Weather Prediction of Heavy Rainfall

Cited by 40 publications

References 55 publications

Intercomparison of deep learning architectures for the prediction of precipitation fields

Intercomparison of deep learning architectures for the prediction of precipitation fields

A Generative Deep Learning Approach to Stochastic Downscaling of Precipitation Forecasts

A Structured Graph Neural Network for Improving the Numerical Weather Prediction of Rainfall

Contact Info

Product

Resources

About