Extreme Precipitation Nowcasting using Transformer-based Generative Models

This paper presents an innovative approach to extreme precipitation nowcasting by employing Transformer-based generative models, namely NowcastingGPT with Extreme Value Loss (EVL) regularization. Leveraging a comprehensive dataset from the Royal Netherlands Meteorological Institute (KNMI), our study focuses on predicting short-term precipitation with high accuracy. We introduce a novel method for computing EVL without assuming fixed extreme representations, addressing the limitations of current models in capturing extreme weather events. We present both qualitative and quantitative analyses, demonstrating the superior performance of the proposed NowcastingGPT-EVL in generating accurate precipitation forecasts, especially when dealing with extreme precipitation events. The code is available at \url{https://github.com/Cmeo97/NowcastingGPT}.

• Publication year

  2024

• Venue

  arXiv.org

• Publication date

  2024-03-06

• Fields of study

  Computer Science, Environmental Science

• Identifiers
  DOI 10.48550/arXiv.2403.03929arXiv 2403.03929
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Preformer: Simple and Efficient Design for Precipitation Nowcasting With Transformers

  2024cited by this paper
• Meta-Transfer-Learning for Time Series Data with Extreme Events: An Application to Water Temperature Prediction

  2023cited by this paper
• Nowcasting of Extreme Precipitation Using Deep Generative Models

  2023influential reference
• Temporally Consistent Transformers for Video Generation

  2022influential reference
• Rainformer: Features Extraction Balanced Network for Radar-Based Precipitation Nowcasting

  2022cited by this paper
• A spatiotemporal deep learning model ST-LSTM-SA for hourly rainfall forecasting using radar echo images

  2022cited by this paper
• AA-TransUNet: Attention Augmented TransUNet For Nowcasting Tasks

  2022cited by this paper
• MaskGIT: Masked Generative Image Transformer

  2022cited by this paper
• MS-nowcasting: Operational Precipitation Nowcasting with Convolutional LSTMs at Microsoft Weather

  2021cited by this paper
• PredRANN: The spatiotemporal attention Convolution Recurrent Neural Network for precipitation nowcasting

  2021cited by this paper
• NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion

  2021influential reference
• AutoNowP: An Approach Using Deep Autoencoders for Precipitation Nowcasting Based on Weather Radar Reflectivity Prediction

  2021cited by this paper
• VideoGPT: Video Generation using VQ-VAE and Transformers

  2021cited by this paper
• Skilful precipitation nowcasting using deep generative models of radar

  2021cited by this paper
• Supplementary Material for - SEVIR : A Storm Event Imagery Dataset for Deep Learning Applications in Radar and Satellite Meteorology

  2020cited by this paper
• A review of radar-based nowcasting of precipitation and applicable machine learning techniques

  2020cited by this paper
• SmaAt-UNet: Precipitation Nowcasting using a Small Attention-UNet Architecture

  2020cited by this paper
• Overview of Observed Clausius-Clapeyron Scaling of Extreme Precipitation in Midlatitudes

  2020cited by this paper
• Spatial and Temporal Evaluation of Radar Rainfall Nowcasting Techniques on 1,533 Events

  2020cited by this paper
• Taming Transformers for High-Resolution Image Synthesis

  2020influential reference
• Modeling Extreme Events in Time Series Prediction

  2019cited by this paper
• Extreme Value Theory

  2019cited by this paper
• Pysteps: an open-source Python library for probabilistic precipitation nowcasting (v1.0)

  2019influential reference
• GENERATIVE ADVERSARIAL NETS

  2018cited by this paper
• Neural Discrete Representation Learning

  2017influential reference
• Global projections of river flood risk in a warmer world

  2017cited by this paper
• Attention is All you Need

  2017cited by this paper
• Convolutional LSTM Network: A Machine Learning Approach for Precipitation Nowcasting

  2015cited by this paper
• Interpolation Methods for the Climate Atlas

  2012cited by this paper
• Extreme value theory and Value-at-Risk: Relative performance in emerging markets

  2004cited by this paper
• An Introduction to Statistical Modeling of Extreme Values

  2001influential reference

• An efficient parallel runoff forecasting model for capturing global and local feature information

  2025cites this paper
• Machine learning in an expectation-maximisation framework for nowcasting

  2025cites this paper
• A Convolutional Neural Network for Lightning Strikes Detection Over the Italian Territory Using SEVIRI@MSG Data

  2025cites this paper
• Deep Learning and Foundation Models for Weather Prediction: A Survey

  2025cites this paper
• BlockGPT: Spatio-Temporal Modelling of Rainfall via Frame-Level Autoregression

  2025cites this paper
• Masked Generative Priors Improve World Models Sequence Modelling Capabilities

  2024cites this paper
• Precipitation Nowcasting Using Physics Informed Discriminator Generative Models

  2024cites this paper