Prithvi WxC: Foundation Model for Weather and Climate

Triggered by the realization that AI emulators can rival the performance of traditional numerical weather prediction models running on HPC systems, there is now an increasing number of large AI models that address use cases such as forecasting, downscaling, or nowcasting. While the parallel developm...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2024-09
Main Authors: Schmude, Johannes, Roy, Sujit, Trojak, Will, Jakubik, Johannes, Daniel Salles Civitarese, Singh, Shraddha, Kuehnert, Julian, Kumar, Ankur, Gupta, Aman, Phillips, Christopher E, Kienzler, Romeo, Szwarcman, Daniela, Gaur, Vishal, Shinde, Rajat, Lal, Rohit, Arlindo Da Silva, Guevara Diaz, Jorge Luis, Jones, Anne, Pfreundschuh, Simon, Lin, Amy, Sheshadri, Aditi, Nair, Udaysankar, Valentine Anantharaj, Hamann, Hendrik, Watson, Campbell, Maskey, Manil, Lee, Tsengdar J, Juan Bernabe Moreno, Ramachandran, Rahul
Format: Article
Language:English
Subjects:
Climate models
Encoders-Decoders
Gravity waves
Nowcasting
Numerical weather forecasting
Parameterization
Parameters
Prediction models
Regional development
Topology
Weather forecasting
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Triggered by the realization that AI emulators can rival the performance of traditional numerical weather prediction models running on HPC systems, there is now an increasing number of large AI models that address use cases such as forecasting, downscaling, or nowcasting. While the parallel developments in the AI literature focus on foundation models -- models that can be effectively tuned to address multiple, different use cases -- the developments on the weather and climate side largely focus on single-use cases with particular emphasis on mid-range forecasting. We close this gap by introducing Prithvi WxC, a 2.3 billion parameter foundation model developed using 160 variables from the Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2). Prithvi WxC employs an encoder-decoder-based architecture, incorporating concepts from various recent transformer models to effectively capture both regional and global dependencies in the input data. The model has been designed to accommodate large token counts to model weather phenomena in different topologies at fine resolutions. Furthermore, it is trained with a mixed objective that combines the paradigms of masked reconstruction with forecasting. We test the model on a set of challenging downstream tasks namely: Autoregressive rollout forecasting, Downscaling, Gravity wave flux parameterization, and Extreme events estimation. The pretrained model with 2.3 billion parameters, along with the associated fine-tuning workflows, has been publicly released as an open-source contribution via Hugging Face.
ISSN:2331-8422