WxC-Bench: A Novel Dataset for Weather and Climate Downstream Tasks

TL;DR

WxC-Bench (Weather and Climate Bench), a multi-modal dataset designed to support the development of generalizable AI models for downstream use-cases in weather and climate research, is introduced.

Abstract

High-quality machine learning (ML)-ready datasets play a foundational role in developing new artificial intelligence (AI) models or fine-tuning existing models for scientific applications such as weather and climate analysis. Unfortunately, despite the growing development of new deep learning models for weather and climate, there is a scarcity of curated, pre-processed machine learning (ML)-ready datasets. Curating such high-quality datasets for developing new models is challenging particularly because the modality of the input data varies significantly for different downstream tasks addressing different atmospheric scales (spatial and temporal). Here we introduce WxC-Bench (Weather and Climate Bench), a multi-modal dataset designed to support the development of generalizable AI models for downstream use-cases in weather and climate research. WxC-Bench is designed as a dataset of datasets for developing ML-models for a complex weather and climate system, addressing selected downstream tasks as machine learning phenomenon. WxC-Bench encompasses several atmospheric processes from meso-$β$ (20 - 200 km) scale to synoptic scales (2500 km), such as aviation turbulence, hurricane intensity and track monitoring, weather analog search, gravity wave parameterization, and natural language report generation. We provide a comprehensive description of the dataset and also present a technical validation for baseline analysis. The dataset and code to prepare the ML-ready data have been made publicly available on Hugging Face -- https://huggingface.co/datasets/nasa-impact/WxC-Bench

Figures (20)

• Figure 1: Illustration showing the applicability of the proposed WxC-Bench in the overall AI/ML workflows after converting raw non-ML ready datasets to ML ready datasets. The ML-ready datasets can then be directly used for training new AI models or fine-tuning pre-trained AI models. The WxC-Bench dataset can be used with AI models relevant to downstream applications across multiple spatial and temporal scales and resolutions.
• Figure 2: Spatial distribution of PIREPs turbulence reports over CONUS from 2003-Present. Note that values are scaled logarithmically.
• Figure 3: Distribution of the coarse-grained input features (scaled background atmospheric state) and output labels (scaled GW momentum fluxes) for a single month (January 2010) from ERA5.
• Figure 4: Illustration of training images for (a) sea level pressure (b) temperature at 2m, for January 01, 2019.
• Figure 5: Example of the input data for the Long-range precipitation forecast task. Each panel displays observations from one of the channels of the three observational datasets that form the input for the Long-range precipitation forecast. Panels (a), (b), and (c) show visible and infrared geostationary observations from the GridSat-B1 dataset. Panels (d), (e), and (f) show corresponding observations from polar-orbiting satellites from the PATMOS-x dataset. Panels (g), (h), and (i) show microwave observations from the SSMI dataset.