A dayside aurora dataset from the Global-scale Observations of the Limb and Disk mission

TL;DR

This paper introduces a dataset of dayside auroral emissions collected by the GOLD mission, offering insights into upper atmospheric electron activity.

Contribution

The novel contribution is a dataset of auroral emissions processed to remove dayglow contamination using machine learning techniques.

Findings

The dataset includes over 47,000 auroral scans with far-ultraviolet emissions from October 2018 to June 2025.

Auroral signals are extracted using a pipeline inspired by computer vision and machine learning.

The dataset enables consistent study of auroral dynamics in the North American and Atlantic sectors.

Abstract

We present a comprehensive dataset of dayside auroral emissions observed by the Global-scale Observations of the Limb and Disk (GOLD) mission from October 2018 to June 2025. The dataset contains over 47,000 unique scans of the northern aurora in three far-ultraviolet spectral channels (OI 135.6 nm, NI 149.3 nm, and N₂ LBH), estimates of the background dayglow, binary masks of auroral locations, and other corresponding spatial and temporal metadata. The OI 135.6 nm, NI 149.3 nm, and N₂ LBH emissions are far-ultraviolet signatures of electron-impact excitation in the upper atmosphere and therefore serve as tracers of auroral electron precipitation. From this dataset, auroral pixels are directly available with no dayglow contamination of the emissions. Auroral signals are extracted through a multi-stage processing pipeline inspired by computer vision and machine learning techniques. This dataset provides a consistent view of the dayside aurora over the North American and Atlantic sectors, enabling studies of auroral dynamics with GOLD observations.

Figures