Prospects for future studies using deep imaging: Analysis of individual Galactic cirrus filaments

TL;DR

This paper develops machine learning techniques to identify and analyze individual Galactic cirrus filaments in deep optical survey data, providing insights into their properties and a framework for future low surface brightness feature studies.

Contribution

It introduces neural network-based methods for isolating Galactic cirrus filaments in SDSS data and offers a detailed analysis of their photometric properties and colour distribution.

Findings

Filaments brighter than 26 mag arcsec$^{-2}$ are identifiable by their colours.

Data processing significantly affects colour estimation.

Most filaments have specific colour ranges in g-r and r-i.

Abstract

The presence of Galactic cirrus is an obstacle for studying both faint objects in our Galaxy and low surface brightness extragalactic structures. With the aim of studying individual cirrus filaments in SDSS Stripe 82 data, we develop techniques based on machine learning and neural networks that allow one to isolate filaments from foreground and background sources in the entirety of Stripe 82 with a precision similar to that of the human expert. Our photometric study of individual filaments indicates that only those brighter than 26 mag arcsec$^{-2}$ in the SDSS $r$ band are likely to be identified in SDSS Stripe~82 data by their distinctive colours in the optical bands. We also show a significant impact of data processing (e.g. flat-fielding, masking of bright stars, and sky subtraction) on colour estimation. Analysing the distribution of filaments' colours with the help of mock simulations, we conclude that most filaments have colours in the following ranges: $0.55 \leq g-r \leq 0.73$ and $0.01 \leq r-i \leq 0.33$. Our work provides a useful framework for an analysis of all types of low surface brightness features (cirri, tidal tails, stellar streams, etc.) in existing and future deep optical surveys. For practical purposes, we provide the catalogue of dust filaments.