NSClean: An Algorithm for Removing Correlated Noise from JWST NIRSpec Images

TL;DR

NSClean is a fast, Python-based algorithm that effectively removes correlated noise such as vertical banding and frame noise from JWST NIRSpec images by fitting and subtracting a background model in Fourier space.

Contribution

The paper introduces NSClean, a novel algorithm and Python package that improves noise removal in JWST NIRSpec images over simpler methods, with efficient Fourier space modeling.

Findings

Removes nearly all correlated noise from NIRSpec images.

Operates efficiently, requiring only seconds per image.

Available as a free Python package from NASA JWST website.

Abstract

NSClean is an algorithm and associated python package for removing faint vertical banding and ``picture frame noise'' from JWST Near Infrared Spectrograph (NIRSpec) images. NSClean uses known dark areas to fit a background model to each exposure in Fourier space. When the model is subtracted, it removes nearly all correlated noise. Compared to simpler strategies like subtracting the rolling median, NSClean is more thorough and uniform. NSClean is computationally undemanding, requiring only a few seconds to clean an image on a typical laptop. The NSClean package is freely available from the NASA JWST website (https://webb.nasa.gov/content/forScientists/publications.html).