[Paper Review] Source Extractor for Dummies v5

Benne W. Holwerda|arXiv (Cornell University)|Dec 6, 2005

Astronomical Observations and InstrumentationEngineering20 citations

TL;DR

This manual serves as a comprehensive, user-driven guide to Source Extractor (SExtractor), a widely used astronomical software for automated source detection and photometry in FITS images. It explains configuration parameters, workflow best practices, and limitations—particularly in crowded fields—offering practical insights beyond the official documentation to improve accuracy and usability for both beginners and advanced users.

ABSTRACT

Source Extractor for Dummies is a user manual for the SEXtractor (Bertin and Arnouts 1996) software package for the detection of astronomical sources in fits-files of fields. It has seen much use as a quick tool and is constantly updated with new features. This fifth installment of the user manual is to catch up with these updates and improve explanations and illustration of SE parameters. The manual was written by a user, not a developer and may be incomplete, inaccurate or clearly copied from the official manual v2.3. In short: more pages, more text, less spelling-errors and don't panic. full pdf at: http://www-int.stsci.edu/~holwerda/se.html

Motivation & Objective

• To provide a detailed, accessible guide to Source Extractor (SExtractor) for researchers and students, addressing gaps in the official documentation.
• To help users configure SExtractor effectively for accurate source detection and photometry in astronomical imaging data.
• To highlight the software’s strengths in speed and scalability while addressing its limitations in accuracy and classification reliability.
• To offer practical parameter settings and workflows tailored for crowded fields and deep-sky surveys like the Hubble Deep Field.
• To serve as a living reference that evolves with user experience and software updates, improving usability beyond the official manual.

Proposed method

• The manual is based on the author’s hands-on experience with SExtractor during PhD research, supplemented by insights from Ed Smith and other users.
• It systematically documents input parameters (e.g., DETECT_MINAREA, DEBLEND_MINCONT, PHOT_APERTURES) and their impact on source detection and photometry.
• The guide includes practical examples using real data, such as the Hubble Deep Field, to demonstrate parameter effects on output catalogs.
• It provides recommended parameter sets (e.g., in Appendix B) derived from simulations of Sextans A and the HDF, validated by success rates in object classification.
• The manual explains how to use weight maps (e.g., from Drizzle) with MAP_RMS for accurate noise modeling in photometry.
• It integrates best practices for handling background estimation (BACK_SIZE, BACK_FILTERSIZE), aperture photometry (PHOT_APERTURES), and image output (CHECKIMAGE_TYPE).

Experimental results

Research questions

• RQ1 How can users configure SExtractor to achieve reliable source detection and photometry in complex, crowded astronomical fields?
• RQ2 What are the most effective parameter settings for minimizing false positives and improving photometric accuracy in deep-sky surveys?
• RQ3 How does SExtractor’s star/galaxy classification perform across different magnitudes, and when does it break down?
• RQ4 In what ways can weight maps from drizzle processing be integrated into SExtractor to improve photometric noise modeling?
• RQ5 What are the trade-offs between speed and accuracy in SExtractor’s source extraction pipeline, and how can users mitigate its limitations?

Key findings

• SExtractor performs well for fast, automated source detection and photometry on large FITS files, especially for wide-field or deep-sky surveys.
• The default parameter set yields reliable results for bright objects, but classification of faint objects as stars or galaxies becomes unreliable due to low signal-to-noise.
• Photometric parameters like MAG_AUTO and CLASS_STAR show systematic deviations when zero-point settings are incorrect, leading to spurious magnitude values.
• The star/galaxy classifier in SExtractor is based on object size and shape (via moments), but it fails for faint objects, behaving like a random classifier beyond a certain magnitude limit.
• Deblending parameters (DEBLEND_NTHRESH, DEBLEND_MINCONT) significantly affect source separation in crowded fields; optimal values (e.g., 32 thresholds, 0.005 flux threshold) improve detection accuracy.
• Using drizzle-derived weight maps with MAP_RMS and proper scaling (via FA factor) enables accurate noise modeling, improving photometric precision in combined HST data.