Abstract

Identifying disc galaxies at cosmic noon (z = 2) is challenging because high-redshift systems subtend smaller angular sizes, often exhibit irregular morphologies, and may lack the clear spiral features that define discs in the local Universe. This thesis develops and tests a combined visual and quantitative approach to disc identification using JWST COSMOS-Web NIRCam F150W imaging. A sample of 50 galaxies—25 visually classified spirals and 25 visually classified ellipticals—was analysed using Sérsic profile fitting with GALFIT and non-parametric morphological measurements with statmorph. The spiral sample has a mean Sérsic index of n¯ = 1.85 (σ = 0.98), whilst the elliptical sample has n¯ = 4.38 (σ = 1.98). Applying the conventional n = 2.5 threshold recovers 76% of spirals as disc-like and 84% of ellipticals as spheroid-like. Among non-parametric diagnostics, concentration C and Gini coefficient G provide the clearest population-level separation; asymmetry A adds meaningful but noisier discrimination; and smoothness S is effectively uninformative at these redshifts. The combination of C, A, and M20 is identified as the recommended non-parametric diagnostic framework, providing the clearest overall population-level separation without dependence on parametric fitting convergence. Several individual galaxies are identified as potential misclassifications or systems with hidden disc components, illustrating that no single diagnostic is sufficient in isolation and that the combination of visual and quantitative methods provides a substantially more robust classification framework than either approach alone.

Advisor

DeGroot, Laura

Department

Physics

Disciplines

Other Physics | Physics

Publication Date

2026

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Download

Available for download on Thursday, May 15, 2031

Share

COinS
 

© Copyright 2026 Akshit Deshpande