Cluster Catalog
This page summarizes the cluster catalogs generated in Brown & Gnedin 2021. The catalog includes the radii, all EFF fit parameters, densities, uncertainties on all these parameters, and a few key LEGUS properties such as mass and age. All data needed to replicate plots 10-17 is included in this catalog. Any references below to equations or figures are from our paper.
The catalog can be downloaded with wget:
wget https://raw.githubusercontent.com/gillenbrown/LEGUS-sizes/master/cluster_sizes_brown_gnedin_21.txt
It can also be directly accessed at that URL.
In what follows we demonstrate how to load and use the data, then detail what each column in the data is.
Reading the Data
The catalog is written with Astropy in the .ecsv format, which can be read by any .csv reader, but allows for preservation of data types when read by Astropy. Here is an example of how to reconstruct the data shown in Figure 11.
from astropy import table
import matplotlib.pyplot as plt
import numpy as np
catalog = table.Table.read("cluster_sizes_brown_gnedin_21.txt", format="ascii.ecsv")
# parse the LEGUS mass errors
catalog["mass_msun_e-"] = catalog["mass_msun"] - catalog["mass_msun_min"]
catalog["mass_msun_e+"] = catalog["mass_msun_max"] - catalog["mass_msun"]
# get the clusters with reliable radii and masses.
mask = catalog["reliable_radius"] & catalog["reliable_mass"]
subset = catalog[mask]
# plot the data
fig, ax = plt.subplots()
ax.errorbar(
x=subset["mass_msun"],
y=subset["r_eff_pc"],
fmt="o",
markersize=2,
lw=0.3,
xerr=[subset["mass_msun_e-"], subset["mass_msun_e+"]],
yerr=[subset["r_eff_pc_e-"], subset["r_eff_pc_e+"]],
)
# plot formatting
ax.set_xscale("log")
ax.set_yscale("log")
ax.set_xlim(1e2, 1e6)
ax.set_ylim(0.1, 40)
ax.set_xlabel("Mass [$M_\odot$]")
ax.set_ylabel("Radius [pc]")
Catalog Columns
Here we detail the meaning of all columns in the data table.
Basic Cluster Properties
field
The identifier for the LEGUS field. Note that some galaxies are split over multiple fields (NGC 1313, NGC 4395, NGC 628, and NGC 7793), and that one field contains multiple galaxies (NGC 5194 and NGC 5195).
ID
The cluster ID assigned by LEGUS. This was done on a field-by-field basis.
galaxy
The galaxy the cluster belongs to. NGC 5194 and NGC 5195 are separated manually (see Figure 1).
galaxy_distance_mpc, galaxy_distance_mpc_err
Distance to the galaxy and its uncertainty, in Mpc. We use the TRGB distances to all LEGUS galaxies provided by Sabbi et al. 2018, except for NGC 1566. See the end of Section 2.4 for more on distances used.
galaxy_stellar_mass, galaxy_sfr, galaxy_ssfr
Stellar masses, star formation rates, and specific star formation rates of the host galaxy, from Calzetti et al. 2015. See the caption of Table 1 for more details. Stellar masses are in M⊙, star formation rates are in M⊙yr-1, and specific star formation rates are in yr-1.
LEGUS Cluster Properties
These properties come from the LEGUS catalogs. If using quantities from those papers, please cite them appropriately: Calzetti et al. 2015, Adamo et al. 2017, Cook et al. 2019.
RA, Dec
Right ascension and declination from the LEGUS catalog.
x_legus, y_legus
X/Y pixel position of the cluster in the field from the LEGUS catalog.
morphology_class
Classification of the morphology of the clusters by LEGUS. Class 1 objects are compact and centrally concentrated with a homogeneous color. Class 2 clusters have slightly elongated density profiles and a less symmetric light distribution. We do not include Class 3 (compact associations) or Class 4 (stars or artifacts) objects in this catalog.
morphology_class_source
The source of the classification of the morphology in the morphology attribute. When available, we use the mode of the classifications from multiple LEGUS team members, called human_mode in this column. Additionally, machine learning classifications (ml) are available for several galaxies. For NGC 5194 and NGC 5195, we use the human classifications for clusters where those are available, and supplement with machine learning classifications for clusters not inspected by humans. In NGC 1566, we use the hybrid classification system (hybrid) created by the LEGUS team, where some clusters are inspected by humans only, some by machine learning only, and some with a machine learning classification verified by humans.
age_yr, age_yr_min, age_yr_max
Cluster age in years and its minimum and maximum allowed value from LEGUS. This uses the deterministic SED fitting method presented in Adamo et al. 2017.
mass_msun, mass_msun_min, mass_msun_max
Cluster mass in M⊙ and its minimum and maximum allowed value from LEGUS using the same SED fitting as age_yr.
Fit Parameters
The catalog includes all the fit parameters and their uncertainties. The EFF profile takes the basic form:
We generalize this (Equations 2-4) to include ellipticity by including an axis ratio and position angle.
The uncertainties on these parameters are marginalized over all other parameters. The lower uncertainty is simply the best fit value of that parameter minus the 16th percentile of the parameter’s distribution for all bootstrap iterations. Similarly, the upper uncertainty is the 84th percentile of the bootstrap distribution minus the best fit value.
x, x_e-, x_e+, y, y_e-, y_e+
The x/y pixel position. This is left as a free parameter in the fit, so it may be slightly different than the LEGUS center.
mu_0, mu_0_e-, mu_0_e+
The central pixel value μ0, in units of electrons. Note that this is the peak pixel value of the raw profile before convolution with the PSF and rebinning (see Equation 8), so it may not be directly useful.
scale_radius_pixels, scale_radius_pixels_e-, scale_radius_pixels_e+
Scale radius a, in units of pixels.
axis_ratio, axis_ratio_e-, axis_ratio_e+
Axis ratio q, defined as the ratio of the minor to major axis, such that 0 < q ≤ 1.
position_angle, position_angle_e-, position_angle_e+
Position angle θ, in radians. Note that since the profile is assumed to be axisymmetric, 0 ≤ θ < π.
power_law_slope, power_law_slope_e-, power_law_slope_e+
Power law slope η.
local_background, local_background_e-, local_background_e+
Value of the local background, in units of electrons.
num_bootstrap_iterations
Number of bootstrap iterations done to calculate uncertainties on fit parameters.
Fit Quality Indicators
radius_fit_failure
Whether a given cluster is identified as having a failed radius fit. We define this as as a scale radius a < 0.1 pixels, a > 15 pixels, or an axis ratio q < 0.3. We also exclude any clusters where the fitted center is more than 2 pixels away from the central pixel identified by LEGUS.
fit_quality_metric
Our metric to evaluate the fit quality, defined in Equation 16. It uses the cumulative light profile to estimate the half-light radius of the cluster non-parametrically, then compares the enclosed light of the model and data within this radius. This value is the fractional uncertainty of the enclosed light of the model.
reliable_radius
Whether or not this cluster radius is deemed to be reliable. To be reliable, a cluster must not have a failed fit (see above), and must not be in the worst 10th percentile of fit_quality_metric among clusters with successful fits. See Section 2.6 for more on this. Our analysis in the paper only uses clusters deemed to be reliable.
reliable_mass
Whether or not we consider this cluster to have a reliable measurement of the mass. This relies on a consideration of the Q statistic (see Section 3.3 for more on this). For any analysis using masses or ages, we only consider clusters with reliable masses.
Effective Radius
Here we include the effective radius Reff (the projected half light radius) and its uncertainties. See Section 2.5 for more on how this is calculated. The uncertainties are marginalized over all other fit parameters. We calculate the effective radius of each bootstrap iteration, then use the percentiles to determine the upper and lower uncertainties. The lower uncertainty is the best fit Reff minus the 16th percentile Reff, while the upper uncertainty is the 84th percentile Reff minus the best fit Reff. The uncertainties for Reff in pixels and arcseconds only include the uncertainty in radius, while the uncertainties for Reff in parsecs also include the uncertainty in galaxy distance.
r_eff_pixels, r_eff_pixels_e-, r_eff_pixels_e+
The cluster effective radius in units of pixels.
r_eff_arcsec, r_eff_arcsec_e-, r_eff_arcsec_e+
The cluster effective radius in units of arcseconds. We provide this to make it easier for future users (i.e. you) to modify the galaxy distance estimates assumed in our paper if so desired.
r_eff_pc, r_eff_pc_e-, r_eff_pc_e+
The cluster effective radius in units of parsecs. The galaxy distances in this table were used to convert from arcseconds to parsecs. The uncertainty here includes the uncertainty in galaxy distance.
Derived Properties
We also calculate some quantities that use both the mass and radius. For the uncertainties on these quantities, we symmetrize both the mass and radius uncertainties, then propagate them analytically. These uncertainties include the uncertainty in galaxy distance. Note that the uncertainties are on the log space quantity, so the range given here should be interpreted as log10(quantity) ± quantity_log_err.
crossing_time_yr, crossing_time_log_err
The cluster crossing time, as defined by Gieles & Portegies Zwart 2011 (our equation 21).
density, density_log_err
The cluster average mass density within the half light radius ρh as defined by Equation 22, in units of M⊙pc-3.
surface_density, surface_density_log_err
The cluster average surface mass density within the half light radius Σh as defined by Equation 22, in units of M⊙pc-2.