MIRION Catalog: Mid-IR Compact Nebulae

• MIRION Catalog is a curated database of 6176 compact mid-IR sources (yellowballs) that trace photodissociation regions around young stellar objects.
• It integrates multi-wavelength photometry, kinematic data, and Bayesian distances across five linked tables to study the transition from isolated to clustered star formation.
• The catalog employs systematic aperture photometry with repeated measurements to improve measurement reliability and refine physical parameters.

Searching arXiv for the specified MIRION catalog paper and closely related yellowball work to ground the article. The MIRION Catalog—Mid-InfraRed Interstellar Objects and Nebulae—is a curated, multi-table catalog built from 6176 compact mid-infrared sources originally identified as “yellowballs” by citizen scientists in the Milky Way Project. In the historical usage retained by the catalog, yellowballs are compact, roughly circular patches in Spitzer three-color images that appear yellow because 8 μm polycyclic aromatic hydrocarbon emission and 24 μm warm-dust emission overlap within a small area. Physically, most MIRION sources are young, compact photodissociation regions surrounding intermediate- and high-mass young stellar objects and compact clusters, although the catalog explicitly treats them as a heterogeneous class of mid-IR bright, compact nebulae associated predominantly with star-forming regions. By compiling photometry, velocities, distances, clump properties, and catalog cross-identifications into five linked tables, MIRION provides an extensive database for studying the transition from isolated low-mass to clustered high-mass star formation and increases the number of candidate intermediate-mass star-forming regions by nearly two orders of magnitude (Devine et al., 5 Dec 2025).

1. Origin, nomenclature, and astrophysical meaning

MIRION sources originate in the Milky Way Project (MWP), where participants identified compact objects now historically known as yellowballs. The designation comes from the appearance of these sources in Spitzer three-color images: they look yellow because green 8 μm emission and red 24 μm emission are spatially coincident over a compact area. In physical terms, the 8 μm band traces UV-excited polycyclic aromatic hydrocarbons in the photodissociation region, while the 24 μm band traces warm dust close to embedded sources (Devine et al., 5 Dec 2025).

Most yellowballs are interpreted as compact photodissociation regions generated by intermediate- and high-mass young stellar objects and compact clusters. Many appear to be precursors to H II regions, including UC H II and compact H II regions, and the population includes a mix of intermediate-mass star-forming regions (IMSFRs; producing stars up to 3–8–10 $M_\odot$) and massive star-forming regions. Because the associated clumps span a wide dynamic range in mass and luminosity, MIRION sources trace the transition between isolated low-mass and clustered high-mass star formation. A plausible implication is that the catalog is especially useful for population-level studies of the poorly sampled regime between classical isolated low-mass cores and established massive H II-region environments.

The sky coverage reflects the underlying infrared survey footprint from which the original identifications were drawn. Sources were identified in Spitzer surveys covering the inner Galactic plane $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$, the Cygnus-X complex $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$, and a portion of the outer Galaxy $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$. All photometry was re-measured in this work at 8, 12, 24, and 70 μm across these regions (Devine et al., 5 Dec 2025).

2. Catalog architecture and linked data products

MIRION is organized as five linked tables, each indexed by a unique Source ID, so that joined queries can be performed across photometry, kinematics, distances, physical clump properties, and external catalog associations. This structure is central to the catalog’s design because it allows filtering by distance, environment, evolutionary indicators, or association with other star-formation tracers.

Table	Core contents	Purpose
MIRION Catalog — Photometry	Positions, source size, hit rate, $F_8$, $F_{12}$, $F_{24}$, $F_{70}$, errors, measurement counts, quality flags	Mid- and far-IR fluxes and photometric QA
MIRION Catalog — Velocities	Adopted $V_{\rm LSR}$, uncertainty, spectral type, survey-specific velocities, dense-gas tracer information	Molecular-cloud kinematics
MIRION Catalog — Distances	Adopted distance, uncertainty, external distances, $P_{\rm far}$, posterior distance solutions, spiral-arm assignments	Bayesian distance inference
MIRION Catalog — Herschel-Matched Sources	Diameter, mass, luminosity, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$0, temperature, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$1, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$2, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$3	Physical properties of associated clumps
MIRION Catalog — Catalog Cross-matches	Matched identifiers from Hi-GAL, ATLASGAL, CORNISH, RMS, and WISE H II catalogs	Environmental and source-type context

The Photometry table includes Galactic longitude GLON, latitude GLAT, MWP radius MWPR, their uncertainties, and the Milky Way Project hit rate HRATE. It also records flux densities and errors in Jy for $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$4, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$5, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$6, and $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$7; the number of measurements per band ($$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$8, $$(|\ell| \lesssim 65^\circ;\ \mathrm{GLIMPSE}+\mathrm{MIPSGAL})$$9, $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$0, $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$1); and several flags: $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$2 for per-band saturation, $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$3 for multiple compact sources in the 8 μm aperture, $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$4 for bands with no obvious source, $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$5 for poor confidence in photometry, and $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$6 for highly circular extended objects likely to be planetary nebulae or compact bubbles (Devine et al., 5 Dec 2025).

The Velocities table stores an adopted $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$7 with uncertainty $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$8 and a spectral classification SType. Its categories are SP for single peak, MP:n for multi-peak spectra, LSN for low signal-to-noise, and OUT for sources outside CO coverage. Survey-specific velocities are preserved for GRS, SEDIGISM, FCRAO OGS/CGPS, ThrUMMS, a low-resolution 1.2 m CO archive, and dense-gas tracers, along with external comparisons from Elia (2021) and Mége et al. (2021) (Devine et al., 5 Dec 2025).

The Distances table provides the adopted distance DIST, its uncertainty, external distance estimates, and the outputs of the Reid et al. (2019) Parallax-Based Distance Calculator v2: $$(\sim 24\ \mathrm{deg}^2\ \mathrm{centered\ on}\ \ell \approx 79.3^\circ,\ b \approx 1^\circ)$$9, two posterior distance solutions, their integrated probabilities, and spiral-arm labels. The Herschel-Matched Sources table holds physical parameters derived from Hi-GAL compact-source crossmatches, after rescaling distance-dependent quantities to the adopted MIRION distances. Finally, the Cross-matches table records positional associations within 24″ to Hi-GAL, ATLASGAL, CORNISH, RMS, and the WISE H II catalog (Devine et al., 5 Dec 2025).

3. Photometric construction and quality control

The catalog’s photometric measurements were made interactively and aperture-based, with explicit attention to the complex backgrounds of the Galactic plane. For each source and band, users selected at least three vertices to define a polygonal mask around the object in an image regridded to the 8 μm pixel scale for consistency. The local background was then estimated by multiquadratic radial basis function interpolation over the masked region. A “source-only” image was produced by subtracting the background model, and the flux density was summed over the mask (Devine et al., 5 Dec 2025).

The procedure was designed to quantify user-selection uncertainty rather than suppress it. Each source in each band received at least five independent measurements, and the cataloged $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$0 values are fractional errors defined as the standard deviation divided by the mean flux. In practice, the measurement statistics reported for sources with fractional uncertainty below 0.5 are 4669 at 8 μm, 4176 at 12 μm, 4545 at 24 μm, and 4757 at 70 μm. This explicit multiplicity of measurements is one of MIRION’s distinguishing methodological features because it treats aperture selection uncertainty as part of the measurement model rather than as an unreported subjective choice (Devine et al., 5 Dec 2025).

The input imaging products were drawn from multiple facilities: Spitzer/IRAC for 8 μm, WISE W3 Atlas coadds for 12 μm, Spitzer/MIPS for 24 μm, and Herschel/PACS Highly Processed Data Products for 70 μm. For the clump physical properties, multi-band Herschel data from Hi-GAL were used by Elia et al., whereas MIRION’s own 70 μm photometry was measured directly by the catalog team. A practical implication is that the catalog separates direct MIRION flux measurement from inherited SED-based clump characterization, which is important when combining photometric and physical-property analyses (Devine et al., 5 Dec 2025).

Color measurements are expressed either as logarithmic flux ratios or as magnitude-like color indices. For bands $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$1 and $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$2, the color is

$$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$3

The propagated color uncertainty from independent flux errors is

$$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$4

MIRION commonly reports colors as logarithmic flux ratios such as $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$5 in order to emphasize physical trends in compactness and polycyclic aromatic hydrocarbon excitation (Devine et al., 5 Dec 2025).

4. Velocities, Bayesian distances, and environmental associations

The kinematic component of MIRION is built from molecular-line data and dense-gas or star-formation tracers. The adopted velocity for each source was determined from 48″ spectral-cube cutouts centered on the MIRION position, a size chosen to encompass typical MIRION scales while minimizing unrelated emission. The default rule was to adopt the Gaussian-fit velocity of the most intense CO peak above a noise floor defined as $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$6 over approximately 30 channels near a spectrum end. Minor peaks had to be < 0.2 $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$7 to count as part of a single major peak; spectra with a secondary peak > 0.8 $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$8 were flagged MP:n. If a dense-gas-tracer peak overlapped the FWHM of a CO secondary peak, the dense-gas-tracer-associated velocity was preferentially adopted when the dense-gas-tracer peak had SNR $$(\ell \approx 102^\circ\text{–}109^\circ,\ b \approx 0^\circ\text{–}3^\circ;\ \mathrm{SMOG})$$9 (Devine et al., 5 Dec 2025).

The velocity inputs came from several surveys. High-resolution CO data were drawn from GRS, SEDIGISM, ThrUMMS, and FCRAO OGS/CGPS; low-resolution CO came from a 1.2 m archive; and dense-gas/star-formation tracers included THOR OH masers, SPLASH OH masers, and HOPS H$F_8$0O masers and NH$F_8$1(1,1). In survey-overlap regions, GRS was preferred over SEDIGISM because of finer spectral resolution; ambiguities were resolved by comparing spectral flags and, if necessary, the integrated distance probability in the distance solution (Devine et al., 5 Dec 2025).

The catalog reports the origins of adopted velocities as follows: GRS (1830; 29.5%), SEDIGISM (2612; 42.3%), FCRAO/CGPS (240; 3.9%), ThrUMMS (597; 9.7%), low-res CO (570; 9.2%), and Mége et al. (327; 5.3%). The corresponding spectral-type distribution is SP (4085; 66.1%), LSN (849; 13.8%), OUT (416; 6.7%), and MP (826; 13.4%; 651 have one additional peak). Dense-gas-tracer velocities overrode CO in 107 cases, or approximately 2% (Devine et al., 5 Dec 2025).

Distances were then computed using the Reid et al. (2019) Parallax-Based Distance Calculator v2, which combines $F_8$2, the adopted $F_8$3, spiral-arm modeling, and a prior $F_8$4 to produce posterior distance estimates and arm assignments. Near-far ambiguity was handled probabilistically through $F_8$5. When a matched Hi-GAL source had $F_8$6, MIRION set $F_8$7; when MDIST or SDIST_GROUP was available from Mége et al. (2021), the non-kinematic Mége distance was adopted directly; for NO_KDA, KDA_NO, TGT_POINT, or missing status, $F_8$8 was used (Devine et al., 5 Dec 2025).

Distances were determined for approximately 94% of MIRION sources. Compared with the subset criteria from Mége et al. (2011), the average distance difference is approximately 0.3 kpc, and 61% agree within 0.7 kpc. Velocity comparisons agree within 5 km s$F_8$9 for approximately 80%. This suggests that the catalog’s kinematic backbone is sufficiently homogeneous for large-scale statistical work, while still requiring caution in complex or low-S/N sightlines (Devine et al., 5 Dec 2025).

Crossmatching further anchors MIRION in the broader Galactic star-formation literature. Within survey coverage, the match fractions are 3979 (65.26%) to Hi-GAL CSC-360, 1623 (30.48%) to ATLASGAL Condensations, 1563 (41.24%) to ATLASGAL CSC2014, 613 (11.03%) to RMS, 2315 (37.48%) to the WISE H II catalog, and 576 (12.08%) to CORNISH. The WISE total includes K = 435 (7.04%), C = 486 (7.87%), G = 113 (1.83%), and Q = 1276 (20.66%). Spurious-association tests using 100 simulated MIRION distributions yielded approximately 3.4% ± 0.2% random matches to Hi-GAL and < 1% to the other catalogs. The mean HRATE is approximately 0.43–0.48 for matched catalogs, compared with 0.32 for unmatched MIRION sources, consistent with many unmatched sources being fainter and/or below detection thresholds of other surveys (Devine et al., 5 Dec 2025).

5. Physical parameter space and the bridge between intermediate- and high-mass star formation

For MIRION sources with Hi-GAL compact-source matches and adopted distances, the catalog provides a physically characterized sample of 3945 objects after removing entries missing any physical property. The associated clumps span a wide range in mass and luminosity: masses extend roughly from 10 to $F_{12}$0, with most lying in $F_{12}$1, while bolometric luminosities span approximately 10–$F_{12}$2, with most in $F_{12}$3. Clump diameters are typical of “clumps”, around 0.2–3 pc (Devine et al., 5 Dec 2025).

The median physical values reported for the matched sample are $F_{12}$4 and $F_{12}$5. The table also includes $F_{12}$6, $F_{12}$7, $F_{12}$8, graybody temperature TEMP, and surface density $F_{12}$9, all showing broad ranges consistent with mixed evolutionary stages and mixed mass regimes. This broad parameter coverage is what allows MIRION to “bridge” intermediate- and high-mass star formation in an operational sense: it densely populates the 10–100 $F_{24}$0 clump regime that had previously been sparsely sampled (Devine et al., 5 Dec 2025).

Distance-dependent quantities in the Herschel-matched table were explicitly rescaled to the MIRION distances. In particular, luminosity and mass were rescaled according to

$F_{24}$1

and

$F_{24}$2

Diameter scales linearly with distance, and the context formula for dust/gas mass from sub-millimeter continuum is

$F_{24}$3

Because MIRION adopts the Hi-GAL SED-derived physical quantities and then rescales them using the adopted MIRION distances, the catalog is internally structured so that distance revisions propagate directly into the inferred physical state of the clumps (Devine et al., 5 Dec 2025).

The catalog also tabulates the fractions of the full Hi-GAL-matched MIRION sample meeting widely used threshold conditions: $F_{24}$4 for 22%, $F_{24}$5 for 35%, $F_{24}$6 for 20%, and $F_{24}$7 for 16%. In the MIRION interpretation, these thresholds respectively mark likely high-mass star-forming regions, more evolved star-forming regions, younger protostellar clumps, and high surface density likely associated with massive star-forming regions (Devine et al., 5 Dec 2025).

6. Infrared color diagnostics, candidate selection, and limitations

One of MIRION’s principal scientific results is that infrared colors trace physical conditions and evolutionary state. For 3116 sources with good 8 μm and 12 μm photometry $F_{24}$8, the mean color is $F_{24}$9. Subsamples show RMS at $F_{70}$0, WISE C/G/K at $F_{70}$1, and CORNISH at $F_{70}$2, all significantly below the typical WISE H II mean color of $F_{70}$3. WISE Q sources, at $F_{70}$4, and the “No Association” population, at $F_{70}$5, are even more negative, indicating particularly compact photodissociation regions and/or softer UV fields, consistent with IMSFRs (Devine et al., 5 Dec 2025).

A second diagnostic uses the color-color plane $F_{70}$6 versus $F_{70}$7. For 2819 sources with good photometry and without NOSRC or PCONF flags, MIRION sources associated with high-mass tracers—RMS, CORNISH, and WISE K/C/G—cluster in the H II-region color space defined by Anderson et al. (2012). Sources without high-mass tracers—especially WISE Q and unmatched MIRION sources—tend toward lower $F_{70}$8, consistent with softer UV fields in which polycyclic aromatic hydrocarbons are excited but not destroyed, while the average dust temperatures inferred from $F_{70}$9 remain similar (Devine et al., 5 Dec 2025).

The luminosity-mass diagrams strengthen that interpretation. RMS and WISE K sources lie largely above the 90th percentile lower limit for H II clumps from Elia (2017), whereas WISE Q and MIRION-only samples lie above pre-stellar upper limits but below the H II locus. Many of these have masses of 10–100 $V_{\rm LSR}$0 and would track toward $V_{\rm LSR}$1 in simple single-star clump models, consistent with IMSFRs. When uncertainties are included, 57–66% of the WISE Q + MIRION-only clumps have $V_{\rm LSR}$2, which the catalog interprets as strengthening their identification as IMSFR candidates (Devine et al., 5 Dec 2025).

The catalog therefore supports several operational sample definitions. High-mass star-forming region candidates can be selected using crossmatches to RMS types “H II region” or “H II/YSO”, CORNISH types UC H II/H II (Diffuse/Dark)/MYSO/IR Quiet, or WISE types K/C/G, together with thresholds such as $V_{\rm LSR}$3 or $V_{\rm LSR}$4 and colors typical of the Anderson H II box. IMSFR candidates are preferentially found among WISE Q and “No Association” MIRION sources, especially when they show compact PAH colors $V_{\rm LSR}$5, lower $V_{\rm LSR}$6, moderate clump masses 10–100 $V_{\rm LSR}$7, and an absence of radio or H II-region tracers. Very young protostellar clumps can be isolated with $V_{\rm LSR}$8 and lower $V_{\rm LSR}$9 (Devine et al., 5 Dec 2025).

The catalog documentation also emphasizes several limitations. MIRION sources are heterogeneous and include a small fraction of non-star-forming objects, for example possible planetary nebulae flagged $P_{\rm far}$0. The parent sample is tied to Spitzer coverage, so faint or extended objects may be underrepresented. Complex mid-infrared backgrounds, saturation, and subjective apertures are mitigated by multiple measurements and flags, but residual photometric systematics remain. In the inner Galaxy, the near-far ambiguity is handled probabilistically rather than deterministically, and distances near the Galactic Center have historically been uncertain. Crossmatch associations use a 24″ radius and are therefore probabilistic rather than guaranteed for every source. A small systematic positional offset was corrected for sources 3035–3101 $P_{\rm far}$1 (Devine et al., 5 Dec 2025).

MIRION is not static. The PERYSCOPE project—People Enabling Research: a Yellowball Survey of the Colors Of Protostellar Environments—provides a classroom-oriented workflow in which volunteers, typically introductory astronomy students, measure photometry for approximately 40 MIRION sources in Google Colab using a simplified version of the MIRION photometry tool. After quality checking against current catalog values and mask overlap, these measurements are periodically incorporated into the online catalog to refine mean fluxes and reduce fractional errors as the number of measurements grows. The latest online version is hosted at https://github.com/astrodevine/MIRION (Devine et al., 5 Dec 2025).