Characterizing Supernova Host Galaxies with FrankenBlast: A Scalable Tool for Transient Host Galaxy Association, Photometry, and Stellar Population Modeling

The Big Picture

Imagine trying to understand a person’s life story only from a photograph of their neighborhood. You can’t speak with them, can’t read their diary, but the street they grew up on, the buildings around them, the economic activity visible from above, all whisper clues about who they might be.

Astronomers face a strikingly similar challenge with supernovae. These cosmic explosions are often too distant, too fleeting, or too numerous to study directly. But their neighborhoods, the galaxies that host them, encode deep clues about what kind of star exploded and why.

The coming decade will sharpen this challenge. The Vera C. Rubin Observatory’s Legacy Survey of Space and Time (LSST) is expected to discover millions of short-lived astronomical events, or transients. The vast majority will never receive spectroscopic follow-up, the detailed light analysis that reveals an explosion’s chemical fingerprint. Spectroscopy remains the gold standard for classifying supernova types, but there won’t be enough telescope time to go around.

So astronomers urgently need scalable tools to squeeze insight from host galaxies alone. That’s the problem FrankenBlast was built to solve. A team led by Anya Nugent at the Center for Astrophysics | Harvard & Smithsonian, including researchers from IAIFI, MIT, and Princeton, has shown that FrankenBlast can characterize the stellar environments of over 9,000 supernovae in a fraction of the time older methods required. Along the way, the analysis revealed surprising new clues about what drives different types of stellar explosions.

Key Insight: The neighborhoods where supernovae happen aren’t just background scenery — they’re diagnostic fingerprints. FrankenBlast makes reading those fingerprints fast, scalable, and publicly available for the era of big-survey astronomy.

How It Works

FrankenBlast builds on an existing platform called Blast (Broader Impact Astro Science Tool), extending it with new capabilities and a pipeline built for industrial-scale throughput. The workflow has three main stages:

1. Host association — Given a supernova’s sky coordinates, FrankenBlast identifies which galaxy it belongs to. It uses a fractional-flux technique, measuring what fraction of the galaxy’s total light comes from the region around the explosion, to pinpoint where within the galaxy the supernova occurred.
2. Photometry — The tool measures the host galaxy’s brightness across multiple wavelengths of light, from ultraviolet to infrared, assembling a multi-color portrait of its total light output.
3. SED fitting — Short for spectral energy distribution fitting, this step compares those brightness measurements against models of stellar populations. Using a code called Prospector, it applies Bayesian inference (a statistical method that finds the most probable answer given the data while rigorously quantifying uncertainty) to infer the galaxy’s physical properties.

What sets FrankenBlast apart is speed. Traditional Bayesian SED fitting can take hours per galaxy. FrankenBlast completes the full pipeline in minutes per object, which matters when you’re processing tens of thousands of events. The team validated FrankenBlast on 14,432 supernovae, roughly half spectroscopically classified and half classified only from their light curves (photometrically), and successfully constrained host properties for 9,262 of those events.

The three properties FrankenBlast pins down form a kind of environmental fingerprint:

• Stellar mass (M*) — the total mass of all stars in the galaxy, indicating how large and evolved it is
• Specific star formation rate (sSFR) — how actively a galaxy is forming new stars relative to its existing stellar mass; high sSFR means a young, churning star factory
• Host dust extinction (A_V) — how much interstellar dust dims and obscures the light

Together, these reveal whether a supernova went off in a young, active star-forming region or a quiet, aging stellar graveyard.

Why It Matters

When the team compared host properties between spectroscopically- and photometrically-classified supernovae of the same type, they found systematic differences. You might assume these reflect real physical differences, but FrankenBlast’s analysis points elsewhere. The deviations stem primarily from contamination: misclassified events sneaking into the photometric sample.

This matters for anyone building classifiers for future surveys. If you’re using host galaxy properties as training features, you need to know whether your training set has been corrupted. FrankenBlast provides a systematic way to diagnose exactly that.

Higher redshifts, meaning the supernovae are farther away and seen as they were earlier in cosmic time, independently drive shifts in the inferred M* and sSFR distributions in photometric samples. Galaxies evolve: those in the early universe look different from nearby ones, and any classifier has to account for this.

The comparisons between spectroscopically-confirmed supernova types yielded the paper’s most intriguing result. All four supernova classes (Type Ia, Ib/c, II, and IIn) show dependence on both stellar mass and star formation rate. But the balance differs:

• Type II and IIn supernovae lean strongly toward star-forming environments, consistent with their massive-star progenitors dying young.
• Type Ia supernovae are more strongly tied to high stellar mass, reflecting their long delay times (the billions of years between when a star system forms and when it finally explodes) and tendency to appear in older, more massive galaxies.
• Type Ib/c supernovae occupy a curious middle ground: more massive and evolved environments than Type II hosts. This suggests their stripped-envelope progenitors, stars that shed their outer layers before exploding, require specific conditions. Perhaps higher metallicity (the abundance of heavy elements built up over generations of stellar evolution) or binary star interactions that only develop in more evolved stellar populations.

Host galaxy properties are increasingly fed into photometric classifiers as features, teaching algorithms to recognize supernova types from galaxy context alone. The quality of that training signal depends entirely on how accurately host properties can be measured at scale.

For fundamental physics, the stakes are even higher. Type Ia supernovae are standard candles, calibrated cosmic distance markers that helped reveal dark energy. A well-documented pattern called the mass step, where Type Ia supernovae in more massive galaxies appear systematically brighter, must be carefully corrected in any measurement of the universe’s expansion history. Systematic errors in characterizing host stellar mass or star formation propagate directly into those cosmological measurements. FrankenBlast’s ability to constrain host properties for thousands of events, quickly and consistently, could sharpen these results.

The team is already developing a version tailored to Rubin Observatory data products. When LSST begins streaming millions of transient alerts, pipelines like FrankenBlast will be frontline infrastructure, linking each explosion to its galactic environment in real time and at scale.

Bottom Line: FrankenBlast turns the slow, expensive task of characterizing supernova host galaxies into a fast, automated pipeline. Testing it on over 14,000 supernovae, the team uncovered why photometric classification samples differ from spectroscopic ones and found new clues about the environmental conditions required for stripped-envelope supernovae.