Reading atmospheric dust in three colors of starlight

Why gamma-ray astronomy needs a weather report

Gamma-ray observatories do not simply collect gamma rays through a lens. Imaging atmospheric Cherenkov telescopes detect brief flashes of visible light from particle showers triggered when high-energy gamma rays enter the atmosphere. Aerosols—airborne particles such as dust—attenuate that light, affecting estimates of the incoming radiation’s energy and brightness.

A preprint by Shefali Negi and colleagues, submitted to arXiv in November 2026, investigates how robotic telescopes can measure the wavelength dependence of that attenuation. This is an observational and instrument-analysis study, not a laboratory demonstration or a purely theoretical proposal. Its results suggest useful atmospheric information can be recovered, but an unexplained measurement offset remains central to the interpretation.

Measuring the dimming of stars

The researchers analyzed observations from 2021–2022 at three FRAM atmospheric-monitoring installations: one on La Palma in Spain and two at an intermediate location at Chile’s Paranal Observatory. The Chilean instruments sit about 1.3 kilometers apart.

FRAM compares measured stellar brightness with catalogued brightness while scanning from near the horizon toward overhead. Light arriving from lower elevations travels through more atmosphere. Fitting the resulting dimming, with modeled molecular extinction included, estimates the vertical aerosol optical depth: a dimensionless measure of aerosol attenuation for an overhead path.

Earlier processing used the blue B filter and the Tycho-2 star catalogue. The new analysis uses Gaia DR3 synthetic photometry, which supplies reference brightnesses in all three required bands: B, V and R, with effective wavelengths of approximately 440, 540 and 620 nanometers. Gaia also makes many more stars available. In five matched scans at one Chilean installation, the updated processing used roughly 25 times as many stars as the older approach.

A color comparison that initially went wrong

The wavelength dependence is summarized by the Ångström exponent. Larger positive values generally indicate stronger attenuation at shorter wavelengths, often associated with finer particles. Values near zero indicate weaker wavelength dependence, characteristic of coarse dust. This is an approximate description, not a unique measurement of particle size or composition.

For the aerosol conditions considered, the researchers expected the greatest optical depth in blue, followed by V and then R. Instead, the measurements consistently placed V highest, R next and B lowest. That ordering produced negative exponents for the B–V and B–R comparisons.

The team tested changes to the photometric aperture, assumed aerosol height distribution, detector-response correction and treatment of stellar image shapes. None resolved the ordering. Agreement between the blue measurements and earlier processing did not establish that blue was correct: errors could affect any or all of the filters.

An empirical correction, not an explanation

The authors fitted straight lines relating optical depths in different filters, accounting for uncertainty in both measurements. These lines had nonzero intercepts. Subtracting the fitted intercept from one filter’s optical depth before calculating the exponent produced more physically plausible results.

This correction assumes an approximately stable exponent within the population used for fitting. It would not work reliably if aerosol properties varied continuously without a stable inter-filter relationship. At La Palma, the team fitted low-aerosol observations separately for each year, then applied each correction to the corresponding year’s data.

Similar offsets across the installations suggest a shared instrumental or processing effect, but do not identify its cause. Recovering plausible exponents therefore does not establish that the absolute optical depths are correct.

Dust and cleaner air emerge

After correction, La Palma showed two prominent populations. The B–V exponent averaged about 0.18 during high-aerosol conditions associated with Saharan dust intrusions, and 1.33 in the low-aerosol background regime.

Nearby AERONET daytime sun-photometer measurements also showed coarse- and fine-particle populations. The comparison covered 300 calendar dates with usable observations from both instruments, but not simultaneous measurements: FRAM observes at night. The coarse-population means were close, about 0.18 for FRAM versus 0.21 for AERONET. The fine-population means differed substantially, about 1.33 versus 1.96. Different wavelength pairs and sampling further limit quantitative agreement.

The two Chilean installations showed single-peaked distributions, with corrected B–V means of approximately 0.84 and 0.60. No corresponding AERONET validation was available there.

What still limits the method

Processing choices contributed more uncertainty than statistical averaging in La Palma’s fine-aerosol regime: roughly 0.2 versus 0.05 for B–V. The closely spaced V and R bands were particularly sensitive to small measurement changes. Seasonal coverage was also uneven, with only 16 high-aerosol FRAM scans in the matched 2022 sample.

The study supports further development of three-color stellar monitoring, particularly using B–V and B–R. It does not yet demonstrate a fully calibrated solution: explaining the persistent filter offsets remains essential.