Cosmic Rain: Colorado Soils Reveal How Millions of Beryllium-10 Atoms Fall Each Year

Every second, high-energy cosmic rays slam into the atmosphere, shattering nitrogen and oxygen atoms and forging rare isotopes that rain quietly onto the landscape below. Among them is beryllium-10, a radioactive atom that clings stubbornly to soil particles once it reaches the ground. For decades, geologists have exploited this cosmic fallout as a clock and a tracer, using it to date landforms, measure erosion, and follow sediment through river basins. But the method has an Achilles heel: nobody knows precisely how fast beryllium-10 falls at any given place. A new study of glacial soils in the Colorado Rockies now delivers one of the most detailed local calibrations ever attempted, revealing that millions of these atoms blanket every square centimeter of mountain terrain each year, and that the rate shifts dramatically with elevation, snowfall, and even the chemistry of the soil itself.

The research, published in the journal Geochronology, was led by José M. Marmolejo-Cossío and Isaac J. Larsen of the University of Massachusetts Amherst, together with Eyal Marder and Alan J. Hidy of Lawrence Livermore National Laboratory. The team focused on the East River watershed near Crested Butte, Colorado, a steep alpine catchment draining the Elk Range of the Rocky Mountains and feeding the Gunnison River, a major tributary of the Colorado. Elevations in the watershed span roughly 2,600 to 4,100 meters, mean annual temperature hovers near zero degrees Celsius, and most precipitation arrives as snow that can pile up several meters deep. Crucially, the bedrock is dominated by Cretaceous Mancos Shale, a quartz-poor rock that thwarts the standard method of measuring erosion with beryllium-10 produced inside mineral grains. That limitation makes the watershed an ideal proving ground for the meteoric variety of the isotope, which can be measured in almost any soil regardless of quartz content.

The researchers’ strategy hinged on a gift from the last ice age. Between roughly 18,000 and 13,000 years ago, Pleistocene glaciers carved cirques into the high peaks and dumped moraines, ridges of glacial debris, across the valley floors. Because previous work had independently dated five of these moraines using in situ-produced beryllium-10 in boulders, the team knew exactly how long each land surface had been exposed to the sky. By digging soil pits into the flattest crests of the moraines, sampling in depth increments down to 145 centimeters, and measuring the total inventory of meteoric beryllium-10 in each profile, they could divide the accumulated atoms by the landform age to recover a millennial-averaged delivery rate. The approach assumes the clock started when the ice retreated, and the team carefully corrected for inherited atoms that had stuck to mineral grains before the sediments were deposited.

The laboratory work was formidable. Roughly half a gram of powdered soil from each depth interval was fused with fluxes in platinum crucibles, spiked with a known quantity of stable beryllium-9, and chemically purified through repeated dissolution and precipitation steps. The resulting beryllium oxide was packed into cathodes and analyzed by accelerator mass spectrometry at Lawrence Livermore, where 10Be/9Be ratios on the order of 10^-12 were measured against calibrated standards, with process blanks several orders of magnitude lower. The team also measured soil pH in water and calcium chloride solutions, because beryllium retention in soils depends strongly on acidity: soils with pH above roughly 4 hold onto the isotope, while more acidic profiles can leach it away. Natural beryllium-9 in the samples turned out to be negligible compared with the added carrier, simplifying the interpretation.

The results are striking. Inheritance-corrected inventories ranged from 7.68 billion to 23.5 billion atoms per square centimeter, and the corresponding fluxes ranged from about 430,000 to 1.8 million atoms per square centimeter per year across the five sites. When the researchers additionally modeled losses from erosion and from desorption, the chemical detachment of beryllium from mineral surfaces and its downward migration in soil water, corrected fluxes spanned 310,000 to 3.7 million atoms per square centimeter per year, deviating from the uncorrected values by factors of 0.3 to 2.1. Desorption losses were predicted to be greatest at the highest, most acidic Copper Creek site, where modeled losses ran an order of magnitude above those at the other moraines, while erosional losses peaked at the most topographically convex Gothic moraine.

Perhaps the most consequential finding is how tightly the flux tracks the mountain environment. Inheritance-corrected fluxes correlated with elevation with an R-squared of 0.91, with mean annual precipitation at 0.82 to 0.85, with mean snow depth at 0.95 to 0.97, and with snow water equivalent at 0.97 to 0.98. The reason is orography: within the East River watershed, precipitation increases with elevation as moist air is forced upslope, so the cosmic isotope rains down faster on the high country than on the valley floor. By applying these regression equations to high-resolution gridded datasets, including 3-meter LiDAR elevation models, downscaled PRISM precipitation data, and Airborne Snow Observatory maps of snow depth and snow water equivalent, the team produced the first spatially resolved maps of beryllium-10 flux for an entire mountain watershed, with watershed-averaged predictions of 1.1 to 3.8 million atoms per square centimeter per year depending on the predictor chosen.

The snow-based maps revealed a subtle topographic fingerprint that precipitation averages alone would miss. Because wind and avalanches strip snow from sharp ridge crests and pile it on adjacent slopes, the models predict lower beryllium-10 accumulation on ridgelines than on nearby hillsides. Yet the snow data also carry risks: they cover only two winters, a vanishingly small slice of the roughly 15,000 years over which the soils have collected the isotope, and some of the deepest snow measurements in the watershed, exceeding 20 meters, are likely measurement artifacts. Elevation, by contrast, is the only predictor immune to temporal bias, though it too can mislead where local topography redistributes snow. The authors suggest that, much like exposure-age analysts who average results from multiple production-rate scaling models, future users of meteoric beryllium-10 may need to adopt an ensemble approach that weighs all of these predictors.

The comparison with existing models was sobering. A widely used empirical model that predicts flux as a function of precipitation and latitude overestimated the measured fluxes by factors of 1.8 to 3.8 relative to inheritance-corrected values, and by 1.0 to 6.0 relative to erosion- and desorption-corrected values. Global circulation models, which simulate atmospheric production and transport of the isotope, fared better at the watershed scale: their predictions of roughly 1.46 to 2.12 million atoms per square centimeter per year generally fell within the range of the locally calibrated watershed averages. But a single grid cell in these global models covers an area larger than the entire watershed, so they completely miss the within-basin variability driven by orographic precipitation. At the site scale, the highest model-based prediction exceeded the lowest corrected field flux by a factor of 7 to 13.

The East River fluxes also fit into a broader regional picture. Soil-inventory-based estimates from the wetter Colorado Front Range and the drier Wind River Mountains of Wyoming cluster around 1.3 to 1.5 million atoms per square centimeter per year, and the East River values of 0.43 to 1.8 million are broadly consistent, with the driest site in the study falling below the regional norms. One complication the authors flag is dust: windblown particles carry beryllium-10 that was scavenged elsewhere, potentially contributing up to about 20 percent of the total flux in some settings. In the Colorado Front Range, dust is thought to contribute less than 10 percent, and if East River conditions are similar, the discrepancy with global models would not change substantially, though the dust contribution there remains unconstrained.

The practical payoff could be considerable. Because meteoric beryllium-10 can be measured in quartz-poor shales and paired with stable beryllium-9 to estimate catchment-wide denudation and chemical weathering, a locally calibrated flux transforms the East River watershed, already an intensive Department of Energy research site for watershed science, into a place where erosion and weathering rates can be quantified across shale terrain that the standard method cannot touch. More broadly, the study demonstrates a template for other alpine landscapes: combine independently dated landforms, careful soil chemistry, and modern high-resolution LiDAR and snowpack datasets to turn a global cosmic drizzle into a locally predictable tool. The atoms fall from the sky whether we measure them or not; this work shows how to read them accurately.

Subject of Research: Measurement of meteoric beryllium-10 deposition fluxes from soil inventories on glacial moraines in the East River watershed, Colorado

Article Title: Meteoric beryllium-10 fluxes from soil inventory measurements in the East River watershed, Colorado, USA

Article References: Marmolejo-Cossío, J. M., Larsen, I. J., Marder, E., & Hidy, A. J. (2026). Meteoric beryllium-10 fluxes from soil inventory measurements in the East River watershed, Colorado, USA. Geochronology, 8(3), 547-565. https://doi.org/10.5194/gchron-8-547-2026

Image Credits: AI Generated

DOI: 10.5194/gchron-8-547-2026

Keywords: beryllium-10, cosmogenic nuclides, glacial moraines, soil inventories, erosion rates, East River watershed, Colorado, orographic precipitation, geochronology, accelerator mass spectrometry, snow water equivalent, watershed science