Crushed Basalt Dissolves in the Topsoil. Getting the Carbon to the Ocean Is the Hard Part

A tractor spreads crushed basalt across a field, and a number appears in a ledger: so much carbon dioxide, removed. The chemistry behind that number is textbook. Rain and soil acid attack the rock, the reaction converts CO₂ dissolved in soil water into bicarbonate, and bicarbonate that reaches the sea holds its carbon there on a thousand-year timescale. The trouble is the word "reaches." Nearly every enhanced-weathering trial takes its measurements in the field. The removal it is claiming happens in the ocean.

That gap is what L. A. Derry of Cornell University, K. Maher of Stanford and O. A. Chadwick of the University of California, Santa Barbara, set out to size in a paper published in Nature on Aug. 26, 2026. They ran no trial. They went instead to the version of the experiment the planet has been running for millions of years: basaltic and intermediate volcanic rock covers about 6% of Earth's land surface, and geochemists have been sampling the rivers that drain it for decades. Compiling published stream chemistry from Iceland, Hawaii, Kamchatka, Guadeloupe, the Philippines and the basalts of the Pacific Northwest, the three asked what most field trials cannot: of the alkalinity that weathering generates, how much actually leaves the watershed?

Every dataset in the paper is previously published, and that is the point: the synthesis exists to give enhanced weathering a reference drawn from real landscapes. Its target is the way removal is currently counted. In most trials, soil is sampled before and after the rock goes down, to a depth of about 30 centimeters. The base cations that have gone missing (calcium, magnesium, sodium and potassium) are read as the fraction of basalt that dissolved, and that fraction is converted into a carbon-removal potential. Two assumptions ride underneath the conversion: that carbonic acid did all the dissolving, and that all the alkalinity generated reaches the ocean.

Volcanic rivers carry less than the ledgers expect

Across the compiled catchments, CO₂ uptake is modest. Median rates run from about 0.16 metric tons per hectare per year in Iceland to roughly 1.0 in the volcanically active Luzon and Bicol arcs of the Philippines, landscapes buried in fresh, fine-grained ash, soaked by tropical rain, and carrying rock loads that dwarf anything a tractor spreads. Some enhanced-weathering deployments claim a removal potential of 4 to 12 metric tons per hectare per year. And the authors think their own natural figures "may be biased high": hydrothermal fluids in active volcanic zones add dissolved solids that have nothing to do with surface weathering, and a river sampled only at low water reads richer than the same river sampled across a full year.

The same gap shows up globally. Derry, Maher and Chadwick report that global rates of natural basalt weathering run a factor of 8 to 30 below present estimates of enhanced weathering's carbon-removal potential. Carried forward at today's rate, the natural basaltic flux would come to between 1.0% and 4.7% of the CO₂ removal the IPCC estimates is needed by 2100: a statement about what volcanoes and rain already do, not a forecast of what rock dust on farms could add. The authors fence the comparison themselves: natural weathering systems differ from engineered deployments, and what these catchments supply is "an empirical baseline on watershed-scale alkalinity export."

Clay and calcite take some of it back

A watershed is not a pipe. Water leaving a treated field passes through soil, weathered rock, groundwater and finally a stream, and nearly every stage offers a reaction that undoes part of the work. Dissolving anorthite, a calcium feldspar common in basalt, consumes eight protons. If the aluminum and silicon it releases then precipitate as kaolinite (their ordinary fate in soil), six protons come straight back out. The calcium that appears in the water says nothing about which acid supplied those protons, and only carbonic acid represents carbon removed: fertilizer, nitrification, oxidizing minerals and exchangeable aluminum all donate protons too. Where secondary clays take up calcium and magnesium directly, more of the ledger goes missing. And where calcium and bicarbonate meet as calcite, half the alkalinity is fixed in a mineral and the other half is returned to the air as CO₂.

Independent studies point the same way, and the paper leans on them. Inverse modeling of Icelandic catchments puts 30% to 67% of the calcium released by silicate weathering into secondary minerals rather than into the river. On the artificial basalt hillslopes of the Landscape Evolution Observatory, irrigated on a controlled schedule, most of the dissolved calcium, magnesium and silicon is retained within days to weeks. In several river systems, calcium and strontium isotopes imply that carbonate precipitation removes a large share of the calcium: in the deep unsaturated zone, in groundwater and in riparian soils, rather than in the channel where models tend to place it. Discharge finishes the job: bicarbonate concentrations fall as rivers rise, so a watershed sampled at low flow reports an export it does not sustain.

What the trials themselves have measured

Enhanced weathering's own field trials give the most direct evidence in the paper. Across the small number of published field and mesocosm studies that measured alkalinity export rather than soil chemistry, basalt amendments of roughly 100 to 200 metric tons per hectare, an enormous dose, produced inferred export equivalent to no more than 0.15 metric tons of CO₂ per hectare per year, most estimates below 0.1 and "frequently statistically indistinguishable from zero." The authors list candidate explanations rather than settle on one: heavy fertilizer use masking the signal, weathering driven by acids from nitrification, acidic exchange sites in the soil and secondary minerals. Most of those studies looked no deeper than 2 meters and ran for a short time.

None of that adds up to a verdict on the technique, and the authors do not offer one. Their strongest sentence about enhanced weathering is that present estimates "may overestimate" how much of the alkalinity escapes the landscape and becomes durable removal. Their recommendation is a change of method: evaluate the idea from source to sink, and settle the question with longer-running experiments that monitor at depth and close a mass balance across an entire watershed. Two of the three argued a version of this in a 2025 paper, and Maher co-authored a verification framework this year; they are participants in the argument, not referees of it. What they have supplied is a reference for the leg between the field and the sea. No enhanced-weathering trial has yet been run that way.