40,000 Samples Later: What Mid-infrared spectroscopy is doing for soil carbon measurement

This week, EarthOptics presented new validation results for mid-infrared spectroscopy (MIR) at CANVAS 2026, the ASA, CSSA, and SSSA International Annual Meeting, at the Oregon Convention Center. CTO Patrick Schwientek presented "Commercial High-Fidelity Mid-Infrared Spectroscopy for Lower-Cost Soil Carbon Stock Determination in Carbon Credit Projects" in the Nutrient Management and Soil & Plant Analysis session on Tuesday, November 3, and lab soil scientist Thor Harkin presented "High-Throughput Mid-Infrared Spectroscopy for Cost-Effective, High-Fidelity Soil Carbon Stock Quantification" in the Wednesday poster session. The work draws on 40,000 soil samples measured by both MIR and dry combustion across 25 states and 900+ fields and ranches.

MIR measures soil carbon by recording how dried soil absorbs infrared light, then converting the spectrum to a carbon value with a calibrated model. It costs about half as much as dry combustion and returns total, organic, and inorganic carbon, plus sand, silt, and clay content, from a single scan. "Over the past year we have seen significant uptake of our MIR offering, to the point where we now measure MIR on the majority of carbon samples coming through our labs," said Patrick Schwientek, CTO of EarthOptics.

"Dry combustion is a well-controlled reference method, and MIR adds an independent second measurement on the same sample," said Schwientek. "When two independent methods agree, verifiers get an additional data set behind every result, which gives the project a stronger evidence base behind the credits. We have the paired data to show how well that works, and we are putting it in front of the scientific community this week."

Registry pathways for spectroscopic methods are available. Verra's VM0042 methodology accepts spectroscopic carbon measurement with documented calibration against dry combustion. EarthOptics is in conversation with other registries, including the Climate Action Reserve, about how spectroscopic methods fit their soil protocols.

What the data show

MIR, calibrated per project against dry combustion on 20 percent of samples, produces field-level organic carbon averages within 7 to 9 percent of dry combustion with no systematic bias. Total and inorganic carbon agree even more closely. EarthOptics presented on MIR at the same meeting in 2025; its September 2025 white paper validated the method on 560 paired samples, and this year's dataset is more than 50 times larger.

One scan costs the lab about half of dry combustion. For a carbon program, that means dry combustion on 25 percent of samples and MIR on all of them; lab costs are 25 percent lower than running dry combustion on every sample, and, because field work is most of what a program pays, the total program cost is 4 to 8 percent lower. The bigger gain is what the savings buy: more cores for the same budget.

What MIR adds beyond the carbon number

Texture on every sample. Each scan returns sand, silt, and clay alongside carbon. Clay, the fraction most closely tied to carbon storage, bulk density, and water holding, comes within 10 percentage points of the laboratory value on three samples in four, with no systematic offset. A measured clay value at every sampled point can replace the soil-survey map unit as input to carbon models and stratification, reducing the uncertainty those models carry forward.

A sample that stays intact. The scan is non-destructive. The same subsample can go on to dry combustion, fractionation, or the archive, and the spectrum itself is stored. When the calibration model improves, or when a model for a new analyte becomes available, archived spectra can be re-analyzed for updated or additional predictions without returning to the field or reprocessing the soil.

The model behind the number

EarthOptics' MIR predictions build on the Open Soil Spectral Library (OSSL), an open resource developed by the Soil Spectroscopy for Global Good initiative of the Woodwell Climate Research Center, OpenGeoHub and the University of Florida. EarthOptics calibrates the OSSL model to its own dry-combustion results, and its paired dataset, among the largest in the industry, provides modest improvements over this generalized library, optimizing accuracy for carbon projects on rangeland and cropland.

"Measurement cost has limited how densely carbon programs can sample," said Adriel Hsu-Flanders, VP of Product at EarthOptics. "MIR gives programs a second independent measurement, texture on every sample, and a path to more cores for the same budget. We are deploying it today on Scope 3 and insetting programs, alongside dry combustion on registry projects, and Verra already accepts it as an alternative method under VM0042. We are in conversation with other registries, and working with customers to bring the first MIR-measured project through a registry."

Where to go next

Our summary white paper sets out the method, the paired-sample results, and their limits in more detail. For a side-by-side pilot on a live project, or the paired-sample data, contact us at www.earthoptics.com.