Rebecca J. Even, Megan B. Machmuller,
Jocelyn M. Lavallee, Tamara J. Zelikova,
& M. Francesca Cotrufo.
White Paper by: Margaret Morgan

Guiding methodologies for soil processing and analyses to support rigorous soil carbon accounting

For a detailed description of the research methods and results, please see our peer reviewed paper in Soil: Even et al., 2025

Interest in agricultural soil carbon (C) and soil health is rapidly gaining traction with producers, corporate sustainability offices, policy makers, C market entities, civil society, and various companies interested in developing and scaling climate solutions. The efficacy of scaling soil C solutions as a climate mitigation strategy relies on the accuracy with which soil organic carbon (SOC) stock changes can be quantified. This is especially important now given incentives to offset greenhouse gas emissions through purchasing soil C credits. Today, rigorous quantification approaches rely on physical soil sampling, either to quantify SOC directly or to calibrate and/or verify indirect measurements (using modeling and remote sensing approaches). The methodology used to process samples can vary substantially between individual labs, with implications for final soil C stocks. With that in mind, our project aimed to quantify the impacts of differences in soil laboratory processing approaches in order to develop general guidance for best practices and improve the rigor of and trust in soil carbon quantification.

THE WAY IN WHICH WE PROCESS SOIL MATTERS.

There is no perfect way, but there are better ways.

Why focus on lab processing procedures?

For reproducibility and comparability of SOC stocks across spatial and temporal scales, sample preparation is considered the first and one of the most important quantification steps. The way soil samples are collected and processed can impact final SOC stock numbers, but there are currently no widely agreed upon guidelines to ensure consistency in how samples are collected in the field or processed in the lab. Today, soil pre-processing protocols and quantification of SOC stocks vary across labs, casting doubt on the quality of measurements they yield.

For example, many commercial testing labs process soils initially using a mechanical grinder instead of hand sieving soil samples prior to running soil analyses. But for soil carbon and nitrogen stock quantification, it is deemed important to remove rocks and roots and account for them in calculations of bulk density (BD). A tiny amount of root in a soil sample can increase SOC stocks while grinding rocks with the bulk sample dilutes SOC. In addition, not considering the impacts of roots and rocks on soil bulk soil mass can bias BD determinations. Even research laboratories that sieve soils use different approaches, both in terms of mesh sizes and how they break up soil aggregates. 

Most stakeholders and data users lack a basic understanding of the different forms of carbon labs measure, such as total (TC), soil inorganic (SIC) or soil organic (SOC). Labs implement different approaches to check for and quantify SIC in order to account for it in TC or remove it to directly measure SOC. These same data from laboratory analyses underpin the implementation of policy programs and the calibration and verification of soil biogeochemical models used to predict how changes in management affect SOC dynamics, making it more imperative that the best approaches are used to obtain accurate and precise SOC data across soil testing laboratories worldwide.

The need for better soil carbon quantification

A range of stakeholders rely on accurate SOC assessments, from start-ups that need to test the accuracy of new tools against high quality lab data for commercialization to modeling platforms that need to benchmark against accurate datasets. Emerging markets need to be built on rigorous C accounting and discrepancies in accounting may undermine climate benefits and goals. Policy makers need to trust the quality of the data that underpin policies for climate change mitigation. Reducing variability and developing clear guidance on soil processing best practices can improve MMRV approaches and credibility of incentive programs, corporate sustainability efforts, and overall climate progress.

While research laboratories use relatively common protocols to process and analyze soils, there are variations of existing procedures that affect soil carbon quantification and there is currently no agreed upon guidance for soil processing that helps ensure consistency.

Glossary + Common Terms

  • Chemical element

  • Soil inorganic carbon consists of mineral forms of carbon, primarily calcium carbonate.

  • Soil organic carbon is composed of decaying plant and microorganisms and various carbon compounds.

  • Bulk density of a material is defined as the ratio of the mass of solids to the bulk volume, reported as g/cm3 for soils.

  • MMRV is a multi-step framework used to account for the greenhouse gas (GHG) emissions and emissions intensity associated with specific sources.

  • A specialized instrument used in analytical chemistry to determine the elemental makeup of a sample.

  • Soil is burned in a muffle furnace and the organic matter is determined by the difference in sample weight pre and post ignition.

  • There are several types of mechanical grinders on the market (e.g., flail or disk grinders) which are used by most commercial labs because they significantly accelerate soil processing time, reducing total costs of analyses.

  • We use a measure of variation to determine the precision of the procedure as it relates to quantification of SOC and define less variables as more precise.

Laboratory analyses

We compared common variations in how soils are processed at each laboratory procedural step, varying one factor per step, while maintaining the remaining steps consistent. Specifically, we:

1

Surveyed commercial and research laboratory procedures to determine the specific methods labs use, in order to evaluate the influence of differential procedures of soil carbon quantification.

2

Quantified the degree to which different soil laboratory processing procedures affect the precision, accuracy, cost and throughput of soil C assessment.

3

Carried out a comparison between our laboratory results and those from 8 external labs to assess the variability within and across labs.

We collected soils from 12 agricultural sites that varied in management as well as soil textures, rock, root, and carbon content, including SIC:SOC ratios.

Soil procedural variations

Visual representation of 11 procedural variations, with reference procedures represented in gray and the remaining steps that are varied represented by different colors. We implemented these procedural variations across 12 contrasting soils (Table S1). Sieving variations are represented as S1-3, grinding variations as G1-2, drying variations as D1-2, and quantification steps as Q1-3. Note that the quantification steps were not singularly tested since each quantification step required a particular preparation step.

Detailed description of each procedural variation

Sieving

Grinding

Total Carbon Quantification

Drying

  • Finely ground soil samples in glass vials are placed with the caps off in a 105°C oven overnight prior to elemental analysis.

  • Finely ground soil samples in glass vials are placed with the caps off in a 60°C oven overnight prior to elemental analysis.

  • Soils are not dried beyond the air-drying step for 2mm sieving but are finely ground prior to elemental analysis.

SOC Quantification

  • SOC is calculated by TC (from the elemental analyzer) minus SIC (pressure transducer).

  • Samples are scanned in the mid-infrared region. The produced spectra are then coupled with predictive models that must be well trained to produce accurate soil C estimates.

  • SOC is quantified by fumigating soils with acid to remove the SIC and running samples on the EA for the estimation of % SOC.

  • Soil is heated to 360°C for 2 hours in a muffle furnace to oxidize organic matter. The soil organic matter content is calculated by comparing the weight of a sample before and after it has been subjected to the high temperatures during ignition. SOC is then calculated using a conversion factor of 0.58.

SIC Quantification

Our Recommendations

Sieving


We recommend manually sieving soils to < 2mm. Fresh soils can either be sieved using an 8mm sieve, followed by 2mm or air-drying the bulk sample and using a rolling pin to break aggregates to move through a 2mm sieve.

We do not recommend solely using the 4mm sieve because it does not remove rocks.

We also do not recommend sieving using a mechanical grinder as this method resulted in significantly lower C measurements and higher variability, particularly in soils with medium to high rock content.

Grinding


We recommend finely grinding the samples beyond 2mm either using a ball mill or roller table to reduce variation among heterogeneous replicates.

The ball mill method resulted in significantly lower variability in soil carbon measurements.

The roller table method was more efficient and sufficiently reduced variability compared to not fine-grinding the samples.

Drying


We recommend drying soils at 105°C. However, in soils with a high proportion of organic matter, a soil moisture correction may be preferable prior to elemental analysis.

We note that none of the service labs we spoke with mentioned implementing a soil-specific moisture correction for mass to account for incomplete drying at lower temperatures but several labs noted using a standard 1-2% correction.

Quantification


We provide evidence that the use of FTIR spectroscopy coupled to predictive modeling was a promising approach to estimate soil TC, SIC, and SOC.

Accurate estimates of soil C using FTIR depend on the spectral libraries available and consistency following sample preparation and instrumentation protocols.

We also recommend the pressure transducer for direct measurements of SIC.

Despite finding a relatively low CV% in SOC, we do not recommend LOI. In our study, the low CV% using LOI was most likely a result of the rigorous sample preparation, ignition temperature, and time of exposure all being consistent across the samples. The range of LOI values are generally larger between labs when they use their own protocols instead of a rigorous standardized protocol.

We recommend that users of service laboratories should at a minimum request information about the soil processing methods. More important than a specific methodology is the consistency in how the methods are carried out and consistency from sample to sample, especially for samples that will be compared to each other.

Detailed Results

Impacts of sieving

  • The amount of rocks and plant material removed from samples varied between the four sieving procedures in soils with medium or high rocks or plant material but not for soils with low rocks or plant material.

  • The sieving procedure mattered more for soils with high rock and/or root content.

  • Leaving more coarse materials in soil samples resulted in higher variability in total and organic carbon estimates.

  • Rocks left in the sample have no organic matter and add mass. Large rocks that were not adequately removed significantly reduced % TC and SOC.

distribution of Coefficient of variance

 The distribution of the coefficient of variance (CV) across all soils (n=12) for each of the three grinding procedures tested, as described in Figure 1. Box plots report the median, first and third quartiles. Whiskers extend to the upper and lower data point that are within 1.5 times the interquartile range. Black dots represent the mean CV % SOC.

Impacts of grinding

  • Finely ground soils had lower variability in soil carbon.

  • Fine grinding helps homogenize heterogeneous samples to minimize between-sample variation.

  • If service laboratories are using elemental analyzers that accommodate larger sample masses, sample grind size and homogeneity may be less important.

Impacts of drying

  • We found lower SOC in air-dried soils, likely due to some moisture remaining in the sample and impacting the sample mass.

  • Different drying temperatures had no effect on the variability of the final carbon numbers, indicating that drying at any temperature is more effective than air-drying.

The difference (∆) in % soil organic carbon (SOC) compared to the reference (R) mean for all drying procedures, including R. Box plots report the median, first and third quartiles. Whiskers extend to the upper and lower data points that are within 1.5 times the interquartile range. Letters indicate the different soils, as described in Table 1, which are arranged on the x-axis by proportion of rock material removed from the R sieving procedure.

Different approaches to quantifying soil carbon

  • We found a strong correlation between the FTIR- predicted % TC and directly measured % TC using the elemental analyzer (EA). The FTIR spectroscopy method may thus be a good alternative to EA as it is both reliable and more time and cost efficient.

  • Getting accurate FTIR-predicted soil C measurements relies on developing or accessing rigorous spectral libraries and consistently following sample preparation and instrumentation protocols.

  • Calculating % SIC as % TC - % SOC  is not as precise as quantifying % SIC directly using FTIR or the pressure transducer method.

All quantification methods for % soil total carbon (TC), % soil inorganic carbon (SIC), and % soil organic carbon (SOC) plotted against the reference method where Q1 is predictions using Fourier transformed infrared spectroscopy, Q2 is acid fumigation, and Q3 is loss on ignition. The dashed line represents a 1:1 relationship. 

External Lab Comparison

We focused our external lab comparisons on variation among labs since we sent homogenized subsamples of the same soils, which would hypothetically yield similar results regardless of lab. These subsamples were also analyzed at CSU using rigorous and consistent methods and serve as a point of comparison. Variation among external labs was notable across several metrics, described in detail below.

  • For subsamples from the same soils, variation across labs was higher for soils with more plant material and higher presence of SIC.

  • Variability across labs was also higher for soils with higher SIC content. We attribute this to the challenges associated with accurately measuring % SIC, which also impacts SOC if it is quantified using TC - SIC.

  • This trend is also true looking across labs. Service labs reported a spread of 0.07 - 5.5 % SIC for soil H, which had the highest % SIC, as quantified by CSU using the reference methodology.

  • Variation across external labs is notable given the fact that the soils were homogenized to the extent possible without sieving before subsamples were sent for analysis or used at CSU for the reference method. Subsampling was performed in the same way for all samples, so the difference in variation between the CSU/Reference method and the commercial labs is more likely due to differences in processing than to soil sample heterogeneity.

Variation within and across labs in total carbon TC (panel a), soil inorganic carbon SIC (panel b), and soil organic carbon (panel C), with sites on the x-axis ordered from low to high in plant material. Circles represent field moist samples, triangles represent air-dried samples. The box plots represent the median, first, and third quartiles for subsamples analyzed across external labs, with gray box plots representing variability across replicates analyzed at CSU. 

The coefficient of variance (CV) of % soil organic carbon (SOC) distributed across soils categorized as having low (panel a), medium (panel b), or high (panel c) plant material for each sieving procedure. Box plots report the median, first and third quartiles. Whiskers extend to the upper and lower data points that are within 1.5 times the interquartile range. Time to sieve is represented as seconds per gram of soil on the x-axis.

Cost Evaluation

  • The most rigorous sieving procedure (8mm -> 2mm) took the most time in every instance, regardless of the total carbon content, but also increased with carbon content. 

  • The 4 mm sieve and the mechanical grinder were generally faster than the other two sieving procedures.

  • There is a trade-off between processing time and soil carbon quantification precision. Soils with higher carbon content were also generally soils with more plant material and the sieving procedures that removed less plant material were associated with more variability in total carbon.

Read the full results from our study here:

Even, R. J., Machmuller, M. B., Lavallee, J. M., Zelikova, T. J., and Cotrufo, M. F.: Large errors in soil carbon measurements attributed to inconsistent sample processing, Soil, https://soil.copernicus.org/articles/11/17/2025/

This work was supported by Environmental Defense Fund with awards from the Bezos Earth Fund, King Philanthropies, and Arcadia, a charitable fund of Lisbet Rausing and Peter Baldwin.