Blog

Why Calibration and Validation Are Critical in Soil Carbon Models

Written by Cara Mathers, PhD | September 05, 2023

Quantifying the environmental benefits of sustainable farming practices requires scientific rigor and transparency in order to build a credible market that ensures corporate financing produces real environmental outcomes for corporations while compensating farmers for transitioning to regenerative management. Indigo Ag has been breaking new ground in soil carbon quantification in close partnership with the scientific community since launching Carbon by Indigo in 2019. Indigo’s hybrid use of modeling and soil sampling sets a high standard for how soil carbon is quantified. Our DayCent-CR biogeochemical model is calibrated, validated, reviewed by independent experts, and then verified by 3rd party entities approved for use by crediting programs, in this case the Climate Action Reserve (CAR)—while accommodating the uncertainties inherent in carbon measurement.

The documentation and review of the model validation for use under the CAR Soil Enrichment Protocol (SEP) are publicly available along with other information for each of our credit issuances. We understand, however, that technical documentation submitted through a registry is not everyone’s preferred reading material. To double down on our commitment to rigor and transparency, we’ve gone further and recently published a peer-reviewed article, “Validating DayCent-CR for cropland soil carbon offset reporting at a national scale,” in the international soils journal Geoderma. This article provides deeper transparency for a key component of our overall MRV (monitoring, reporting, and verification) processes.

Indigo is currently the only agricultural offset producer to have both registry approval of our model validation in addition to academic peer-review, demonstrating our commitment to the scientific integrity of our programs. Publication of Indigo's calibration and validation work in Geoderma will increase stakeholder (including buyers, verifiers, academics, government, and others) confidence in the quality of Indigo's sustainability programs and carbon credits.

“As a company, being science led is a core principle to our business success,” said Ron Hovsepian, CEO of Indigo Ag. “While this paper validates our models for offsets, the approach we take is consistent across our solutions and is the starting point to account for soil organic carbon removals in Scope 3 programs. This is incredibly important to customers of our sustainability programs, including Carbon by Indigo and our Market+ Source sustainable grain program, as well as our farmers and agribusiness partners to ensure that they are financing real, meaningful environmental benefits resulting from sustainable practices.”

You can read about our model calibration and validation in the Geoderma article. Below we’ve addressed some commonly cited questions about soil carbon modeling.

What is DayCent-CR, and why do we need to calibrate and validate it?

DayCent-CR is a biogeochemical process model that simulates changes in soil organic carbon (SOC) at the soil-plant-atmosphere interface. It is the main tool that we use to quantify the number of soil carbon credits produced through regenerative agriculture. DayCent is the daily time-step version of the CENTURY model originating out of Colorado State University, and the “CR” signifies the proprietary “credit-ready” version used in Indigo’s carbon crediting program.

While DayCent comes with a standard “factory setting,” we need to calibrate it to the various crops, practices, and geographies that are within Indigo’s project. We then need to measure how well the calibrated model performs through a validation exercise. If a model is used without being calibrated, or shows poor performance through validation, it will produce inaccurate estimations. As credits are purchased by companies to offset their own emissions, it’s of incredibly high importance to ensure that any carbon credit produced and sold is based on real carbon sequestration.

What data do we use in the calibration and validation process?

The model is calibrated and validated with measurements of SOC from published field studies obtained through a literature search. The CAR Soil Enrichment Protocol (SEP) model validation guidance sets the criteria for what SOC data can and cannot be included. In total, we used 668 SOC stocks for calibration and validation of DayCent-CR.

But, we’re still hungry for data! Of 152 experimental sites that we initially evaluated, only 41 met the criteria for inclusion into our calibration and validation process. While we were able to effectively perform model calibration and validation with data from those experimental sites through a k-fold cross validation technique, our model will be improved over time through ingesting additional, high-quality data.

Addressing Model Uncertainty

To produce credits, we use our calibrated DayCent-CR coupled with actual soil carbon measurements collected from the field. The SOC measurements initialize our model. This process enables us to produce credits on a much larger acreage and on a shorter timescale than soil sampling alone.

Validating DayCent-CR for cropland soil carbon offset reporting at a national scale” also explains how we calculate model uncertainty and the implications of model uncertainty as well as this position paper: Managing Quantification Uncertainty in Agricultural Carbon Markets. With model improvements from DayCent 1.0 to DayCent 1.0.2, we were able to reduce model prediction uncertainty from 37% to 18%. No model will ever give estimates that are 100% certain, so we deduct credits from the overall total that we produce to account for uncertainty and ensure we are selling an appropriately conservative number of credits.

Why do we report our model results?

In order for our credits to undergo verification against the CAR SEP, we must explain our model validation process and showcase the results in a validation report. The reports are reviewed by an independent 3rd party modeling expert. Once approved, these reports appear on CAR’s website and are accessible for anyone to read and scrutinize.

Since the inception of the carbon program, we have completed two model validation reports: DayCent-CR version 1.0 in 2021 and DayCent-CR version 1.0.2 in 2022. Every time we add a new crop or potential practice change to our model, we must re-calibrate it and create an updated model validation report.

"Validating DayCent-CR for cropland soil carbon offset reporting at a national scale” highlights the results from Indigo’s second validation report, where we added cotton as newly eligible crop in 2022.

What’s next?

In 2019, we started a long-term, large-scale soil carbon experiment on cropland fields that have transitioned to regenerative management. We are now beginning to analyze soil carbon stock change at early (< 5 year) time scales. We plan to publish the data, and then add them to our model calibration and validation dataset to improve DayCent-CR. The experiment is ongoing, and we will continue to monitor SOC and additional soil health properties.

Indigo scientists are currently preparing other manuscripts for publication across a range of topics. Additionally, we will be presenting research at the Tri-Societies Annual Meeting and American Geophysical Union Fall Meeting. We look forward to meeting you there!