Carbon released by agricultural land clearing moves toward open measurement

Food companies facing tighter corporate carbon rules will gain public access to the computational pipelines that calculate greenhouse gas emissions when forests and savannas become farms.

When a tract of tropical forest or wooded savanna is felled to plant crops, the carbon stored within living trees, undergrowth, and root systems does not simply vanish. It escapes into the atmosphere as carbon dioxide, transforming a living biological sink into a sudden burst of greenhouse gas emissions. For agricultural supply chains, this physical clearing of natural landscapes represents one of the largest drivers of climate impact, yet calculating the exact volume of carbon released has long remained an elusive accounting problem.

The dilemma stems from how agricultural commerce actually operates. A single chocolate bar or processed ingredient contains raw agricultural commodities gathered from thousands of scattered smallholders, passed through local middlemen, consolidated at regional drying stations, and shipped in bulk across oceans. By the time the commodity arrives at a processing facility, the physical connection to the specific parcel of soil where it grew has disappeared. Connecting that finished product to the precise hectare of forest cleared to plant it requires combining satellite observations of forest disturbance with complex statistical models of trade flows.

To solve that accounting disconnect, corporate carbon accounting relies on a technical discipline known as land use change modelling. In September 2026, enterprise climate software developer Watershed Technology, Inc. and geospatial research firm AdAstra Sustainability announced a partnership to publish the computational pipeline of Orbae, a dedicated land use change emissions dataset, as open-source software under the Cornerstone Sustainability Data Initiative.¹ The collaboration aims to replace closed, proprietary carbon calculators with transparent computational tools that outside scientists and corporate auditors can inspect directly.

Before such models can generate meaningful figures, their internal algorithms must mirror the physical sequence of land clearing. First, natural vegetation must be removed and the ground turned over for crop cultivation, releasing carbon from decaying wood and exposed soil. Second, regional agricultural yields decide how much harvest that cleared land actually produces each season. Third, domestic and international shipping routes distribute that harvest into common trade streams. Finally, an accounting framework must distribute the initial carbon pulse from the clearing across the subsequent years of agricultural production. If this sequence is modelled inaccurately, corporate climate reporting either ignores severe deforestation or penalises farmers who cultivate long-established fields.

Why is measuring carbon from cleared land so difficult?

Carbon accounting across agricultural supply chains breaks down because tracking which patch of forest was felled for a specific harvest requires matching fragmented trade records with satellite observations. Across global food production, land conversion can dominate a business footprint. For high-risk commodities such as cocoa, emissions from clearing land can account for over 90% of total product emissions, according to figures released by Watershed and AdAstra Sustainability.¹ When businesses attempt to calculate those numbers, they frequently rely on closed databases that produce widely differing estimates for identical regions.¹

The urgency to reconcile those conflicting figures follows new international regulatory mandates. In early 2026, the Greenhouse Gas Protocol published its Land Sector and Removals Standard, which establishes binding guidelines for agricultural accounting.¹ Under this regulatory standard, commercial enterprises will be required to measure land use change emissions starting January 1, 2027, in preparation for mandatory corporate disclosures beginning in 2028.¹ Companies are also attempting to demonstrate genuine reductions against targets set by the Science Based Targets initiative Forest, Land and Agriculture guidance, creating immediate corporate demand for credible emissions data.¹

Michael Steffen, the head of climate data at Watershed, stated in the announcement that land conversion often represents the single largest line item in a food company footprint as well as its primary opportunity for emissions reductions.¹ Yet Steffen noted that the measurement of land clearing has historically suffered from opacity and methodological friction, a dynamic intensified by the arrival of simultaneous regulatory requirements from standard-setting bodies.¹

How does opening the underlying code change emissions accounting?

Publishing the computational code behind land conversion models allows external researchers and auditors to examine every mathematical step between raw satellite pixels and reported corporate footprints. Although AdAstra Sustainability made the jurisdictional direct land use change data and methodological documentation within Orbae openly accessible in early 2025, the underlying programmatic pipeline that processes those numbers remained proprietary.¹ Moving that computational machinery into the public domain enables independent scientists to stress-test the calculations and modify the code for localized farming contexts.

Xavier Bengoa, the co-founder and chief executive of AdAstra Sustainability, explained in the announcement that open-sourcing the algorithmic pipeline represents a necessary progression toward a shared scientific vocabulary.¹ Bengoa noted that public access to the underlying code allows the broader research community to scrutinize, improve, and deploy the methodology at scale, rather than relying on unexaminable estimates as corporate deadlines for 2030 approach.¹

The joint technical roadmap plans several sequential dataset releases through the Cornerstone platform. In 2026, the partnership will release an open-source global statistical land use change dataset designed to comply with the Greenhouse Gas Protocol Land Sector and Removals Standard.¹ AdAstra Sustainability will also issue an interim proxy dataset later in 2026 based on jurisdictional principles, which will subsequently be open-sourced under Cornerstone in 2027.¹ By 2027, Watershed and AdAstra Sustainability intend to release a next-generation open-source direct land use change codebase that supports both farm-level evaluations and regional calculations covering more than 90% of global cropland.¹

What limits remain in tracking agricultural land conversion?

Statistical estimates and regional proxies cannot eliminate the physical uncertainty of matching commodities traded through bulk elevators to specific farm boundaries. The datasets produced by these pipelines are model-driven historical allocations, not continuous, real-time measurements of atmospheric greenhouse gases over every field. They rely on satellite imagery and regional crop production averages, meaning they can approximate the probability that a commodity originated from recently cleared land without proving the physical provenance of an individual shipment. When local traceability stops at a regional collection center, mathematical models must apportion land clearing across all participating farms, which can overestimate emissions for producers operating on older plots while undercounting clearing on aggressive agricultural frontiers.

Integrating land clearing calculations into broader corporate accounting also requires merging disparate economic models. The Cornerstone Sustainability Data Initiative, launched in 2025 by environmental consulting firm ERG, the Stanford Doerr School of Sustainability Sustainable Solutions Lab, and Watershed, was created to maintain open access to core emissions models.¹ The initiative already houses the United States Environmentally-Extended Input-Output model, originally created by the United States Environmental Protection Agency to cover more than 400 economic sectors, alongside the CEDA model, which tracks emissions across 400 commercial sectors across 148 countries and regions.¹ Steering committee members, including Dr. Wesley Ingwersen and Dr. Steve Davis of Stanford University together with Dr. Sangwon Suh and Dr. Mo Li of Watershed, are currently harmonizing those macroeconomic frameworks to handle fine-grained land data.²

Food and consumer goods companies, including multinational dairy producer Danone and agricultural supplier ofi, already employ Orbae within their commercial supply operations.¹ The forthcoming releases will determine whether providing open code can create an agreed standard across corporate boardrooms, third-party certifiers, and environmental regulators before the 2027 measurement mandates take legal effect.

This piece was prepared from the public announcement and records; the authors have not been interviewed.

References

This article is based on 2 sources, listed in the order they are cited.

1. 1 WT Watershed Technology, Inc. announcement · 10 Sep 2026 Watershed and AdAstra partner to open-source Orbae, the leading land use change dataset See the source
2. 2 W Watershed Cornerstone Sustainability Data Initiative See the source