A scalable tool for farm-level carbon accounting: evidence from UK agriculture

Abstract Accurately quantifying greenhouse gas (GHG) emissions at the farm level is crucial for agricultural decarbonisation, yet doing so remains challenging—particularly for smaller farming operations that lack resources for comprehensive data collection. To address this gap, we present a new framework to predict farm-level direct emission intensities (tCO 2 e/£) from a minimal set of input variables. Our approach leverages a unique and comprehensive dataset that integrates survey based GHG audits from 482 farms across the United Kingdom (UK) with financial transaction data, obtained through a collaboration with a major high-street bank and a leading UK charity promoting sustainable agriculture. By combining granular farm-level emissions metrics with financial records, we demonstrate that a limited set of input variables focused on the highest impact areas—particularly dairy and beef cattle intensities—yield robust predictions that can explain up to 91% of the variation in farm-level emissions. Although this model does not replace the depth and specificity of established carbon calculators, it provides a pragmatic, scalable alternative for emissions reporting. It offers farms a clear entry-point to carbon accounting and provides financial institutions with a data-efficient method for more accurate assessments of the environmental impact of their agricultural portfolios.