A Classification Framework for Carbon Sequestration Potential: Modelling Soil Carbon Pools With Environmental Data

This study aimed to develop a comprehensive spatial assessment of soil carbon sequestration potential for supporting future decisions on soil carbon stock monitoring and management. Firstly, the spatial distribution of soil organic carbon (SOC) and mineral associated organic carbon (MAOC) was mapped at the 500 m resolution at the northern half of Ireland using machine learning tools with environmental covariates and soil geochemistry. Random forest models achieved robust estimations of SOC and MAOC, with R 2 values of 0.69 and 0.73 respectively. Afterwards, carbon sequestration potential was assessed using two complementary approaches: (1) quantile regression to estimate MAOC saturation and deficit, and (2) a data‐driven method using agroclimatic‐landcover units to estimate achievable SOC levels. These approaches revealed contrasting spatial patterns, reflecting different carbon storage mechanisms constrained by mineralogy versus climate‐land cover conditions. Hence, we integrated both methods into a comprehensive framework to support decision‐making for carbon farming. Classified by below or above the mean value of estimated SOC seq and MAOC seq , a four‐classification framework of high/low SOC seq by high/low MAOC seq was established and mapped to delineate areas in each class with distinct sequestration potential characteristics and corresponding management strategies. These included dual saturation areas (Low SOC seq –Low MAOC seq : 34.87%) requiring protection strategies and dual high potential areas (High SOC seq –High MAOC seq : 32.93%) ideal for carbon farming projects, with other areas having specific limitations in either MAOC (High SOC seq –Low MAOC seq : 20.24%) or SOC (Low SOC seq –High MAOC seq : 11.95%) accumulation. This provides a practical and effective framework that supports targeted policymaking for soil carbon management and climate mitigation.