Modeling soil organic carbon stocks and changes in agricultural cropping systems using a decision support tool and process-based model

Soil organic carbon (SOC) is a key indicator of soil health, yet changes in SOC often take decades to detect. Assessment tools, accessible to non-expert users, are needed to accurately and cost-effectively quantify SOC stocks. Soil sampling and analysis are ideal for quantification but are also resource- and time-intensive. A decision-support tool, COMET-Farm, could improve the transition to low-C agricultural supply chains; however, it was designed to estimate ΔSOC rather than SOC stocks directly, and its accuracy across diverse agricultural settings remains uncertain. We investigated whether the COMET-Farm tool could accurately estimate SOC stocks on 15 working farms in Maryland with variable management practices, including tillage, cover cropping, manure application, and irrigation. Field-measured SOC stocks (0–20 cm soil depth) of 17.1 to 73.6 Mg ha −1 were compared with model-predicted values using farmer-reported data in (i) COMET-Farm with default inputs supplied to DayCent, versus (ii) expert-applied DayCent. We found COMET-Farm predictions were lower than measured SOC stocks (R 2 = 0.101; RMSE = 18.80 Mg ha −1 ), while expert-applied DayCent predictions more closely matched field measurements (R 2 = 0.73; RMSE = 7.89 Mg ha −1 ). Poorly drained soils and long-term manure applications caused greater discrepancies between measured and predicted SOC. These findings suggest that COMET-Farm estimates could improve with better model initialization, more accurate site-specific soil data, such as measured soil texture and drainage conditions, and more regionally representative land-use histories. With these improvements and further testing, COMET-Farm could be expanded beyond its original purpose of estimating ΔSOC to directly estimate SOC stocks for non-expert users.