Modelling the impacts of organic matter quality on soil carbon turnover and storage : continued testing of a novel dual-pore model against data collected from a long-term study site in Offer, Sweden

Abstract: As we look for solutions to the worsening climate crisis, carbon sequestration has become a popular strategy to mitigate CO2 emissions. Carbon models are used to predict or emulate changes in future soil organic carbon (SOC) stocks, but most fail to account for the influence of pore size on soil structure and SOC storage. This study examines a new dual-pore model against a long-term study site in Offer, Sweden, to determine if it is possible to predict SOC storage capacity based upon SOM retention and quality, by accounting for pore size and distribution. Data from the long-term study site provided soil samples from two different cropping treatments (A and D); yearly records of crop yield measurements; some SOC and bulk density (BD) measurements; and water content and pressure head data for 1956 and 2019. The model simulates an increase in soil organic matter (SOM) for a prolonged ley treatment (A) and a decrease in an annual cropping treatment with tillage (D). There is a 12 percent gain of SOM in treatment A and a 32 percent loss of SOM in treatment D, attributed primarily to fluctuations in storage of microbially-processed SOM. While the model does not precisely fit the data, there is a clear correlation between the simulated and modelled SOC values - r values of 0.91 and 0.83, RMSE and MAE values below 0.002, and NSE values of 0.71 and 0.55 for treatments A and D, respectively. For BD, the model does not fit the data for treatment A very well; but, for treatment D, a positive NSE value of 0.06 and r value of 0.66 suggests that there was some correlation. There is a near perfect fit for treatment D when the final BD measurement is removed. When accounting only for SOC and BD and assigning them equal weights, the EF values were 0.21 for treatment A and 0.31 for treatment D, indicating that the dual-pore model is relatively successful overall. Error in the model may be due to a lack of data points, but despite this, it was possible to model SOC well, as the trends in the figures demonstrate. The most important finding of the study is that the model successfully calibrates an organic matter retention coefficient for treatment A (0.29) that is twice for as large as treatment D (0.15), reflecting higher organic matter quality and the developed soil structure of a prolonged grass ley. This is a promising outcome in a second test of a novel dual-pore, dual-carbon model.