Simulation of soil carbon efflux from an arable soil using the ECOSSE model: Need for an improved model evaluation framework?

Globally, it is estimated that ~1500PgC of organic carbon is stored in the top meter of terrestrial soils. This represents the largest terrestrial pool of carbon. Appropriate management of soils, to maintain or increase the soil carbon pool, represents a significant climate change mitigation opportunity. To achieve this, appropriate tools and models are required in order to more accurately estimate soil carbon fluxes with a view to informing and developing more effective land use management strategies. Central to this is the evaluation of models currently in use to estimate soil carbon emissions. In the present study, we evaluate the ECOSSE (Estimating Carbon in Organic Soils – Sequestration and Emissions) model which has its origins in both SUNDIAL and RothC and has been widely used globally to model soil CO2 fluxes across different locations and land-use types on both organic and mineral soils. In contrast to previous studies, the model was found to poorly represent observed soil respiration at the study site, an arable cropland on mineral soil located in south-east Ireland. To isolate potential sources of error, the model was decomposed into its component rate equations or modifiers. This investigation highlighted a deficiency in the model simulated soil water, resulting in significant inhibition of the model simulated CO2 flux relative to the observed data. When measured values of soil water at the site were employed, the model simulated soil respiration improved significantly (r2 of 0.775 vs 0.154). This highlighted model deficiency remains to be evaluated at other sites; however, the research highlights the need for a more comprehensive evaluation of soil carbon models prior to their use in informing policy, particularly models which are employed at larger scales and for climate change projections. [Display omitted] •Soil carbon modelling is essential for quantification of GHG flux.•Commonly used soil C model was evaluated and discrepancies were found.•Deconstruction of model rate modifiers shows deficiency in water component.•r2 Values improve significantly using observed soil water content instead of modelled.•We recommend stronger evaluation of biogeochemical models.