A vertical representation of soil carbon in the JULES land surface scheme (vn4.3_permafrost) with a focus on permafrost regions

Abstract. An improved representation of the carbon cycle in permafrost regions will enable more realistic projections of the future climate–carbon system. Currently JULES (the Joint UK Land Environment Simulator) – the land surface model of the UK Earth System Model (UKESM) – uses the standard four-pool RothC soil carbon model. This paper describes a new version of JULES (vn4.3_permafrost) in which the soil vertical dimension is added to the soil carbon model, with a set of four pools in every soil layer. The respiration rate in each soil layer depends on the temperature and moisture conditions in that layer. Cryoturbation/bioturbation processes, which transfer soil carbon between layers, are represented by diffusive mixing. The litter inputs and the soil respiration are both parametrized to decrease with increasing depth. The model now includes a tracer so that selected soil carbon can be labelled and tracked through a simulation. Simulations show an improvement in the large-scale horizontal and vertical distribution of soil carbon over the standard version of JULES (vn4.3). Like the standard version of JULES, the vertically discretized model is still unable to simulate enough soil carbon in the tundra regions. This is in part because JULES underestimates the plant productivity over the tundra, but also because not all of the processes relevant for the accumulation of permafrost carbon, such as peat development, are included in the model. In comparison with the standard model, the vertically discretized model shows a delay in the onset of soil respiration in the spring, resulting in an increased net uptake of carbon during this time. In order to provide a more suitable representation of permafrost carbon for quantifying the permafrost carbon feedback within UKESM, the deep soil carbon in the permafrost region (below 1 m) was initialized using the observed soil carbon. There is now a slight drift in the soil carbon ( <  0.018 % decade⁻¹), but the change in simulated soil carbon over the 20th century, when there is little climate change, is comparable to the original vertically discretized model and significantly larger than the drift.