Accurate simulation of soil organic carbon (SOC) dynamics by terrestrial biosphere models is hampered by poorly constrained parameters and parameter equifinality, amongst other issues. To address this, we use Bayesian optimization to constrain the 16 SOC-related parameters in the Canadian Land Surface Scheme Including biogeochemical Cycles (CLASSIC). We employed a global sensitivity analysis (Sobol’) to develop four parameter sets based upon different sensitivity criteria. We then optimized each set against observed SOC (World Soil Information Service; WoSIS) and soil respiration (Soil Respiration Database; SRDB). Using two different loss functions; one focused on reproducing the observational mean value, and the other explicitly accounting for an estimated observational uncertainty. The best optimized parameter sets from each loss function had an average relative difference of 61%. Thus the choice of loss function impacts what parameter values are deemed optimal and should be considered carefully. The final selected optimal parameter set saw a 12% improvement against WoSIS and SRDB, had global SOC totals in line with literature estimates, and better simulated high-latitude SOC stocks evaluated against the Northern Circumpolar Soil Carbon Database (RMSD: 16.39 vs. 17.61; bias: -5.57 vs. -10.78 kg C m2) compared to the default CLASSIC parameters. However, some parameters were not well constrained, in particular those of needle-leaf deciduous trees which dominate Siberian boreal forests, a region relatively poorly observed in WoSIS and SRDB. Future work should apply further constraints on the optimization framework and address observational gaps.

Land surface/Earth System models depend upon accurate simulation of evapotranspiration (ET) to avoid excessive biases in simulated energy, water, and carbon cycles. The Canadian Land Surface Scheme including biogeochemical Cycles (CLASSIC), the land surface scheme of the Canadian Earth System Model (CanESM) shows reasonable ET fluxes globally, but CLASSIC’s partitioning into evaporation (E) and transpiration (T) can be improved. Specifically, CLASSIC exhibited a high soil evaporation (Es) bias in sparsely vegetated areas during wet periods, which can deplete soil water and decrease photosynthesis and T later in the year. A dry surface layer (DSL) parameterization was implemented to address biases in Es through an increased surface resistance to water vapour and heat fluxes. In arid/semi-arid regions, the DSL decreased Es, leading to improved seasonality of ET and increased gross primary productivity (GPP) due to an increase in soil moisture. The DSL simulations significantly (t-test, p<0.01) increased T/ET from 0.25 in baseline CLASSIC to 0.30 in the DSL simulations. T/ET was further increased to 0.41 (p<0.01), comparable to the CMIP5 model mean, by allowing T to occur from the dry canopy fraction while water evaporates from the wet fraction. This mainly affected densely vegetated areas, where T and ET increased significantly (p<0.01) and canopy E was reduced (p<0.01). In seasonally dry tropical forests, higher T and ET reduced GPP. Despite increases in arid/semi-arid regions, the reduced GPP in tropical forests resulted in ∼1.6% lower global GPP (p=0.018) than baseline CLASSIC. Including these modifications in CanESM might reduce biases in climate.