MISMIP+ (the third Marine Ice Sheet MIP), ISOMIP+ (the second Ice Shelf-Ocean MIP) and MISOMIP1 (the first Marine Ice Sheet-Ocean MIP) prescribe a set of idealized experiments for marine ice-sheet models, ocean models with ice-shelf cavities, and coupled ice sheet-ocean models, respectively. Here, we present results from ISOMIP+ and MISOMIP1 experiments using several ocean-only and coupled ice sheet-ocean models.

Among the ocean models, we show that differences in model behavior are significant enough that similar results can only be achieved by tuning model parameters (the heat- and salt-transfer coefficients across the sub-ice-shelf boundary layer) for each model. This tuning is constrained by a desired mean melt rate in quasi-steady state under specified forcing conditions, akin to tuning the models to match observed melt rates. We compare the evolution of ocean temperature transects, melt rate, friction velocity and thermal driving between ocean models for the five ISOMIP+ experiments (Ocean0-4), which have prescribed ice-shelf topography. We find that melt patterns differ between models based on the relative importance of overturning strength and vertical mixing of temperature even when the models have been tuned to achieve similar melt rates near the grounding line.

For the two MISOMIP1 experiments (IceOcean1 without dynamic calving and IceOcean2 with a simple calving parameterization), we compare temperature transects, melt rate, ice-shelf topography and grounded area across models and for several model configurations. Consistent with preliminary results from MISMIP+, we find that for a given coupled model, the use of a Coulomb-limited basal friction parameterization below grounded ice and the application of dynamic calving both significantly increase the rate of grounding-line retreat, whereas the rate of retreat appears to be less sensitive to the ice stress approximation (shallow-shelf approximation, higher-order, etc.). We show that models with similar mean melt rates, stress approximations and basal friction parameterizations produce markedly different rates of grounding-line retreat, and we investigate possible sources of these disparities (e.g. differences in coupling strategy or melt distribution).