The significance of the melt-pond scheme in a CMIP6 global climate model

Abstract The impact of melt ponds on sea-ice albedo has been observed and documented. In general circulation models, ponds are now accounted for through indirect diagnostic treatments (‘implicit’ schemes), or prognostic melt-pond parametrisations (‘explicit’ schemes). However, there has been a lack of studies showing the impacts of these schemes on simulated Arctic climate. We focus here on rectifying this using the general circulation model HadGEM3, one of the few models with a detailed explicit pond-scheme. We identify the impact of melt ponds on the sea ice and climate, and associated ice-ocean-atmosphere interactions. We run a set of constant forcing simulations for three different periods and show, for the first time, using mechanistically different pond-schemes can lead to very significantly different sea-ice and climate states. Under near-future conditions, an implicit scheme never yields an ice-free summer Arctic, whilst an explicit scheme yields an ice-free Arctic 35% of years and raises autumn Arctic air temperatures by 5 to 8 °C. We find that impacts on climate and sea ice depend on the ice state: under near-future and Last Interglacial conditions, the thin sea ice is very sensitive to pond formation and parametrisation, whilst during the Pre-Industrial, the thicker sea ice is less sensitive to the pond-scheme choice. Both of these two commonly-used parametrisations of sea-ice albedo yield similar results under pre-industrial conditions, but in warmer climates, lead to very different Arctic sea ice and ocean and atmospheric temperatures. Thus changes to physical parametrisations in the sea-ice model can have large impacts on simulated sea ice, ocean and atmosphere.