Influences of Cloud Super-Parameterization on Land-Atmosphere Energetics and Surface Temperature Biases

Most modern GCMs suggest that regions of strong land-atmosphere coupling (i.e. where local rainfall and soil moisture fluctuations are correlated) will be especially sensitive to climate change. However, in these regions (e.g. Central US) most current GCMs exhibit strong surface temperature biases, and are not able to realistically capture organized convection and rainfall intensity. These biases in present-day simulations undermine confidence in projections of future climate change. We evaluate the roles of clouds, radiation, and precipitation processes in contributing to surface temperature biases in a GCM that explicitly resolves convection using cloud super-parameterization (i.e. using embedded cloud-resolving models rather than conventional parameterizations to represent rainfall and turbulence processes). We find super-parameterization impacts both the net radiative input to the land-surface and the partitioning of sensible and latent heating with a distinct signature. We separate the effects of atmospheric processes, especially cloud radiative properties, from land-surface processes, especially soil-moisture dynamics, using a series of free-running and hindcast simulations.