Toward Eliminating the Decades-Old “Too Zonal and Too Equatorward” Storm-Track Bias in Climate Models

Abstract

Generations of climate models exhibit biases in their representation of North Atlantic storm tracks, which tend to be too far near the equator and too zonal. Additionally, models have difficulties simulating explosive cyclone growth. These biases are one of the reasons for the uncertainties in projections of future climate over Europe, and the underlying causes have yet to be determined. All three biases are shown to be related, and diabatic processes are pointed to as a likely cause. To demonstrate this, two hemispherically symmetric storm tracks forming downstream of an idealized sea surface temperature (SST) front on an aquaplanet are examined using the seamless ICOsahedral Non-hydrostatic weather and climate model (ICON) and its grid refinement capabilities. The analyzed perpetual boreal winter has a global grid spacing of 20 km, two bi-directionally interacting grid nests over the Northern Hemisphere that refine the grid to 10-km spacing over much of the stormtrack and further to 5-km spacing near the SST front. In contrast, no grid refinement is performed for the Southern Hemisphere. Feature-based cyclone tracking shows that the poleward propagation in the NH is enhanced, so the high-resolution storm track is less equatorward and less zonal; explosive deepening rates are more frequent and precipitation rates are amplified. The implication is that resolving diabatic processes on the storm scale improves all three intersecting biases in the representation of storm tracks. While new challenges arise at cloud resolving scales, much improvement for the representation of storm tracks will be gained once climate models resolve the meso-γ scale.