Abstract/Summary

This study presents an assessment of the ability of the Hadley Centre Global Environment Model version 2—Earth system configuration (HadGEM2-ES)—in simulating the mid-latitude storm tracks over the Southern Hemisphere (SH). The storm tracks are primarily assessed using cyclone tracking using data from a 4 member ensemble of 27-year simulations of HadGEM2-ES over the historical period, and the European Centre for Medium-Range Weather Forecasts Interim Reanalysis. Both winter and summer periods are considered and contrasted. Results show that the storm track (ST) climatology of HadGEM2-ES presents similar patterns to those of the reanalysis. However, the model tends to represent the austral winter ST position with an equatorward bias and a zonal bias in the spiral towards the pole. The main differences were found during the austral winter, with large track density biases over the Indian Ocean indicating a poor representation of the ST in this specific region. This was found to be related to two factors. First, the large negative genesis biases over South America, Antarctic Peninsula and the Antarctic coast. Second, the model resolution and the representation of the Andes Mountains in South America. The link between STs and the large-scale circulation is examined and shows at upper levels an equatorward jet position bias of the subtropical jet and a negative bias in the eddy-driven, associated with a large cold bias over the extratropical and polar regions. The analysis of the large-scale circulation shows that the split jet during winter has problems in the model linked to these biases, including geopotential anomaly and sea surface temperature biases. Consequently, in general the track densities over the Southern oceans are underestimated in the austral winter. During summer, the results show the STs move poleward and there is a single eddy-driven jet, which is represented relatively well compared with the winter situation. These factors tend to reduce the differences seen in the cyclone track distribution biases. Although the model has biases in the ST behaviour in the SH it is still considered that these do not preclude this model being used for perturbation and future projection studies.