Skip to content

Response to CO2 doubling of the Atlantic Hurricane Main Development Region in a High-Resolution Climate Model

April 15th, 2013


Key Findings

  • Response of tropical Atlantic to doubling of CO2 explored with new high-resolution GFDL global climate model, CM2.5.
  • Year-to-year variations in Atlantic hurricane development region: sea surface temperature increase and shift towards summer in response to increased CO2.
  • Long-term average Atlantic hurricane frequency decreases in response to CO2 doubling, but hurricane frequency in extremely active years remains similar.
  • While there is some indication of a shift in interannual variation in the observed record, further analysis is needed.
  • Future climate extremes are impacted by changes in average conditions, climate variations and changes in climate variations.

Takeshi Doi, Gabriel A. Vecchi, Anthony J. Rosati, and Thomas L. Delworth. Journal: Journal of Climate. DOI:10.1175/JCLI-D-12-00110.1

Summary

The authors simulated the response of sea surface temperature (SST) in the Atlantic Hurricane Main Development Region (MDR) to a doubling of CO2, using a cutting-edge global high-resolution coupled model developed at GFDL (CM2.5). This model has been shown to produce a very faithful simulation of the observed seasonal cycle and year-to-year (or interannual) variability in the tropical Atlantic. The skillful representation of Atlantic interannual variability enables the exploration of the response of interannual variability to increasing CO2 – in addition to exploring changes in the average conditions in the Atlantic.

This study reveals a significant increase in the amplitude of interannual variations of SST in the Atlantic MDR: the amplitude of these SST variations increases by about 25% in response to CO2 doubling. Further, the seasonal timing of the peak in interannual SST variations moves from northern hemisphere spring to early summer, moving that peak variability closer to the North Atlantic hurricane season (which begins in June). The changes to the interannual variations of the SST in the Atlantic Hurricane Main Development Region are due to an increase in effectiveness of the Wind-Evaporation-SST positive feedback.

A statistical refinement allows estimation of the impact of CO2 changes on Atlantic hurricane frequency. The average number of hurricanes is found to decrease in response to increasing CO2, consistent with previous studies. Nevertheless, the frequency of Atlantic hurricanes in the most active years is as high or higher after doubling CO2, because of the enhanced year-to-year variability in Atlantic MDR SSTs and its alignment with the Atlantic hurricane season.

Exploration of the observed record suggests some evidence for an enhancement of the interannual variation of SST in the Atlantic Hurricane Main Development Region in boreal early summer since the late 19th century, although there are substantial uncertainties in the observational record.

The combined impact of natural climate variations, changes in average conditions, and changes in variability is key to the response of climate extremes to increasing greenhouse gases. This change in the interannual variability of SST will be one of the factors in projecting the risk of extremes in Atlantic Hurricanes, and for drought (or flood) in the Sahel and South America in a future climate.

Figure: (a) Monthly standard deviation of the interannual variation of the SSTMDR for the present-day Control (thick blue line) and the CO2 doubling run (thick red line) in model years 91-140 (°C). The thin blue lines show the four other 50-years mean standard deviations in the Control run. A running mean of three months is applied. (b) Same as (a), but for observation averaged in three periods of 1881-2009, 1960-2009, and 1881-1930 in the ERSSTv3 (black) and HadISST data (grey).
Figure: (a) Monthly standard deviation of the interannual variation of the SSTMDR for the present-day Control (thick blue line) and the CO2 doubling run (thick red line) in model years 91-140 (°C). The thin blue lines show the four other 50-years mean standard deviations in the Control run. A running mean of three months is applied. (b) Same as (a), but for observation averaged in three periods of 1881-2009, 1960-2009, and 1881-1930 in the ERSSTv3 (black) and HadISST data (grey).