Translating historical extreme weather events into a warmer world

How would the consequences of Storm Ulysses (1903) be different if the same storm occurred today?

Whenever a severe weather event occurs with harmful impacts in a particular region, it is often asked by disaster responders, recovery planners, politicians, and journalists whether climate change caused or affected the event. The harmful impacts are caused by the unusual weather, but climate change may have made the weather event more likely, more severe, or both.

We introduce a novel approach to use the 20th Century Reanalysis v3 system (20CRv3) to generate counter-factual reconstructions of extreme events by rerunning the reanalysis for those events with perturbed SST boundary conditions and assimilating the same surface pressure observations. As atmospheric temperature changes will be primarily mediated by imposed changes to SSTs, this perturbation translates the reanalysis of the events into a changed climate.

As a proof of concept, we have applied this approach to a severe extratropical windstorm known as Storm Ulysses, which occurred in February 1903, to examine how the consequences of this historical event may have been different had it occurred in a warmer world.

Three experiments with the same reanalysis system are considered: (1) the original 20CRv3, (2) an improved version of 20CRv3 with added historical surface pressure observations and a small change to the data assimilation scheme, and (3) same as (2) but with a spatially uniform +2 K perturbation added to the SST boundary conditions.

Figure 1: Translating Storm Ulysses into a warmer world.

Storm Ulysses is not well represented in the original 20CRv3, but the improved reanalysis shows a far more intense storm, mainly from adding new rescued pressure observations; it is a credible reconstruction of one of the most extreme windstorms to occur over the British–Irish Isles in the past 120 years. Figure 1 (top row) shows the wet-bulb potential temperature at 850 hPa during the storm for the different experiments. The storm in the improved reanalysis has a hook-shaped Shapiro–Keyser-type structure which is not present in the original 20CRv3. As expected, the structure of the storm in the warmer-world reanalysis is constrained to be very similar to the improved reanalysis, except for a roughly 2.5 K offset. Specific humidity and surface air temperatures are also higher (not shown).

The warmer-world reanalysis of the storm generates stronger maximum winds (middle row) and increased rainfall (bottom row) compared to the improved reanalysis, suggesting that the consequences of the same surface circulation pattern would be more severe in a warmer world. While the increase in wind footprint appears small over land (around 1 m/s in some places), it represents a significant increase in damage (>10 %). The total rainfall over land during the storm increases by 26 %.

These experiments have reconstructed an intense windstorm in different climates and demonstrated the potential of using a surface-pressure-based reanalysis to develop heavily-constrained storylines which examine the thermodynamic and local dynamical changes in extreme weather events. This approach also offers the opportunity to attribute changes in observed weather events to human influence by translating reanalyses of modern events into cooler worlds. The 20CRv3 reanalysis currently ends in 2015 but could be extended to run in near real time to examine extreme events soon after they occur.