While several attribution studies have quantified the influence of climate change on tropical cyclone rainfall, very few have investigated tropical cyclone winds because most climate models can not accurately simulate strong storms. IRIS’s innovative methodology provides a solution to this problem by modelling synthetic tropical cyclones and applying the concept of potential intensity.

Potential intensity is a powerful metric that predicts the maximum possible wind speed. It is calculated from the sea surface temperature, air moisture, and air temperature. It will vary by location and month.

To understand the potential intensity of these tropical cyclones in a counterfactual, preindustrial climate without human-caused warming, recent decadal trends in potential intensity are extrapolated backwards. This backwards shift is only small as the global mean temperature in the 1980s, the earliest weather data is very high-quality, was only 0.4°C warmer than pre-industrial climate, compared to 1.2°C today. 

The difference in the storm intensity and likelihood of this storm intensity between the counterfactual climate and today’s climate can be attributed to climate change. The model uses historical tracks to simulate realistic landfall locations. 

For example, an IRIS attribution analysis on Typhoon Haiyan, a Category 5 cyclone that hit the Philippines in 2013, found the typhoon was extremely unlikely without climate change. 

According to IRIS, in a cooler climate without human-caused warming, typhoons as powerful as Haiyan are expected to hit the Philippines about once every 9,300 years, but in the climate with 0.8°C of warming, similar typhoons are expected to occur about once every 130 years. 

The intensity changes predicted by IRIS is similar to full physics numerical model studies of Haiyan such as Takayabu et al., 2015 and Delfino et al., 2023, and will be described in a forthcoming paper submitted to Atmospheric Science Letters. A preprint is available here: Climate Change Attribution of Typhoon Haiyan with the IRIS model (PDF).