Hurricanes - WorldWeatherAttribution/wwa-wiki GitHub Wiki
Attribution of tropical cyclones
-
Attribution of tropical cyclones (TCs) can now include rainfall, wind speeds and the environmental conditions leading to the event, including potential intensity and sea surface temperatures. In any given case the variables studied should be selected to match the impacts of an event.
-
Seasonality is very important, especially where cyclone seasons may overlap with monsoon seasons (such as in the North Indian Ocean) or even interact with monsoon systems (such as in parts of the western North Pacific). In some regions, ENSO also plays a major role in TCs and should be checked where the standard WWA protocol is applied.
-
Rainfall: use standard world weather attribution protocol combining observations and climate models, ideally of 25 km resolution or less, and certainly not more than 50 km. CORDEX, FLOR and AM2 models are commonly used. The event definition should be limited to land regions where possible to increase the number of observational datasets that can be used, though this may not be feasible in island regions such as the Philippines.
-
Potential Intensity: select an area around the track of the storm and average for the month in which the storm occurred. This is usually calculated using ERA5 and CMIP6 models and modelled using a normal distribution that shifts with GMST.
-
Sea Surface temperatures: can be attributed using the same method as potential intensity or using the ‘Climate Shift Index: Ocean’ provided by Climate Central, described here. The latter provides an estimate of likelihood and magnitude changes due to climate change locally at 6-hour intervals along the storm track. As such, it enables reporting of both peak and average likelihood and magnitude shifts.
-
Wind speeds are attributed using the Imperial College Storm Model (IRIS), described here and applied here. This is a stochastic model of storm tracks and intensities, based on historical tracks, estimating landfall intensities by modelling the decay rate from peak intensity. It does not rely on climate models and instead uses the estimated zonal mean change in potential intensity.
IRIS gives a return curve for a given region around (typically within 2 degrees) of the event’s landfall. This enables estimation of the return periods of the event in the current and preindustrial climate and the change in magnitude for a given return period event. It is not possible to combine these results into an overarching statement. However, it is important to describe the compound nature of such events and consistent or inconsistent signals between the different facets are important. Furthermore, when discussing the change in wind speeds, it is important to highlight the non-linear relationship between magnitude and impacts.