Skip to main content
U.S. flag

An official website of the United States government

Publication Date
1 January 2021

Anthropogenic Influence on Hurricane Dorian's Extreme Rainfall

Hindcast attribution simulations suggest that anthropogenic climate change increased the likelihood of Hurricane Dorian's extreme 3-hourly rainfall amounts and total accumulated rainfall.
Print / PDF
Powerpoint Slide

The work explores the potential impact of climate change on the rainfall associated with Hurricane Dorian using the CAM5 hindcast attribution framework. The analysis indicates that the simulated extreme 3 hourly rainfall rates associated with Hurricane Dorian increase in both frequency and magnitude due to climate change. The likelihood of 3 hourly rainfall accumulations above 0.136 m (observed) increases by 16%.


Tropical cyclones are very costly natural disasters due to a diverse set of impacts, including high winds, extreme rainfall, storm surge, and fresh and/or saltwater flooding. A recent review estimates that the global mean near-storm rainfall increases at the Clausius-Clapeyron rate of about 7% per 1°C. This work continues the authors’ advances in attribution frameworks to quantify the effect of climate change on individual hurricanes. Investigations of individual storms using various attribution methodologies suggest that changes in rainfall can exceed the Clausius-Clapeyron estimate, although there are uncertainties associated with the use of different rainfall metrics.


This work uses the variable-resolution configuration of the Community Atmosphere Model version with a 28-km nest over the North Atlantic. Following our previous methodology, short 7-day ensemble hindcasts are initialized both in advance of and after Hurricane Dorian's landfall in the Bahamas at 12-h increments starting at 1200UTC 30 August and ending at 0000 UTC 4 September for a total of 10 initialization times. The CAM5 hindcasts are initialized to construct an ensemble under "actual" and counterfactual climate and weather conditions. Twenty ensemble members are created at each initialization time by perturbing a set of three parameters in the deep convection parameterization. The "counterfactual" ensemble is constructed by removing the anthropogenic signal from the air temperature, specific humidity, and sea surface temperature (SST) initial conditions used for the actual ensemble. The anthropogenic signal is approximated by computing the difference between the All-Hist and Nat-Hist CAM5 simulations completed under the DOE RGMA sponsored Climate of the Twentieth Century (C20C+) Detection and Attribution Project (available at for the average of August and September for the last 20 years (1996-2016).

Point of Contact
Michael Wehner
Lawrence Berkeley National Laboratory (LBNL) - Computational Research Division
Funding Program Area(s)