Assessing Extreme Events for Anthropogenic Influence: Examples of Recent Cases for Australian Temperatures, U.S. Precipitation, and Hurricane Sandy

Thursday, 18 December 2014: 4:15 PM
Thomas R Knutson1, Fanrong Jenny Zeng1, Andrew Thorne Wittenberg1, Phil Duffy2, J R Arnold3, Chris Massey4, Michael F Wehner5, Dáithí A Stone6, Morris Bender1 and Matt Morin1, (1)NOAA Princeton, Princeton, NJ, United States, (2)Lawrence Livermore National Laboratory, Livermore, CA, United States, (3)U.S. Army Engineer Institute for Water Resources, Univ. of Washington, Seattle, WA, United States, (4)U.S. Army Engineers Research and Development Center, Coastal and Hydraulics Laboratory, Vicksburg, MS, United States, (5)Lawrence Berkeley National Laboratory, Berkeley, CA, United States, (6)Lawrence Berkeley National Lab, Berkeley, CA, United States
The degree to which particular extreme weather and climate events are assessed as being attributable to anthropogenic climate change (e.g., that anthropogenic forcing influenced their probability of occurrence or other characteristics) can vary dramatically from case to case.

One example assessed at GFDL is the record or near-record annual mean temperature over a large region of Australia in 2013. According to this analysis of the CMIP5 models, the event was largely attributable to anthropogenic forcing of the climate system. Another 2013 case was the extreme positive annual mean precipitation anomalies in 5x5 degree gridded (GHCN) precipitation data that were observed along the U.S./Canadian border region. This is a region with a detectable long-term increase of precipitation since 1900. Nonetheless, the 2013 event is assessed as primarily attributable to internal (unforced) climate variability and only partly attributable to external forcing (natural and anthropogenic combined). There are many outstanding challenges to these studies. Among these are the limitations to historical data length, data quality, model ensemble size, and model control run length. Furthermore, there is room for improvement in addressing model biases, station/gridcell scale mismatches, modeling the extreme ends of the distributions e.g. with Generalized Extreme Value methods, etc.

Another project assesses anthropogenic influences on the track and evolution (but not the likelihood) of Sandy-like storms. Assuming the existence of a Sandy-like storm under non-industrial conditions, we use CMIP5 model simulations, a global atmospheric model time slice experiment, and regional hurricane model idealized simulations to suggest that the unusual left turn the storm took may have been made more likely by anthropogenic climate forcing. This does not imply that Sandy-like events are less likely in the non-industrial climate, because we assumed the existence of such a storm to begin with.