Atmospheric sciences colloquia

Back to Listing

Seminar coordinator for AY 2017-2018: Prof. Ryan Sriver (


Trends and Projections of Extreme precipitation over North America

Event Type
Department of Atmospheric Sciences
Room 112 of the Transportation Building
Mar 6, 2013   3:00 pm  
Emily Janssen, Graduate Student, DAS, University of Illinois
Shirley Palmisano

Changes in frequency of extreme precipitation events can affect crops, cause property damage and even loss of life. Thus, understanding current trends and future projections of extreme precipitation events over the United States is of great importance. An increase shown in the number of observed extreme precipitation events over the U.S. during the last 50 years along with CMIP3 model results have been analyzed in a number of studies and assessments, including the USGCRP 2009 National Climate Assessment (NCA). There are a variety of ways to define and analyze extreme precipitation events. The method used here is the “Extreme Precipitation Index (EPI)”, which is a measure of frequency of heavy precipitation events for a given duration and return period. Duration is the number of days over which precipitation accumulated and return (or recurrence interval) is the average number of years between events. Analyses are done for the contiguous U.S. as a whole and as 7 “sub-regions” consistent with the 2009 NCA. Observed measures of extreme precipitation are compared with the newly released CMIP5-based historical simulations. There is a positive trend in 2-day duration, 5-year return EPI percent anomalies for the most recent 4 decades, over the period 1906-2005. The historical multi-model median of CMIP5 simulations (for the continental U.S.) also shows a positive trend in 2-day 5-year EPI percent anomalies for the past for 4 decades, however, with a smaller signal than observations and a large spread among the models. Differences in future CMIP5 projections, from RCP 4.5 and RCP 8.5, are also examined. A clear increase for both RCP scenarios is shown through the year 2100 in the multi-model median EPI with much better model agreement for the future analyses than was found in the historical simulations.

link for robots only