Statistical trend detection of a global change signal in regional climate over the U.S.
Airong Cai, Univ. of Illinois, Chicago, IL; and K. Hayhoe, G. C. Tiao, and D. J. Wuebbles
Predicting near-term (5 to 30 year) regional climate is inherently difficult, despite the fact that the latest scientific understanding suggests that in many regions, temperatures already regularly exceed the observed historical mean due to global-scale changes in climate that are occurring as a result of anthropogenic activities. Even accounting for the impact of atmospheric circulation indices on regional climate variability, historical model simulations essentially display little correlation with deseasonalized observed climate data in certain locations such as the U.S. Midwest. How long will it take before the upward trend in temperature – already identified at the global scale through pattern-based detection and attribution techniques – becomes significant at the regional level across the U.S.? Over what time horizon will decision makers from various regions need to take into account climate change projections for that region? These are the questions we address through application of trend detection techniques to observed historical data and model-simulated historical and future climate projections in order to estimate for the six primary U.S. regions (NE, Midwest, SE, Gulf Coast, SW and NW): (1) the year in which a significant trend in seasonal mean temperature and/or precipitation is likely to either have been or be detected in that region, and (2) the slope of the trend over this century and the degree to which the slope depends on the emission scenario used to drive the simulations.
Poster Session 1, Observed climate change
Monday, 30 January 2006, 2:30 PM-4:00 PM, Exhibit Hall A2
Previous paper Next paper
Browse or search entire meeting
AMS Home Page