Topic:Combined Score Equations for Spatial Extremes Modeling with Application to Detection and Attribution of Climate Change
Speaker:Jun Yan, University of Connecticut
Time:Monday, June 25, 14:00-15:00
Place:Room 217, Guanghua Building 2
Detection and attribution analysis for changes in climate extremes is a major concern in climate research. The widely used optimal fingerprinting method for changes in average climate states has no satisfactory analog in climate extremes. We propose two marginal and one joint modeling method in the framework of extreme value analysis, where the fingerprints of specific external forcings estimated from climate model simulations are incorporated into the location parameter of the generalized extreme value (GEV) distributions of observed extremes. The combined score equation (CSE) method combines the score equations of the marginal GEV models through a weight matrix to improve efficiency. Under working independence, it reduces to the independence likelihood method, which is shown to be robust to uncertainty in signal estimation. The joint modeling method specifies the dependence structure with a max-stable process and estimates the parameters with the pairwise likelihood. The performances of the estimators were compared in simulation studies. The marginal approaches were applied to detection and attribution analyses of four annual daily temperature extremes in East Asia, in which the anthropogenic impact was detected in the observed changes after controlling the natural climate variation.
Dr. Jun Yan is a Professor in the Department of Statistics, University of Connecticut. He received his Ph.D. in statistics from University of Wisconsin--Madison in 2003 and was an assistant professor at the University of Iowa before moving to UConn in 2007. Dr. Yan's methodological research interests include survival analysis, clustered data analysis, spatial data analysis, spatial extremes, estimating functions, and statistical computing. He has been involved in a number of collaborative research projects in climate change, animal ecology, and public health. He is committed to making his statistical methods available via open source software, coauthoring and maintaining a collection of R packages in the public domain.
Your participation is warmly welcomed!