Uncertainty Quantification for Regional Climate Model Output Using Space-Time Self-Exciting Point Processes

• Activity Number: 643
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Souparno Ghosh*+ and Alan Gelfand and Stephan Sain
• Companies: and Duke University and Geophysical Statistics Project
• Keywords: Climate change ; NARCCAP ; Self-exciting point process ; Global climate model ; Regional climate model

Abstract

We analyze the simulated daily precipitation output obtained from several regional climate models (RCMs) using a space-time self-exciting point process model. We consider precipation times as a point pattern. Then, the proposed model uses the space-time generalization of Hawkes' self-exciting process but incorporates a lag-structure, arguing that the seasonal nature of precipitation supports delayed self excitation. The inferential framework thus developed will enable comparison among the outputs generated by different climate models. Furthermore, this approach may be useful in the North American Regional Climate Change Assessment Program which investigates the variability between the outputs of six different RCMs each driven by a set of four global climate models (GCMs) and a reanalysis model with each GCM-RCM combination being run under the present scenario and a future scenario. Using the framework developed herein, we can compare the outputs obtained from these two model runs in a statistically coherent way. The inference drawn from this comparison can be used to assess the changes in climate under two different scenarios.