|
Activity Number:
|
71
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, July 31, 2016 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
|
Abstract #318595
|
View Presentation
|
|
Title:
|
Bivariate Quantile-Based Calibration of Numerical Model Outputs with Application to Climate Projections
|
|
Author(s):
|
Brooke Alhanti* and Howard Chang
|
|
Companies:
|
Emory University and Emory University
|
|
Keywords:
|
climate ;
quantile ;
calibration ;
copula ;
bayesian
|
|
Abstract:
|
Climate models are mathematical representations of climate that are inherently biased compared to observations. There is evidence calibrating variables outputted from climate models with historically observed values improves predictability. Traditional calibration methods use univariate approaches that can perform poorly at distribution tails. Here we use copulas to perform a bivariate quantile calibration method that simultaneously calibrates the entire distribution while capturing dependence between variables. This method estimates bias between quantile functions for climate models and monitoring data and applies this estimated bias to future climate projections. The Gumbel and Frank copulas were used to estimate the dependence between two climate variables: daily average temperature and daily total solar radiation. We apply our method to projections from four different climate models under the same emission scenarios in Atlanta and and evidence for higher mean temperature and lower mean solar radiation for the period 2041-2070 compared to 1991-2000. Our results indicate that calibrating climate model outputs can decrease between-model variability in projections.
|
Authors who are presenting talks have a * after their name.
