The consistency between climate model output and corresponding observations can be evaluated by comparing modeled and observed time series. We propose to measure this consistency with a likelihood: if the climate system behaves as a climate model simulates it, what is the probability that an observed value of a given statistic would be obtained? The likelihood is calculated using a Monte Carlo procedure to estimate the sampling distribution of the selected statistic from the model time series, and then locating the observed value of the statistic in that distribution. In this talk, we demonstrate the procedure using specific humidity time series data from CMIP5 model runs and observed data from JPL's AIRS instrument.

download presentation slides ›

Dr. Amy Braverman is a Senior Statistician in JPL’s Science Data Understanding Group. Her research interests lie in many areas of applying statistics to climate and Earth science, including the development and implementation of new data reduction methodologies for massive datasets based on information-theoretic principles, and the development of statistical methods to compare climate model output to observations. Dr. Braverman received her B.A. in Economics from Swarthmore College in 1982, and worked in litigation support consulting before earning an M.A. in Mathematics and a Ph.D. in Statistics from UCLA in 1992 and 1999 respectively. She was a Caltech postdoctoral scholar at JPL until 2001, and since July 2008 she is an Adjunct Associate Professor of Statistics at UCLA in addition to her role at JPL.