Raquel Barata, Raquel Prado and Bruno Sanso
04/17/2019 10:24 AM
Statistics
We assess the behavior of large-scale spatial averages of surface temperature in climate model simulations and in reanalysis products. We rely on univariate and multivariate Dynamic Linear Model (DLM) techniques to estimate both long-term and seasonal changes in the externally-forced temperature. The residuals capture the internal variability of the climate system and exhibit complex temporal autocorrelation structure. To characterize this internal variability, we explore the structure of the residuals using univariate and multivariate autoregressive (AR) models. The climate model data analyzed here consist of three different types of numerical experiments from phase 5 of the Coupled Model Intercomparison Project (CMIP5): preindustrial control runs, simulations of historical climate change, and decadal predictions. Our focus is on results from one particular model (MIROC5), as well as on two different reanalysis-based estimates of observed changes in climate (from NCEP-2 and ERA-Interim). The climate variable of interest is monthly-mean 2 meter surface temperature over the time period from January 1981 to December 2010, spatially averaged over four different domains (global, tropical, Northern Hemisphere, and Southern Hemisphere). Our results illustrate differences in all components of the climate ``signal'' (the response to changes in external forcings), most notably between the reanalysis products and the model-generated simulations. Despite the differences in underlying factors contributing to variability, the three types of simulation yield very similar spectral estimates of internal temperature variability. This is of particular interest for the decadal simulation runs as influence from initialization might be expected. In general, we find there is no evidence that the MIROC5 model systematically underestimates the amplitude of observed surface temperature variability on multi-decadal timescales -- a finding that has considerable relevance for efforts to identify human-caused ``fingerprints'' in observational surface temperature data.
