An 'Observational Large Ensemble' to Compare Observed and Modeled Temperature Trend Uncertainty due to Internal Variability

Initial condition climate model ensembles show that simulated regional temperature trends can be highly variable on decadal timescales due to characteristics of internal climate variability. In order to contextualize recent observed temperature changes, it is therefore crucial to quantify the role of internal variability in historical trends. Importantly, internal variability simulated by a model may be inconsistent with that in observations, so observation-based methods of studying internal variability are needed as well. Here, we consider trend uncertainty due to internal variability in historical 50-year (1966-2015) winter near-surface air temperature trends over North America, comparing simulated trend variability in the NCAR CESM1 Large Ensemble (LENS) to trend uncertainty in observations. We use a statistical resampling approach to create an `Observational Large Ensemble' (OLENS), which generates spatially coherent fields of temperature trends that might have been observed given a different realization of internal variability consistent with the observations (and consistent with the forced temperature response in CEMS1). The OLENS suggests that it is more difficult to identify a climate change signal in any given CESM1 simulation than in observations. That is, uncertainty in wintertime temperature trends over North America due to internal variability is largely overestimated by CESM1, with 78% of North America showing larger trend variability in the LENS than in the OLENS, on average by a factor of 26%.