Evaluation of Interannual-to-Decadal Climate Variability in CMIP6 Models Using the New “Climate Variability Diagnostics Package for Large Ensembles”

Abstract:

Evaluation of interannual-to-decadal climate variability in CMIP6 models is a challenge due to the brevity of the observational record and confounding of external forcing and internal variability on regional scales. Here, we introduce a new automated tool for evaluating modes of variability in CMIP6 models called the “Climate Variability Diagnostics Package for Large Ensembles” or CVDP-LE. This tool facilitates model evaluation and inter-comparison by providing a suite of diagnostics on the major modes of internal variability on interannual-to-decadal timescales in the ocean, atmosphere and sea ice. The diagnostics are saved to a data repository for later usage, and provided in succinct graphical and netcdf formats. The CVDP-LE operates on a user-specified set of observational and model data sets, including Obs4MIPS and CMIP6 initial-condition Large Ensembles, respectively. A key feature of the CVDP-LE is the ability to place the single observational record within the context of model ensemble-spread. This is crucial in order to avoid confounding model structural uncertainty with uncertainty due to limited temporal sampling, and essential for proper model evaluation and inter-comparison. In our presentation, we shall illustrate the salient features of the CVDP-LE as applied to the CMIP6 archive, highlighting overall model performance based on 11 key metrics of internal climate variability.