Biases Beyond the Mean in CMIP6 Extreme Precipitation: A Global Investigation.

Climate models are crucial for assessing climate variability and change. A reliable model for future climate should reasonably simulate the historical climate. Here, we assess the performance of CMIP6 models in reproducing statistical properties of observed annual maxima of daily precipitation. We go beyond the commonly used methods and assess CMIP6 simulations on three scales by performing: (a) univariate comparison based on L‐moments and relative difference measures; (b) bivariate comparison using Kernel densities of mean and L‐variation, and of L‐skewness and L‐kurtosis, and (c) comparison of the entire distribution function using the Generalized Extreme Value (GEV) distribution coupled with a novel application of the Anderson‐Darling Goodness‐of‐fit test. The results reveal that the statistical shape properties (related to the frequency and magnitude of extremes) of CMIP6 simulations match well with the observational datasets. The simulated mean and variation differ among the models with 70% of simulations having a difference within ±10% from the observations. Biases are observed in the bivariate investigation of mean and variation. Several models perform well with the HadGEM3‐GC31‐MM model performing well in all three scales when compared to the ground‐based Global Precipitation Climatology Centre data. Finally, the study highlights biases of CMIP6 models in simulating extreme precipitation in the Arctic, Tropics, arid and semi‐arid regions. Plain Language Summary: Annual maxima of daily precipitation are widely used to design critical infrastructures such as dams and stormwater networks. Climate change is expected to increase the frequency and intensity of extreme events. Climate model projections offer a glimpse into the future and can help assess potential changes and impacts. It is reasonable to assume that climate models simulating accurately the historical climate may also simulate well the future. Here, we assessed the latest generation of climate models, that is the CMIP6 models, to reproduce the historical annual maximum daily precipitation. We compared simulations and observations of extreme precipitation using advanced summary statistics, novel probability similarity measures, and robust statistical tests. The results indicate that models, in general, reproduce well the behavior of annual maximum precipitation, but biases exist especially in the simultaneous behavior of mean and variance. Shortcomings of CMIP6 models are highlighted in the Arctic, Tropics, arid, and semi‐arid regions. Key Points: Assessment of CMIP6 biases in precipitation extremes using L‐moments, novel probability similarity measures and statistical testsCMIP6 models reproduce fairly well observed annual maximum precipitation but biases exist in the joint behavior of mean and variabilityShortcomings of CMIP6 models are highlighted in the Arctic, Tropics, arid, and semi‐arid regions [ABSTRACT FROM AUTHOR]