Task 2.5 Combining Predictions and Projections to Assess Predictive Skill of Heat Extremes

There is an increasing demand for consistent climate information for up to a 30-year horizon.
However, users of climate information have a number of options for obtaining data from global
climate models with different temporal resolutions for the next few years and decades. This data is
sometimes based on different sources and models and is generally generated independently, which
can lead to inconsistencies in the respective information content when this data is viewed together.
For example, the current set of scenario-based global climate projections, as produced for CMIP6,
can be supplemented with initialised global decadal climate forecasts, which in turn are produced
periodically and annually by multiple modelling centres around the world. However, since their
forecast lead time ends at 10 years, projections are not a seamless continuation of decadal climate
forecasts, any inconsistencies in the overlap periods between decadal climate forecasts and multidecadal climate information must be corrected so that users can receive seamless and consistent
forecast information.

In recent years, some successful methods for combining and merging climate projections with
decadal climate forecasts have been developed in this regard (e.g. Befort, 2020; Mahmood et al.,
2021, Befort et al., 2022). Now basically, these methods show that consistent climate information,
sometimes with the respective forecast quality, can be obtained over a period of 20 years. Here, we
investigate to what extent the combination of observations, decadal forecasts and climate projections can be used to provide a seamless estimate of extreme temperatures and associated heat stress metrics for a period of up to 30 years. We apply different large ensembles of initialised as well as non-initialized (driven with historical and scenario-based constraints) experiments. We focus on existing ensembles from the Max-Planck-Institute Earth System Model (MPI-ESM) and new
experiments with ICON-XPP. The combination should first be checked globally for mean
summer temperatures with a focus on Europe and Germany. However, we will expand our
assessment to include metrics of individual and combined heat events. Specifically, these metrics
include multivariate heat stress metrics associated with socioeconomic and health impacts (e.g.,
exceeding muggy heat thresholds; sustained abstinence from nocturnal cooling), and heat stress
indicators such as those already used by the DWD for weather forecasts, e.g. perceived temperature
and associated threshold values (e.g. "health risk heat"). The aim is to create seamless probabilities
of occurrence for extreme temperatures and their stress metrics for a period of up to 30 years.

Contributors

Max Planck Institute for Meteorology
Dr. Wolfgang Müller
Dr. Ned Williams

Karlsruhe Institute for Technology

University of Hamburg

Deutscher Wetterdienst
Dr. Clementine Dalelane
Dr. Andreas Paxian