Datensatz

Zusammenfassung

In this dissertation I investigate the temporal development of internal
variability under global warming. Understanding internal variability
is essential to understand the past and possible future trajectories of
our climate, yet it is often assumed to be a property of the climate
system that does not change under global warming. I use a novel large
ensemble, the Max Planck Institute Grand Ensemble, to introduce a
new perspective on internal variability.
Internal variability can be described as the seemingly random fluc-
tuations of the climate system over time. Due to nonlinearity in the
climate system, small perturbations may grow to large anomalies over
time that are associated with anomalous or even extreme events. In
my first chapter, I quantify internal variability and investigate whether
it changes under global warming. The change in the external forcing is
the same for all of these realisations, the initial conditions are different
for each realisation. Thus, each realisation follows its own, unique tra-
jectory. For each time step, the distribution of all realisations provides
an estimate of the possible states of the climate system.
I develop an analysis framework based on a large ensemble to detect,
quantify and attribute changes in internal variability in a transient
climate. Rather than analysing variability over time, I use the ensemble
dimension of a large ensemble to quantify internal variability. This
approach allows a clean separation of the forced signal from internal
variability and ensures stationarity of the statistics even when the
forcing is changing with time. My non-parametric approach provides
an objective quantification of changes in internal variability and their
robustness.
In my second chapter I apply this analysis framework to investigate
rainfall in the tropical Atlantic region in the past and its possible
future trajectories. I can show that simulated internal variability in
the Sahel encompasses all observed values for the
20th century. The
model suggests an externally forced increase in rainfall towards the
end of the
20th century. However, due to large internal variability,
it is not possible to detect this forced change in a single realisation.
In future projections, I find an increase in the mean rainfall over the
Sahel, accompanied by an increase in the variability. This implies
that the average rainfall will increase, but individual years may show
deviations from this mean value that are larger than under present-day
conditions.
In the tropical Atlantic region, most state-of-the-art coupled climate
models show large biases in the simulated sea surface temperature
and rainfall when compared to observations. These model errors