1. Abstract
Master Thesis: “Changes of meteorological anomalies in context of the recent climate change
in South Bavaria”
Subjective perception leads to the impression of an increasing amount of extreme weather events such
as heat waves, heavy precipitation and storms. Especially in the last two decades where the topic of
climate change has been highly visible in the media the coherence seems only logical. Also statistical
theory implies a link between a small shift in the mean resulting into higher variability and the
frequency of such extreme events. This thesis therefore aims for detecting significant changes in the
frequency and intensity of meteorological extremes in the last century which can be attributed to the
recent climate change.
For this purpose six indices for temperature, precipitation and wind extremes are analyzed. These
indices have been calculated from the daily observational data of three stations with a recording
history of at least 60 years in South Bavaria, Germany. Extreme temperatures are defined as daily
events over the threshold of the 90th
percentile of each station. The extreme precipitation indices are
based on the 95th
percentile of the distribution of the daily precipitation data. The choice of the
different threshold is due to the IPCC Report which used similar percentiles in the extreme indices.
The definition of storms is the absolute wind speed of 21 m/sec according to the definition of the
German Climate Service Center. The data has been seasonal resolved.
Two methods are used: the statistical tool of a trend analysis and return levels based on the extreme
value model of the Generalized Pareto Distribution (GPD). The trend analysis is applied for both
frequency and intensity of extreme values. The trends are tested for significance via F-Test, Mann-
Kendall-Test and the approach of the Signal-to-Noise-Ratio. Before fitting the values to the GPD the
data sets of each station are separated into two equally sized time slices. The estimated return values
of the GPD are compared in the next step.
The temperature data show overall a significant change in all three stations in South Bavaria. Extreme
temperature events increase in frequency and intensity. The tendency to warmer winters is especially
striking. Modeling of the extreme temperature data with the GPD also exposes a shift to a higher level
of temperature values. It indicates a higher probability in the return levels of extreme values. The
results in temperature extremes prove to be resilient when compared to similar research in this domain
in Europe. However, in general the application of numerical modeling appears more common than
statistical modeling.
The extreme precipitation values display overall, both in frequency and intensity a positive signal in
spring and autumn. Amixed signal in summer and a negative tendency in winter is observed. However,
the results of the analysis of the extreme precipitation indices show high variability. This applies to
2. the data sets internally as well as compared with each other. So no significant trend can be
acknowledged. The goodness of fit to the GPD is almost as high as in the temperature values and
validates the mixed results. In the regional case of South Bavaria based on the daily observational
data of the selected stations a significant signal change in precipitation anomalies cannot be found.
The current literature agrees in the difficulties when analyzing precipitation. Because of its high intern
variability coherent patterns in the progression of the precipitations extremes cannot be found yet in
Europe.
A similar variability problem occurs in the wind speed indices. No significant trend in either storm
frequency or wind speed in the mixed tendencies can be detected. More observations are needed for
more conclusive results. Current research treating local wind speed data and local progression of
storms is only available to a very limited extent. Testing for reliance is therefore difficult. The
observations in this field of study aim more commonly for the progression of cyclone activity in a
higher scale.
All data sets are tested for homogeneity because of several changes of sensors through the years. The
temperature data appears to be reliable for the greatest part. The data series of precipitation and wind
expose irregularities. In case of precipitation the irregularities could be due to the natural variability.
So no attempt to homogenize the data has been made. However, the time range of the wind data series
is significant smaller. The inhomogeneity has therefore a strong influence on the results and proves
to falsify data to a certain degree. Via standard deviation and long-term average a simple effort to
homogenize the wind data sets has been made. The outcome displays in one case a remarkable change
from a positive to a negative tendency, even though it exposes no significant trend in wind speeds.