Work at the Alfred-Wegener-Institute
Excerpts from my work ...

I did a voluntary internship at the Alfred-Wegener-Institute in september 2020 and since then I have been working there besides my studies with other students and scientists in the section "Dynamics of the Paleoclimate". Together with Gerrit Lohmann and others, I Investigate various outputs from different climate models and performed cluster, EOF and teleconnection analyses, among other things, to capture and compare the differences between the climate models, as well as their time spans. I am also interested in web development and occasionally create websites to illustrate physical processes.

Here are some websites with learning content, which I have created over time:

Daisyworld Orbital Theory of Ice Ages Energy Balance Models Fourcaults Pendulum Brownian Motion Random Systems Growth Model

If you are interested in bias corrections, this may be exciting as well:

April - June 2022
Bachelor Thesis

I wrote my bachelor thesis with the topic "The influence of bias corrections on variability, distribution, and correlation of temperatures in comparison to observed and modeled climate data in Europe" in cooperation with the AWI. gave an introduction to climate modeling, followed by various methods to adjust bias between modeled and observed climate data and implemented them in the programming languages Python and C++. These implementations are presented with simplified examples.

In the second part of the bachelor thesis the analysis and evaluation of the procedures takes place by using different statistical methods, to show and evaluate deviations between modeled, observed and bias corrected time series of the region Europe between 1981-2010.

Bias correction

This bachelor thesis can be viewed here. Currently a technical paper is in planning since the algorithms, which I wrote and published in the meantime, bring a tremendous added value for research, as they are simple and understandable and especially the C++ implementation has an extremely good performance even on large datasets.

The repositories of the Python and C++ implementations developed during this work are listed below:

July 2021
Brownian Motion Website

Since I had already studied random systems before, I now came to the Brownian Motion. For this I created a website, which again offers interactive diagrams for changing parameters. Different configurations can be tested and so the behavior of the Brownian Motion can be understood.

Brownian Motion image

Further down on the website is a diagram, which should show again, how exactly this change from one state to the other happens. This can be found here.

July 2021
Random Systems Website

Once again, I have prepared a website for a course on the topic of "Random Systems". There are illustrative examples of natural random processes.

Diffusion image

There is the possibility to interact with all diagrams, so that processes like random walks, but also diffusion can be understood. This can be found here.

May - June 2021
Daisyworld Model Website

Gerrit Lohman teaches the course "Modeling of the Earth System" at the university of Bremen and I have implemented an illustrative example of the Daisyworld model in javascript for him and his students.

The Daisyworld model was first published in 1983 by Andrew J. Watson & James E. Lovelock in "Biological homeostasis of the global environment: the parable of Daisyworld" and describes a theoretical cloudless planet with negligible greenhouse gases, inhabited only by black and white daisies with different albedos.

Daisyworld image

It illustrates how a change in one environmental condition can cause a change in a second condition, which in turn can change the first condition in a feedback loop. In this case, the feedback loop relates to the different albedo of the different vegetation types.
This can be found here.

since 2021 January
Clusteranalysis based on correlation matrices of LGM, MIS3 and PI data

Together with Gerrit Lohmann, I am investigating climate data derived from different simulations for different periods of earth history. This includes simulations of the last ice age, MIS3 and preindustrial data.

cluster image

The annual mean temperatures of the last 200 years of the model runs are of interest here. Taking into account significance tests, several results on the differences of the individual models were obtained.
With the help of correlation matrices, similarly correlated locations are assigned to the same clusters, whereby we focus exclusively on the surface temperatures. It can be seen that here, as in the analysis with the european data, clusters are increasingly found on the water.

mean anomalys image
September - October 2020
Comparison of output data from global and regional climate models for the continent of Africa in the period of 1950 - 2055

This was my first project that I did at the Alfred-Wegener-Institute. It was primarily about learning how to handle and analyze data from climate models. I learned and applied different methods to analyze regional and global climate data over the continent of africa. I focused on mean temperatures and precipitation. It was amazing to me how much information you can get from such a multidimensional dataset in netcdf format.

I did this analysis mainly with python, but also learned how to use cdo. The result of my analysis was that the regional data, as expected, had much more accurate point values than the global data. This is clear because the regional data have a much higher resolution, but this was a very good and important exercise in dealing with climate data.

since March 2022
Growth Model of the atlantic cod

For a colleague at the AWI I created a website where students and interested people can try out a growth model for the atlantic cod. Here, as on many other of my websites, various parameters can be set so that the development of the fish's weight over time can be displayed. The model used in this process and the subsequent analyses are recorded in a technical document, for which I have also created various diagrams.

This website can be found here.

since July 2021
Teleconections in several datasets

Some time ago I compared data from different models of the european region and in the course of my work I developed a method to detect and visualize teleconnections between areas that are strongly negatively correlated to each other. This method is based on the the one described in "Teleconnections in the geopotential height field throughing the northern hemisphere winter" by John M. Wallace and David S. Gutzler

I have referred mainly to european data in this analysis, but the picture presented here is the result of the analysis on annual mean temperatures over a 30-year period in the global datasets.


In the analysis, for example, it was noticed that the minimum correlation for each location (one-point correlation map) for the european bias corrected data ranged from -0.19 to 0.4. All other models have much stronger negative correlations. This suggests that the bias correction procedure has too strong an effect on the data.

April - May 2021
Energy Balance Models Website

During my internship at the Alfred-Wegener-Institute I was studying energy balance models and came across the paper "How Climate Model Complexity Influences Sea Ice Stability". I rewrote the models described there in javascript to make them easy to use from inside the browser.

Since the operation is quite intuitive and it is quite easy to make changes to environmental parameters and see their effects on the energy balance, this is now being used for the course "Modeling of the Earth System" at the UNI Bremen to give students a visual opportunity to apply what they have learned.

complex ebm contour

The model is an idealized representation of sea ice and climate on an aquaplanet and shows seaonal and latitudinal variations. It is possivle to change It is possible to change a large number of parameters to visualize their influence on the climate. In addition to the general temperatures and the suface enthalpy, the sea ice boundary is also shown in seasonal progression.

complex ebm line

A simple EBM is also implemented on the website, which responds to changes in the input parameters with different plots of curves representing temperatures across the northern hemisphere. This can be found here.

December 2020
Comparison and Analysis of historical global, regional, bias-corrected and observation data for the region Europe in the period of 1970 - 2000

In this analysis, I examined the extent to which bias-corrected data differ from regional and global data. In doing so, the corrected data were compared with both the original global data and the regional data for the Europe region. In addition, comparisons with observational data was also made.

First, I looked at the general and seasonal distributions and examined the data sets across the spatial and temporal scales for correlation. This was done globally, but also at the local level. I then asked myself at which temperatures the highest differences between the models occur and examined this at maximum, minimum and mean temperatures.

distributions 2

Finally, i performed a cluster analysis based on correlations. Similar correlating locations are assigned to the same clusters. Because of the different distributions of the clusters, the question arose as to whether there were more clusters on the water or on the land. An examination of this has shown that, taking into account the different cell sizes, an increased number of clusters are found on the water than on the land. This indicates a higher variability of temperatures on water.

cluster image