Possible topics for B.Sc. and M.Sc. theses in the Bioclimatology Group
In order to make the assignment and supervision of B.Sc. and M.Sc. theses in our Bioclimatology Group more transparent and fairer, the following application procedure should be applied.
Theses Day
Once every semester, we invite all Bachelor and Master students that are interested in writing a thesis in the Bioclimatology Group to our Theses Day. On this day, we supervisors introduce ourselves, present ideas for thesis topics, possible research questions, and the necessary methodology. Please find below the dates for the upcoming Theses Days and sign up for it under stud.IP (Course: Bioclimatology Theses Day).
How It Works
If you are interested in a specific thesis topic, you can apply for it by writing an email to the respective supervisor (within one month after Theses Day, see deadline below). This email may already include your motivation, possible research questions and hypotheses, your skills, and a rough time schedule.
The supervisor will organize a meeting with all interested students to discuss the details and (if needed) select a candidate. Feedback by the supervisor will be provided shortly thereafter.
More information about the application, possible topics, and the general supervision will be given on the Theses Day.
| Writing thesis during... | Thesis Day | Application Deadline |
|---|---|---|
| ... summer semester 2025 | Jan 14, 2025, 10:15-11:30 FSR 2.7, Büsgenweg 2 |
Feb 07, 2025 |
| ... winter semester 2024/25 | Jul 11, 2024, 14:00-15:00 FSR 2.7, Büsgenweg 2 |
Aug 01, 2024 |
Possible Topics
Some topics listed below can be worked on as B.Sc. or M.Sc. thesis and thematically adapted accordingly. The thesis can be written in German or English in our group. Own suggestions for topics are also always welcome. Please, also note that each supervisor will only supervise 1-3 theses per semester.
Please find more information here.
Drought has emerged as a significant stressor for forest ecosystems in Central Europe, with events since 2018 highlighting their vulnerability to extreme climatic conditions. These events have had profound impacts on forest vitality, such as declining canopy health, increased tree mortality, and shifts in species resilience. Understanding these effects requires a comprehensive assessment of how drought is defined, measured, and studied within the scientific community.
This study combines a meta-analysis and literature review to evaluate the methodologies used to investigate drought impacts on forests. Key aspects include the definitions of drought, the environmental parameters assessed (e.g., drought indices, mortality rates, and canopy condition), and the datasets employed, whether terrestrial or satellite-based. The analysis aims to identify consistent patterns and discrepancies across studies, evaluate the explanatory power of different methodologies, and map regional trends in research density. Special attention is given to the differential responses among tree species and the role of legacy effects from past droughts.
Within the SIGNAL project, we have collected data over the last 5 years from a total of 12 stations, located at 10 different agroforestry and monocropping agricultural systems. We have time series of meteorological data, CO2 and energy fluxes (latent heat LE and sensible heat H). The goal of this thesis project would be to work with the time series of all the sites and perform a multiple regression, based on some Python or R package, to understand which are the main meteorological drivers (radiation, vapor pressure deficit, etc.) for the different flux components (CO2, LE, H). The relations will be evaluated at different scales, from the multi-year to the seasonal and monthly scale. Some initial hypotheses can be evaluated, such as the following: solar radiation is the main driver of fluxes in both agroforestry and monocropping systems; vapor pressure deficit is a more important driver of fluxes in the monocropping, due to the absence of buffering effect of trees in temperature and relative humidity; etc.
No data collection will be performed during this thesis. The data to be used will be provided. An initial goal could be to analyze first only two adjacent field sites (one agroforestry and one monocropping) and perform the multiple regression across the whole available time period. Depending on time and performance of the student, the analysis can be extended further beyond.
Requirements:
- Basic/medium programming knowledge (preferably Python, but any other language is welcome)
- Basic courses in bioclimatology (interactions ecosystem-atmosphere) or in meteorology
An important feature when studying ecosystems’ carbon cycle is the partitioning of Net Ecosystem Exchange (NEE), gathered from eddy covariance measurements, into the photosynthesis (Gross Primary Production, GPP) and ecosystem respiration (Reco) components. Furthermore, when evaluating carbon fluxes above an ecosystem, it is important to understand the distribution of sources and sinks of CO2, especially if the ecosystem is heterogeneous. Land surface heterogeneity induces irregular spatial distribution of sources and sinks of CO2.
In the SIGNAL project, we have collected data over the last 5 years from a total of 12 stations, located at 10 different agroforestry and monocropping agricultural systems. We have time series of meteorological data, CO2 and energy fluxes (latent heat LE and sensible heat H). All the agroforestry sites and some of the monocropping sites are heterogeneous, with different crops being grown together and different management practices happening in different areas around our measuring stations.
The goal of this MSc thesis would be to run a simple footprint model (from Kljun et al., 2015) corresponding to a time series of few months during the growing season, for two of the sites in SIGNAL, two adjacent agroforestry and monocropping sites, and relate the footprint information to the magnitude and the sign of the carbon fluxes. The time series of meteorological data, NEE, GPP and Reco will be provided. By identifying specific periods and classifying data in wind directions, the student will be able to separate the behavior of specific parts of the ecosystem and to understand the relevance of those periods in the whole ecosystem development.
Requirements:
- Basic/medium programming knowledge (preferably Python, but any other language is welcome)
- Basic courses in bioclimatology (interactions ecosystem-atmosphere) or in meteorology
- Basic/medium skills in working with spatial information (to combine footprint modeling with flux information) – either using a programming software or some GIS processing tool
The basic assumptions of the eddy covariance technique and the intrinsic difficulties in its deployment cause always missing data in the time series of CO2, latent heat and sensible heat fluxes. In order to perform year or multi-year carbon and water balances, different techniques are employed to gap-fill the time series. Despite the large acceptance of some methods in the community, their robustness is compromised when the gaps in the time series are very long (few weeks or months), and especially if the gaps are not well distributed across the year, e.g. a lot of missing data in winter and not in summer. Due to this, different machine learning algorithms have been used in the last years to provide more robust gap-filling.
In the SIGNAL project, we have collected data over the last 5 years from a total of 12 stations, located at 10 different agroforestry and monocropping agricultural systems. We have time series of meteorological data, CO2 and energy fluxes (latent heat LE and sensible heat H). Meteorological data have been filled by using the nearby German Weather Service (DWD) stations. Flux data have been filled by using the Extreme Gradient Boosting algorithm, adapted from the code used in Vekuri et al. (2023). The performance of the algorithm seems satisfying, in the sense that it keeps diel and seasonal patterns and that it understands well the relations between the meteorological drivers and the fluxes. However, this algorithm does not provide an uncertainty estimation for the filled data. Using the root mean squared error (RMSE) or the mean bias in the 30-min period is not robust enough, because we have very long gaps.
The goal of this MSc thesis would be to develop...
- ... a quality flag for the filled data, assigning a coding system (such as from 0 to 3) to the filled fluxes depending on their reliability. This reliability would be tested depending on how long the gap is, how much information is there around the specific gap being filled, whether the meteorological data used to drive the gap-filling in that specific period are measured or filled, etc.
- ... a more robust uncertainty estimator, based on the quality flag, which is independent on the length of the time series, but can be propagated when performing cumulative sums for annual or multi-annual budgets of carbon or evapotranspiration.
Requirements:
- Basic/medium programming knowledge (preferably Python, but any other language is welcome)
- Basic courses in bioclimatology (interactions ecosystem-atmosphere) or in meteorology
In the context of this work, transpiration should be measured at the agricultural experimental farm Reinshof (i) by means of hand measurements with a photosynthesis device (LI-6800) and (ii) the water fluxes measured on leaf scale should be scaled to the ecosystem and compared with direct eddy covariance measurements. The measurements should be carried out under different environmental conditions over the course of a growing season. The results should give an indication of the reliability of energy fluxes measured at the Reinshof site.
Various approaches to check and assess the quality of meteorological measurements of our study sites should be compared. This study will include a literature research for the various approaches, the application of these approaches with the R or python programming languages to the existing measurements, and a thorough comparison of the filtered data. For the various meteorological variables a different combination of approaches can be of advantage.