In her doctoral thesis, Stefanie Penzel developed new methods for modelling measurement uncertainties and fuzzy classification in order to improve the reliability of on-site optical analysis and the assessment of water quality.
On 14 September 2026, Stefanie Penzel successfully defended her doctoral thesis entitled ‘Modelling measurement uncertainties using data-driven fuzzy methods for optimised water monitoring in on-site analysis’. The collaborative PhD programme was carried out in partnership with Chemnitz University of Technology, HTWK Leipzig and the Helmholtz Centre for Environmental Research (UFZ) in Leipzig.
The doctoral thesis
In view of climate change, environmental pollution and other anthropogenic influences, the reliable monitoring of water quality is becoming increasingly important. Optical in-situ analysis systems enable the water quality to be assessed in real time directly at the water body, without the need for time-consuming sample transport and laboratory analyses. However, environmental factors such as temperature fluctuations, turbidity or interfering substances can affect the measurement data and limit the reliability of the analysis. In her PhD thesis, Stefanie Penzel investigated how these measurement uncertainties can be systematically recorded, mathematically modelled and taken into account during evaluation, in order to improve the reliability of optical measurement methods and create a robust basis for optimised water monitoring.
Research for more reliable water monitoring
The thesis focuses on a sensor system that combines UV/Vis and fluorescence spectroscopy and is integrated into an immersion probe. With the aid of specially developed algorithms for temperature correction, spectral filtering and turbidity compensation, the quality of the measurement data has been significantly improved. For example, the error in the analysis of the fluorescence spectra was reduced from 71.68 µg/l to 4.37 µg/l through temperature correction. For the UV/Vis spectra, the corresponding error was minimised from 0.5935 mg/l to 0.0218 mg/l through turbidity compensation.
On this basis, Penzel developed a systematic approach to feature extraction that takes the entire spectrum into account. Using an iterative Gaussian approach, characteristic spectral properties are identified which serve both substance identification and concentration determination. The features derived from this also form the basis for the systematic determination of measurement uncertainties in accordance with the international ‘Guide to the Expression of Uncertainty in Measurement’ (GUM).
A particular focus of the work was on the mathematical refinement of the fuzzy pattern classifier. Its model structure was extended so that the previously determined measurement uncertainties are taken into account not only during model building but also during the classification of unknown measurement objects. By combining spectral feature extraction, quantified measurement uncertainties and fuzzy classification, it is possible to make more reliable statements regarding the identity and concentration of substances, even under varying measurement conditions. Finally, the overall concept for the detection of potentially harmful algae was expanded and tested in real water bodies in the Leipzig area. Among other things, a water body infested with cyanobacteria was successfully identified.
The results show that the combination of measurement uncertainty modelling, data-driven pre-processing and fuzzy classification can significantly improve the reliability of optical water analysis and lay the foundations for more precise water monitoring. Further research is planned as part of a forthcoming research visit to the Massachusetts Institute of Technology (MIT) in the USA. There, Stefanie Penzel will further her research into the modelling of measurement uncertainties in environmental monitoring.
Congratulations
The Faculty of Engineering warmly congratulates Stefanie Penzel on the successful completion of her doctoral thesis and wishes her all the best and every success in her future career and personal life.
