Christoph Tholen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Christoph Tholen has a total of 36 co-authors. Here are the top co-authors by number of shared publications:
- Carolin Leluschko
German Research Center for Artificial Intelligence
5 shared publications - Frederic Stahl
German Research Center for Artificial Intelligence
5 shared publications - Lars Nolle
Jade University of Applied Sciences
4 shared publications Lars Nolle
Jade University of Applied Sciences, Wilhelmshaven, Germany
3 shared publications- Oliver Zielinski
Institute for Chemistry and Biology of the Marine Environment, University of Oldenburg, Oldenburg, Germany
3 shared publications - Eike Rodenbäck
German Research Center for Artificial Intelligence, 26129 Oldenburg, Germany
2 shared publications
Latest Publications
- Open Access
- 2 September 2026
- Published by Cantonal and University Library Fribourg in Anwendungen und Konzepte der Wirtschaftsinformatik
- No. 22,p. 7
On the Optimisation of Machine Learning Models for Predicting the Photosynthetically Available Radiation in the Water Column
Photosynthetically Available Radiation (PAR) is a crucial parameter in oceanography. This study explores the optimization of machine learning models to predict PAR in the water column using selected wavelengths of downwelling irradiance. By leveraging Genetic Algorithms (GA), optimal wavelength combinations were identified for two machine learning models: Linear Regression (LR) and Regression Trees (RT). The models were trained on data from the HE533 expedition and validated using datasets from multiple ship expeditions across different geolocations. Experimental results indicate that the LR model, with an optimal wavelength combination of Ed(469), Ed(501), and Ed(600), achieved the highest prediction accuracy (R² = 0.9992, MAE = 5.78). The RT model, using Ed(433), Ed(586), and Ed(687), performed slightly worse (R² = 0.9954, MAE = 16.37). While both models generalised well to unseen datasets, significant prediction errors were observed for small PAR values at lower water depths.
- 2 September 2026
- Published by Cantonal and University Library Fribourg in Anwendungen und Konzepte der Wirtschaftsinformatik
- No. 22,p. 7
Open AccessOn the Optimisation of Machine Learning Models for Predicting the Photosynthetically Available Radiation in the Water Column
Photosynthetically Available Radiation (PAR) is a crucial parameter in oceanography. This study explores the optimization of machine learning models to predict PAR in the water column using selected wavelengths of downwelling irradiance. By leveraging Genetic Algorithms (GA), optimal wavelength combinations were identified for two machine learning models: Linear Regression (LR) and Regression Trees (RT). The models were trained on data from the HE533 expedition and validated using datasets from multiple ship expeditions across different geolocations. Experimental results indicate that the LR model, with an optimal wavelength combination of Ed(469), Ed(501), and Ed(600), achieved the highest prediction accuracy (R² = 0.9992, MAE = 5.78). The RT model, using Ed(433), Ed(586), and Ed(687), performed slightly worse (R² = 0.9954, MAE = 16.37). While both models generalised well to unseen datasets, significant prediction errors were observed for small PAR values at lower water depths.
- Open Access
- 1 September 2026
- Published by Elsevier in SSRN Electronic Journal
Spatial and temporal variability of beached plastics near river mouths assessed using UAS surveys and machine learning
Within the context of increasing plastic pollution, rivers act as the main source of these materials into the seas and oceans. A better understanding of their role as drivers of plastic dispersal is still necessary. A selected part of the northeastern coast of Sicily (southern Italy) is used as a case study to fill this gap. Uncrewed aerial system surveys combined with machine learning-based analysis are used to detect and quantify plastic accumulated on beaches following flash flood events and to assess the associated depositional mechanisms. Polystyrene foam was the dominant beached plastic across all datasets, accounting for the highest number and largest total accumulated size. Polyurethane foam and polyethylene terephthalate bottles were the next most frequently recorded plastics. Spatial analysis revealed that plastic concentration was highest immediately adjacent to the river mouth and decreased with distance, consistent with the river acting as the main source of plastics during flash floods, although simultaneous wave action disrupted this pattern at some sites, redistributing plastic along the shoreline. No relationship was observed between plastic size and distance from the river mouth, though polystyrene foam size varied across rivers and years, reflecting differences in fragmentation associated with fluvial energy, wave action, and dry period duration. Drainage basin size and population density also influence the amount of accumulated plastic. This study shows that flash flood-driven plastic accumulation displays significant spatial and temporal variability across rivers and events. A deeper understanding of these processes can lead to more precise predictions and solutions for plastic pollution.
- 1 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSpatial and temporal variability of beached plastics near river mouths assessed using UAS surveys and machine learning
Within the context of increasing plastic pollution, rivers act as the main source of these materials into the seas and oceans. A better understanding of their role as drivers of plastic dispersal is still necessary. A selected part of the northeastern coast of Sicily (southern Italy) is used as a case study to fill this gap. Uncrewed aerial system surveys combined with machine learning-based analysis are used to detect and quantify plastic accumulated on beaches following flash flood events and to assess the associated depositional mechanisms. Polystyrene foam was the dominant beached plastic across all datasets, accounting for the highest number and largest total accumulated size. Polyurethane foam and polyethylene terephthalate bottles were the next most frequently recorded plastics. Spatial analysis revealed that plastic concentration was highest immediately adjacent to the river mouth and decreased with distance, consistent with the river acting as the main source of plastics during flash floods, although simultaneous wave action disrupted this pattern at some sites, redistributing plastic along the shoreline. No relationship was observed between plastic size and distance from the river mouth, though polystyrene foam size varied across rivers and years, reflecting differences in fragmentation associated with fluvial energy, wave action, and dry period duration. Drainage basin size and population density also influence the amount of accumulated plastic. This study shows that flash flood-driven plastic accumulation displays significant spatial and temporal variability across rivers and events. A deeper understanding of these processes can lead to more precise predictions and solutions for plastic pollution.
- Open Access
- 2 July 2026
- Published by IOS Press
Logic Programming for Developing an Interactive Explainable Naval Decision Support System
The increasing complexity of naval traffic necessitates robust decision support systems that enhance safety through compliance with the International Regulations for Preventing Collisions at Sea (COLREGs). This paper proposes using logic programming to develop explainable decision support systems tailored for encounter situations in maritime navigation. Logic programming is utilized to provide recommendations with clear, human-understandable explanations. The framework addresses various encounter scenarios by encoding the COLREGs rules into logical constructs to conclude appropriate actions and facilitates automated reasoning, ensuring rapid and accurate response. Our system is designed to collect sensor data, assess risk levels, and recommend optimal actions that comply with maritime regulations. The inherent explainability of logic programming allows generating traceable and transparent decision paths. Mariners can comprehend the underlying rationale that supports trusted decision-making onboard. The paper explores the implementation of this system, detailing its architecture, inference mechanisms, and integration with Automatic Identification System (AIS) messages. An Interactive eXplainable Decision Support System (IXDSS) was developed to demonstrate its effectiveness in diverse traffic scenarios while providing logical and visual explanations, including interactive ‘what-if’ analyses. IXDSS was evaluated using a standardized set of encounter scenarios, showing accurate scenario prediction, and its explainability was assessed qualitatively.
- 2 July 2026
- Published by IOS Press
Open AccessLogic Programming for Developing an Interactive Explainable Naval Decision Support System
The increasing complexity of naval traffic necessitates robust decision support systems that enhance safety through compliance with the International Regulations for Preventing Collisions at Sea (COLREGs). This paper proposes using logic programming to develop explainable decision support systems tailored for encounter situations in maritime navigation. Logic programming is utilized to provide recommendations with clear, human-understandable explanations. The framework addresses various encounter scenarios by encoding the COLREGs rules into logical constructs to conclude appropriate actions and facilitates automated reasoning, ensuring rapid and accurate response. Our system is designed to collect sensor data, assess risk levels, and recommend optimal actions that comply with maritime regulations. The inherent explainability of logic programming allows generating traceable and transparent decision paths. Mariners can comprehend the underlying rationale that supports trusted decision-making onboard. The paper explores the implementation of this system, detailing its architecture, inference mechanisms, and integration with Automatic Identification System (AIS) messages. An Interactive eXplainable Decision Support System (IXDSS) was developed to demonstrate its effectiveness in diverse traffic scenarios while providing logical and visual explanations, including interactive ‘what-if’ analyses. IXDSS was evaluated using a standardized set of encounter scenarios, showing accurate scenario prediction, and its explainability was assessed qualitatively.
- 24 November 2025
- Published by Springer Nature
- p. 172-184
Measuring and Comparison of the Energy Consumption of Different Machine Learning Methods
- 24 November 2025
- Published by Springer Nature
- p. 172-184
Measuring and Comparison of the Energy Consumption of Different Machine Learning Methods
- 23 October 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 383 (2307) , 20240427
Flash-flood-driven litter accumulation on beaches studied with aerial imagery and machine learning on Sicily’s northeastern coast, southern Italy
This study examines the distributions and dynamics of litter accumulation in a part of the northeastern coast of Sicily, southern Italy, using a multidisciplinary approach with manned and unmanned aerial systems imagery, a machine-learning algorithm and analysis of meteorological data. Aerial imagery shows that litter accumulations are present on beaches near the river mouths and potentially after flash floods. At least five distinct flash flood events depositing litter have been inferred since March 2015 to January 2024 in the study area. Meteorological data show that high precipitation, resulting in flash floods, and prevailing wind direction from the sea are the primary contributors to litter accumulation in this area. The algorithm applied on drone imagery on one of the accumulations detected a composition of polystyrene (42.6%), wood (35.6%), polyurethane (6.1%) and polyethylene terephthalate (5.4%). The general composition of the litter accumulation and its distribution near river mouths suggests that its source is associated with the input from nearby towns through mismanagement on the riverbeds, followed by flash floods. This study highlights the importance of considering river floods when investigating litter dynamics in coastal environments, as well as the potential of using aerial imagery and machine learning to help assess this problem. This article is part of the Theo Murphy meeting issue ‘Sedimentology of plastics: state of the art and future directions’.
- 23 October 2025
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 383 (2307) , 20240427
Flash-flood-driven litter accumulation on beaches studied with aerial imagery and machine learning on Sicily’s northeastern coast, southern Italy
This study examines the distributions and dynamics of litter accumulation in a part of the northeastern coast of Sicily, southern Italy, using a multidisciplinary approach with manned and unmanned aerial systems imagery, a machine-learning algorithm and analysis of meteorological data. Aerial imagery shows that litter accumulations are present on beaches near the river mouths and potentially after flash floods. At least five distinct flash flood events depositing litter have been inferred since March 2015 to January 2024 in the study area. Meteorological data show that high precipitation, resulting in flash floods, and prevailing wind direction from the sea are the primary contributors to litter accumulation in this area. The algorithm applied on drone imagery on one of the accumulations detected a composition of polystyrene (42.6%), wood (35.6%), polyurethane (6.1%) and polyethylene terephthalate (5.4%). The general composition of the litter accumulation and its distribution near river mouths suggests that its source is associated with the input from nearby towns through mismanagement on the riverbeds, followed by flash floods. This study highlights the importance of considering river floods when investigating litter dynamics in coastal environments, as well as the potential of using aerial imagery and machine learning to help assess this problem. This article is part of the Theo Murphy meeting issue ‘Sedimentology of plastics: state of the art and future directions’.
- Open Access
Multi-Resolution Remote Sensing Dataset for the Detection of Anthropogenic Litter: A Multi-Platform and Multi-Sensor Approach
The dataset developed within the PlasticObs+ project aims to facilitate a multi-resolution approach for detecting and quantifying anthropogenic litter through areal images. Traditional detection methods often suffer from narrow, use-case-specific limitations, reducing their transferability. To address this, an image dataset was created featuring various spatial and spectral resolutions. The highest spatial resolution images (ground sampling distance = 0.2 cm) were used to generate a labeled dataset, which was georeferenced for mapping onto coarser-resolution images.
Open AccessMulti-Resolution Remote Sensing Dataset for the Detection of Anthropogenic Litter: A Multi-Platform and Multi-Sensor Approach
The dataset developed within the PlasticObs+ project aims to facilitate a multi-resolution approach for detecting and quantifying anthropogenic litter through areal images. Traditional detection methods often suffer from narrow, use-case-specific limitations, reducing their transferability. To address this, an image dataset was created featuring various spatial and spectral resolutions. The highest spatial resolution images (ground sampling distance = 0.2 cm) were used to generate a labeled dataset, which was georeferenced for mapping onto coarser-resolution images.
- 29 November 2024
- Published by Springer Nature
- p. 270-275
Bi-directional LSTM Applied to the Maritime Target Motion Analysis Problem
- 29 November 2024
- Published by Springer Nature
- p. 270-275
Bi-directional LSTM Applied to the Maritime Target Motion Analysis Problem
- 29 November 2024
- Published by Springer Nature
- p. 47-60
On the Development of a Pixel-Wise Plastic Waste Identification System for Multispectral Remote Sensing Applications
- 29 November 2024
- Published by Springer Nature
- p. 47-60
On the Development of a Pixel-Wise Plastic Waste Identification System for Multispectral Remote Sensing Applications
- 8 November 2023
- Published by Springer Nature
- p. 519-525
Goals and Stakeholder Involvement in XAI for Remote Sensing: A Structured Literature Review
- 8 November 2023
- Published by Springer Nature
- p. 519-525
Goals and Stakeholder Involvement in XAI for Remote Sensing: A Structured Literature Review
- 8 November 2023
- Published by Springer Nature
- p. 481-486
On the Application of Feed-Forward Artificial Neural Networks to the Maritime Target Motion Analysis Problem
- 8 November 2023
- Published by Springer Nature
- p. 481-486
On the Application of Feed-Forward Artificial Neural Networks to the Maritime Target Motion Analysis Problem