Current research projects on data science

BALTZOS - Mapping of Zostera meadows in the western Baltic

Baseline mapping of Zostera meadows in the western Baltic Sea to assess Zostera distribution and depth range, and Zostera distribution trends. The data contribute to maps, publications, public communication and management decisions, and large-scale calculations of areal extent and CO2 sequestration, etc.

Project duration: since 2009

Contact at GEOMAR: Philipp Schubert; pschubert(at), Prof. Thorsten Reusch; treusch(at)

Included data types: RGB video and nmea-Data from towed camera systems



Global in situ imaging with the Underwater Vision Profiler 5 (UVP5) yields large amounts of in situ plankton and particle images. With PlanktonID we combine state-of-the-art deep learning image recognition algorithms and a citizen scientist approach, to enhance our abilities for high-throughput image annotation needed for an efficient sensor-to-information pathway. Involvement of citizen scientists also leads to a direct dissemination of results to the general public and will contributes to our outreach activities. 

Project duration: since 2016

Contact at GEOMAR: Dr. Rainer Kiko; rkiko(at)

Included data types: plankton and particle image data, obtained in situ with the Underwater Vision Profiler, multiple annotations of plankton images by Citizen Scientists


ARENA 2 laboratory

The overall goal is to provide tools that directly improve the scientific workflow by facilitating visual data exploration.

Project duration: since 2018

Contact at GEOMAR: Dr. Tom Kwasnitschka; tkwasnitschka(at)

Included data types: 2-4D geospatial data, model runs, large format video, stills, telepresence



Route tracking seismoacoustic 3D submarine cable detection - sensor and navigation development for CATRA: design and construction of a 3D ultra-high resolution multichannel seismic acquisition platform capable to detect small objects (> 20 cm) buried up to 2 m depth in marine sediments.

Project duration: 2020-2023

Contact at GEOMAR: Dr. Jörg Bialas jbialas(at)

Included data types: marine near vertical seismic data (8 kHz)


Climvar Baltic Sea

Investigation of the changing impact of the large-scale atmospheric circulation on the regional climate variability of the Baltic Sea for the period 1950-2022

Project duration: 2022 - 2024

Contact at GEOMAR: Dr. Andreas Lehmann; alehmann(at)

Included data types: time series gridded data atmosphere & oceanography,  observational data (ERA5 data; Model-Data, INDEX Data)


SOI Virtual Vents

One of the first comprehensive photogrammetric surveys of an entire hydrothermal field in the Northeastern Lau Basin.

Project duration: 2016 - 2024

Contact at GEOMAR: Dr. Tom Kwasnitschka; tkwasnitschka(at)

Included data types: Video, stills, photogrammetric models, geologic & habitat maps, analytics



In-situ measurements of key physical parameters in pyroclastic eruptions using UAVs and resilient sensor packages.

Project duration: 2018 - 2024

Contact at GEOMAR: Dr. Tom Kwasnitschka; tkwasnitschka(at)

Included data types: vehicle telemetry, in-situ probe data


Deep Quanticams

Visual mapping and quantitative machine vision in the deep sea. In particular the project will advance automated visual mapping of the ocean floor from photo or video, including the steps camera calibration, finding correspondending points in underwater images, underwater multi-view geometry, image registration, visual localization, seafloor surface geometry estimation and faithful color correction in order to facilitate the next generation ocean mapping and measurement applications.

Project duration: 2019-2024

Contact at GEOMAR: Dr. Kevin Köser; kkoeser(at)

Included data types: Marine Image Data



TechOceanS will develop nine innovative technologies and methods for deep sea sensing, sample collection and on-board analysis, and AI-driven image processing and transmission.

Project duration: 2020-2024

Contact at GEOMAR: Dr. Kevin Köser; kkoeser(at)

Included data types: Marine image data, data from different (novel) sensors



CASCADIA CO2 supports pre-site studies of a possible CO2 storage location in marine basalt complexes. Based on four components ocean-bottom seismometer investigations the project adds missing information on shear parameters in marine basalt complexes.

Project duration: 2022-2024

Contact at GEOMAR: Dr. Jörg Bialas jbialas(at)

Included data types: marine wide angle seismic data (compressional and converted shear wave)



Marispace-X aims to create a digital maritime data space based on the data sovereignty, security, interoperability and modularity of Gaia-X. Marispace-X offers new directions in maritime Big Data processing and in the analysis of sensor data via edge, fog and cloud computing. In MARISPACE, GEOMAR focuses on the development of smart AUVs that can automatically analyze image data and create and use 3D information.

Project duration: 2021 - 2024

Contact at GEOMAR: Dr. Timm Schoening; tschoening(at)

Included data types: Bathymetric data, Images & Videos, Remote Sensing (photos & radar), Time series data


ProBaNNt (Professional intelligent munitions assessment using 3D reconstructions and Bayesian Neural Networks)

ProBaNNt aims to improve decision-making capabilities on various levels and generate a comprehensive tool to support offshore explosive ordnance disposal (EOD) campaigns. The project integrates sustainable convergence, use, and analysis of existing EOD data with new data acquisition techniques, such as 3D photogrammetry. All of this information will be integrated into EOD decision-making software to propose the most viable clearance option for a given munitions item at a given location.

Project duration: 2021 - 2024

Contact at GEOMAR: Torsten Frey; tfrey(at)

Included data types: Marine image data (photomosaics and digital terrain models), high frequency sonar “camera” data, munition clearance data



The Ghostnetbusters project addresses the automatic detection of lost fishing gear, known as ghost nets, using seafloor data. This data will be used as training for an AI to automatically identify ghost nets in order to significantly speed up the search for lost fishing gear and thus its recovery.

Project duration: 2023 - 2024

Contact at GEOMAR: Mia Schumacher; mschumacher(at)

Included data types: Hydroacoustic Data (from Side Scan Sonar, MBES), satellite images


BlueHealthTech Hyperquant


Immersive visualization of hyperpolarized MRT-scans.

Project duration: 2022-2025

Contact at GEOMAR: Dr. Tom Kwasnitschka; tkwasnitschka(at)

Included data types: MRT scans, annotations, user interaction data


SAM - Smart AUV-based Magnetics

Munitions in the sea pose a threat to life and work of people in coastal areas worldwide. With its smart and autonomous navigation routines based on in-situ magnetic measurements, SAM has the potential to accelerate the clearance of offshore UXO (UneXploded Ordnance).

Project duration: 2023-2025

Contact at GEOMAR: Dr. Marc Seidel; mseidel(at)

Included data types: 3D Magnetic data, time series data, AUV navigational data



DIPLO (Digital Plankton Ocean)

The goal is to use AI models to effectively quantify relationships between different plankton images,to facilitate comparability between different imaging systems.

Project duration: 2023-2026

Contact at GEOMAR: Veit Dausmann; vdausmann(at), Jan Taucher; jtaucher(at)

Included data types: marine image data (Zooplankton)



Seismic and electromagnetic characterization and monitoring of CO2 sequestration sites in marine basalt complexes. Marine 2D and 3D seismic and electromagnetic investigations will be applied in a joint analyses and inversion of a possible storage site for giga-tons of CO2. The project is embedded in the international PERBAS consortium of 10 partners from Norway, USA and India.

Project duration: 2023-2026

Contact at GEOMAR: Dr. Jörg Bialas jbialas(at)

Included data types: 2D and 3D  multichannel and four component marine seismic data, 2D and 3D controlled source electromagnetic data


AI-quifer - Using AI to detect off-shore groundwater as key to coastal water management

We propose that globally available geospatial data on surface features (e.g. Digital Elevation Model, Land Surface Models, shelf bathymetry and the GRACE datasets), in conjunction with climatic data, can be used to predict the largely hidden offshore occurrence of coastal freshwater aquifers. Specifically, we aim to derive a reliable machine learning method that will account for the complex underlying hydrological mechanism of offshore groundwater emplacement and preservation. At the end, we expect to provide a tool that allows us to produce global probability maps of the location of the coastal fresh/saline water interface (a proxy for OFG), that can be adjusted to future climate scenarios.

Project duration: 2023-2026

Contact at GEOMAR: Dr. Laura Haffert; lahaffert(at)

Included data types: 2D hydrogeological modelling data, DEM, LSM, GRACE, climatic data


Imaging Marine Life in an Ocean of Change (IOChange)

Global change impacts the distribution and activity of marine life at local to basin-wide scales, with major consequences for oceanic oxygen dynamics, nutrient cycles and the transfer of carbon dioxide from the atmosphere to the deep sea. Within IOChange, we use augmented image observations that integrate autonomous camera and environmental sensor systems with state-of-the-art artificial intelligence solutions and ecophysiological approaches to yield a novel image of zooplankton and detrital particle (Z&P) dynamics in a changing ocean.

Project duration: 2022 - 2027

Contact at GEOMAR: Dr. Rainer Kiko; rkiko(at)

Included data types: plankton and particle image data, obtained in situ or using benchtop imaging systems; see also


ValidITy (Validation of Intelligent Terrain and Feature Recognition Methods for Hydrographic Data)

ValidITy project aims to develop software that uses artificial intelligence and terrain classification dictionaries to detect objects in bathymetric measurement data. The aim is to significantly reduce the cost of data analysis.

Project duration: 2023 - 2025

Contact at GEOMAR: Prof. Jens Greinert; jgreinert(at)

Included data types: MBES data


KIMERA - Artificial Intelligence applied to mapping of the seafloor and marine spatial planning

Development of a method for predicting the characteristics of the seabed using machine learning as a basis for marine resource assessment and habitat exploration.

Duration: 2023 - 2026

Contact at GEOMAR: Dr. Philipp Brandl; pbrandl(at), Dr. Sven Petersen; spetersen(at)

Included Data types: MBES und geological Data


SpotKI- Digital Twins, Robotics, AI and Sensors for the monitoring of large-scale technical systems - Safety as a Service

Use of a "robot dog" SPOT from Boston Dynamics, an agile autonomous platform that is optimized for monitoring tasks and whose software framework is prepared for the connection of additional sensors. GEOMAR supplements this capability with sensors for high-resolution detection of trace gases and the integration of digital twins from ChemParks (NorthDocks GmbH).

Duration: 2023 - 2026

Contact at GEOMAR: Roberto Benavides; rbenavides(at)

Included Data types: Environmental measurement data (especially gases)


The "Helmholtz School for Marine Science" (MarDATA) is a concept for a graduate school of the Helmoltz Association. MarDATA aims to define and train a new type of marine scientist. It aims to integrate young scientists from the computer and mathematical sciences directly into the marine sciences. Scientists from GEOMAR and AWI train PhD students together with the respective partner universities.

Project duration: 2022 - 2027

Contact at GEOMAR: Prof. Dr. Arne Biastoch; abiostoch(at); Dr. Enno Prigge; eprigge(at)


MARDATA: Automated observation and data assimilation

In this project methods are developed that analyze AUV (Autonomous underwater vehicle) sensor data such as temperature, salinity, oxygen and turbidity of the water in (near) real-time to drive the AUV towards regions of interest to enhance data acquisition and quality. By this the degree of autonomy of GEOMARs Girona500 AUVs build by iquarobotics shall be increased.

Project duration: 2019 - 2023

Contact at GEOMAR: Tim Benedikt von See; tsee(at)

Included data types: real-time AUV sensor data


MARDATA: The Digital Lab Book

Development of a framework to capture, compare and disseminate spatially immersive visualization workflows as practiced in the ARENA2 lab.

Project duration: 2021 - 2024

Contact at GEOMAR: Armin Bernstetter; abernstetter(at)

Included data types: Bathymetric time series, imagery, video, analytics, user interaction data, annotations


MARDATA: Deep neural networks for the prediction of sediment accumulation at the seafloor

Using different deep learning techniques, we aim to provide a global map of sedimentation rates from in-situ measurements and predictor grids, that might influence the sedimentation rates.

Project duration: 2021 - 2024

Contact at GEOMAR: Naveen Kumar Parameswaran;  nparameswaran(at)

Included data types: Predictor grids (spatial data), sedimentation rate measurements (tabular form)


MARDATA: Links between climate and volcanism

A recent and fascinating discovery from studies of glaciated regions is that changes in climate may affect volcanic activity. However, the binary nature of the volcanic records (eruption or no eruption) adds complexity to correlating this data to real valued climate proxies. The aim of the project is to develop appropriate statistical procedures to overcome this difficulty and apply them to test the hypothesis that climate influences volcanism. 

Project duration: 2021 - 2024

Contact at GEOMAR: José Kling;  jkling(at)

Included data types: Eruption records, climate proxy time series


MARDATA: Learning underwater visual appearance for robust seafloor surveys

This project aims at identifying and analyzing the characteristics of underwater image appearance, especially in the deep ocean where no natural light is present anymore. Specialized feature matching algorithms that robustly adapt to the underwater conditions will be developed considering both, data-driven methods and methods modelling the physics of underwater appearance. The goal is to achieve algorithms that allow detection of loop closures in AUV tracks, visually correct for drift in inertial underwater navigation or improve structure-from-motion application in challenging underwater settings.

Project duration: 2021 - 2024

Contact at GEOMAR: Patricia Schöntag;  pschoentag(at)

Included data types: Marine image data (simulations and photos), time series data (navigation data) and some other sensor data (optics measurements, e.g. turbidity, scattering, absorption)


MARDATA: Semantic seaflloor mapping

Within this project, a generic workflow and associated software for the classification of geological, biological and anthropogenic information from images will be developed. Based on this, algorithms for automatic analysis, segmentation and classification of large marine image datasets will be formed.

Project duration: 2019 - 2023

Contact at GEOMAR: Benson Mbani; bmbani(at)

Included data types: Marine Image Data