Related Experiment Video
Updated: Jul 27, 2025

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging
Published on: April 18, 2025
Ecodatacube.eu: analysis-ready open environmental data cube for Europe
Martijn Witjes1, Leandro Parente1, Josip Križan2
1OpenGeoHub, Wageningen, Netherlands.
An open-access European data cube integrates Landsat and Sentinel-2 satellite data with a digital terrain model (DTM) for continental-scale machine learning. This harmonized data cube enhances land cover classification accuracy, providing valuable resources for environmental research.
Area of Science:
- Earth Observation
- Geospatial Data Science
- Environmental Monitoring
Background:
- Satellite data availability and processing challenges hinder large-scale spatiotemporal analysis.
- A consistent, analysis-ready dataset is crucial for advancing machine learning applications in environmental science.
- Open-access data initiatives are vital for democratizing scientific research and fostering collaboration.
Purpose of the Study:
- To create an analysis-ready, open-access European data cube integrating Landsat, Sentinel-2, and digital terrain model (DTM) data.
- To provide a harmonized, multidimensional feature space for continental-scale spatiotemporal machine learning tasks.
- To assess the accuracy and usability of the data cube components for environmental applications.
Main Methods:
- Systematic spatiotemporal harmonization, compression, and imputation of missing values for satellite data.
- Aggregation of Sentinel-2 and Landsat reflectance into quarterly averages and percentiles to capture seasonal variance.
- Imputation of missing Landsat data using a temporal moving window median (TMWM) approach, with accuracy assessed across different European regions.
Main Results:
- The data cube successfully integrates Landsat, Sentinel-2, and DTM data, offering a consistent spatiotemporal feature space.
- The temporal moving window median (TMWM) imputation method showed varying accuracy across Europe, performing better in Southern Europe than in mountainous regions.
- Land cover classification experiments demonstrated that utilizing the full data cube (including DTM, Landsat, and Sentinel-2) yields the highest classification accuracy.
Conclusions:
- The developed European data cube significantly enhances the accessibility and usability of satellite and terrain data for spatiotemporal machine learning.
- The comprehensive dataset, available under a CC-BY license, supports a wide range of environmental research and monitoring applications.
- The EcoDataCube platform provides a valuable resource for researchers, offering open vegetation, soil, and land use/land cover (LULC) maps alongside the core data cube.
Related Concept Videos
GIS Software, Hardware, and Sources of GIS Data
Selected Data About Geographic Locations
Levels of Use of a GIS
Manipulation and Analysis
Applications of GIS: Disaster Management and Emergency Response
Design Example: Analyzing Capacity Contours for Flood Risk Assessment

