This land cover classification of Germany was created using Sentinel-2 imagery from the years 2015 to 2017 and LUCAS 2015 in-situ reference data (https://ec.europa.eu/eurostat/web/lucas). It contains seven land cover types: (1) artificial land, (2) open soil, (3) high seasonal vegetation, (4) high perennial vegetation, (5) low seasonal vegetation, (6) low perennial vegetation and (7) water with a spatial resolution of 10m x 10m. For further information, please see the following publication: https://doi.org/10.1016/j.jag.2020.102065

This dataset includes the normalized difference vegetation index (NDVI) derived from Sentinel-2 imagery. Using the Google Earth Engine, all granules with a cloud cover below 60% were used as input. Cloudy pixels (referring to quality layer QA60) were masked as well. Eventually, a median mosaic was composed over the whole observation period. It was also used as input for a land cover classification (see: Land Cover DE - Sentinel-2 - Germany, 2015).

The Sentinel-2 fractional vegetation cover (fCover) product for the Netherlands was produced as part of the NextGEOSS project at the German Aerospace Center (DLR). The goal is to derive abundance maps from atmospherically corrected Sentinel-2 multispectral images for: photosynthetically active vegetation (PV); and for combined non-photosynthetically active vegetation (NPV) and bare soil (BS). The fCover product for the Netherlands has been generated by processing 10 cloud-free Sentinel-2 tiles which covered the country on 8 September 2016. The map has a spatial resolution of 60m x 60m. The Sentinel-2 scene classification layer was used to ensure that the spectral unmixing was only performed on areas of vegetation or soil. The abundance maps were made by performing MESMA unmixing on each pixel from an endmember library of PV and combined NPV + BS spectra. The purest pixels in a scene, called endmembers, were extracted using the Spatial-Spectral Endmember Extraction (SSEE) approach. The PV and NPV+BS endmembers were classified with a random forest approach and selected to form the spectral library. The spectral library was used in the µMESMA unmixing to get the PV and NPV+BS abundances.

The World Settlement Footprint (WSF) 2019 is a 10m resolution binary mask outlining the extent of human settlements globally derived by means of 2019 multitemporal Sentinel-1 (S1) and Sentinel-2 (S2) imagery. Based on the hypothesis that settlements generally show a more stable behavior with respect to most land-cover classes, temporal statistics are calculated for both S1- and S2-based indices. In particular, a comprehensive analysis has been performed by exploiting a number of reference building outlines to identify the most suitable set of temporal features (ultimately including 6 from S1 and 25 from S2). Training points for the settlement and non-settlement class are then generated by thresholding specific features, which varies depending on the 30 climate types of the well-established Köppen Geiger scheme. Next, binary classification based on Random Forest is applied and, finally, a dedicated post-processing is performed where ancillary datasets are employed to further reduce omission and commission errors. Here, the whole classification process has been entirely carried out within the Google Earth Engine platform. To assess the high accuracy and reliability of the WSF2019, two independent crowd-sourcing-based validation exercises have been carried out with the support of Google and Mapswipe, respectively, where overall 1M reference labels have been collected based photointerpretation of very high-resolution optical imagery.

The World Settlement Footprint (WSF) 3D provides detailed quantification of the average height, total volume, total area and the fraction of buildings at 90 m resolution at a global scale. It is generated using a modified version of the World Settlement Footprint human settlements mask derived from Sentinel-1 and Sentinel-2 satellite imagery in combination with digital elevation data and radar imagery collected by the TanDEM-X mission. The framework includes three basic workflows: i) the estimation of the mean building height based on an analysis of height differences along potential building edges, ii) the determination of building fraction and total building area within each 90 m cell, and iii) the combination of the height information and building area in order to determine the average height and total built-up volume at 90 m gridding. In addition, global height information on skyscrapers and high-rise buildings provided by the Emporis database is integrated into the processing framework, to improve the WSF 3D Building Height and subsequently the Building Volume Layer. A comprehensive validation campaign has been performed to assess the accuracy of the dataset quantitatively by using VHR 3D building models from 19 globally distributed regions (~86,000 km2) as reference data. The WSF 3D standard layers are provided in the format of Lempel-Ziv-Welch (LZW)-compressed GeoTiff files, with each file - or image tile - covering an area of 1 x 1 ° geographical lat/lon at a geometric resolution of 2.8 arcsec (~ 90 m at the equator). Following the system established by the TDX-DEM mission, the latitude resolution is decreased in multiple steps when moving towards the poles to compensate for the reduced circumference of the Earth.

Grassland mowing dynamics (i.e. the timing and frequency of mowing events) have a strong impact on grassland functions and yields. As grasslands in Germany are managed on small-scale units and grass grows back quickly, satellite information with high spatial and temporal resolution is necessary to capture grassland mowing dynamics. Based on Sentinel-2 data time series, mowing events are detected throughout Germany and annual maps of the grassland mowing frequency generated. The grassland mowing detection approach operates per pixel, including preprocessing of the Enhanced Vegetation Index (EVI) time series and a calibrated rule-based grassland mowing detection which is specified in more detail in Reinermann et al. 2022, 2023.

SWIM Water Extent is a global surface water product at 10 m pixel spacing based on Sentinel-1/2 data. The collection contains binary layers indicating open surface water for each Sentinel-1/2 scene. Clouds and cloud shadows are removed using ukis-csmask (see: https://github.com/dlr-eoc/ukis-csmask ) and are represented as NoData. The water extent extraction is based on convolutional neural networks (CNN). For further information, please see the following publications: https://doi.org/10.1016/j.rse.2019.05.022 and https://doi.org/10.3390/rs11192330

The product contains information of tree canopy cover loss in Germany per district (Landkreis) between January 2018 and April 2021 at monthly temporal resolution. The information is aggregated at from the 10 m spatial resolution Sentinel-2 and Landsat-based raster product (Tree Canopy Cover Loss Monthly - Landsat-8/Sentinel-2 - Germany, 2018-2021). The method used to derive this product as well as the mapping results are described in detail in Thonfeld et al. (2022). The map depicts areas of natural disturbances (windthrow, fire, droughts, insect infestation) as well as sanitation and salvage logging, and regular forest harvest without explicitly differentiating these drivers. The vector files contain information about tree canopy cover loss area per forest type (deciduous, coniferous, both) and per year (2018, 2019, 2020, January-April 2021, and January 2018-April 2021) in absolute numbers and in percentages. In addition, the vector files contain the district area and the total forest area per district.

The Tree Species Germany product provides a map of dominant tree species across Germany for the year 2022 at a spatial resolution of 10 meters. The map depicts the distribution of ten tree species groups derived from multi-temporal optical Sentinel-2 data, radar data from Sentinel-1, and a digital elevation model. The input features explicitly incorporate phenological information to capture seasonal vegetation dynamics relevant for species discrimination. A total of over 80,000 training and test samples were compiled from publicly accessible sources, including urban tree inventories, Google Earth Pro, Google Street View, and field observations. The final classification was generated using an XGBoost machine learning algorithm. The Tree Species Germany product achieves an overall F1-score of 0.89. For the dominant species pine, spruce, beech, and oak, class-wise F1-scores range from 0.76 to 0.98, while F1-scores for other widespread species such as birch, alder, larch, Douglas fir, and fir range from 0.88 to 0.96. The product provides a consistent, high-resolution, and up-to-date representation of tree species distribution across Germany. Its transferable, cost-efficient, and repeatable methodology enables reliable large-scale forest monitoring and offers a valuable basis for assessing spatial patterns and temporal changes in forest composition in the context of ongoing climatic and environmental dynamics.

The PolarLakes dataset provides bi-weekly observations of supraglacial lakes on Antarctic ice shelves, utilizing imagery from Sentinel-2 and Sentinel-1 to address time series gaps caused by frequent cloud cover. These observations detect the extents of supraglacial lakes with a U-Net model for every two weeks from November to March, with each sensor operating independently before the data is merged. The resulting bi-weekly product reflects the maximum lake extents for the first and second halves of each month. When combined for an entire season, the dataset consolidates all bi-weekly records over these five months, allowing for analysis of the maximum lake extent per season and the frequency of lake formation, which can occur up to ten times (5 months á two weeks). The year indicated in the dataset corresponds to January of the melt season, as this month typically experiences the highest melt rates (e.g., 2023 refers to the season from November 2022 to March 2023). The aggregation of all annual datasets creates a recurrence layer that illustrates the frequency of lake presence throughout the entire observation period, which spans from 2014 to 2024, depending on satellite data availability for each ice shelf. The PolarLakes dataset provides valuable insights into the dynamics of supraglacial lakes and serves as a crucial resource for hydrological and climate modeling.