March 2021

Geo-harmonizer project

Prepared by: OpenGeoHub, mundialis and Terrasigna

Executive summary

We have prepared the following list of layers:

• Density of building areas according to OSM (2021) and Copernicus (impervious built-up – 2018);
• Density of commercial, industrial and residential buildings according to OSM (2021);
• Harmonized protected areas derived from OSM (2021) and NATURA2000 (2019)
• Harmonized administrative areas (county-level) derived from OSM (2021) and NUTS (2021);
• Probabilities of 18 OSM land cover classes (2021);
• Density of highways and railways according to OSM (2021);

The layers were prepared at 10-m spatial resolution, then aggregated to 30-m spatial resolution to produce density/probability maps that are compatible with other products and cover the same coverage. All processing was implemented using python in combination with GDAL and can be used to update the layers on a monthly basis.

All layers are available publicly as Cloud-Optimized GeoTIFFs for download and for viewing from: https://maps.opendatascience.eu.

Geo-harmonizer project

Prepared by: OpenGeoHub, CTU in Prague and Terrasigna

Executive summary

The layers that are currently available:

• Annual dominant land cover class based on CORINE (2000–2019);
• Annual probabilities and uncertainties for 33 CORINE land cover classes (2000–2019);
• Annual Landsat RGB composites for the summer season (June–September for 2000–2019);
• Annual Landsat NDVI composite derived by season (winter, spring, summer and fall – 2000–2020);

The annual land cover product for continental Europe was based on Ensemble Machine Learning (EML), which used training samples obtained from other European Commission funded projects, such as the LUCAS (Land Use and Coverage Area frame Survey) and the Copernicus Land Monitoring Service, and several harmonized raster layers (e.g. GLAD Landsat ARD imagery and Continental EU DTM) to predict the dominant land cover, probabilities and uncertainties for 33 classes compatible with CLC (CORINE Land Cover) over 20 years (2000–2019).

All layers are available publicly as Cloud-Optimized GeoTIFFs for download and for viewing from: https://maps.opendatascience.eu.

A space-time extremely randomised trees model was used to estimate PM_2.5 concentrations over Europe. Satellite remote sensing data, meteorological data, and land variables were used as the independent variables, PM_2.5 ground-observations were used as the dependent variable to create our model. Used data is shown in Table 1.

The outcome of this study is a daily full coverage PM_2.5 dataset with 1 km spatial resolution for a three-year period 2018–2020.

We selected November 2020 for downloading demonstration data, the same principle applies to the remaining months by changing the month and the year in the links.

• https://s3.eu-central-1.wasabisys.com/eumap/enq/enq_pm25_et.eml_m_1km_na_2020.11.01..2020.11.30_eumap_3035_v1.tif

Geo-harmonizer project

Prepared by: OpenGeoHub, CTU in Prague, mundialis and Terrasigna

Executive summary

Environmental quality maps

Seamless environmental quality maps related to milestone 7 are currently represented by Monthly Particulates (PM2.5) Predictions computed using Machine Learning for years 2018-2020.

We have prepared the following list of layers:

• PM2.5 monthly maps for the year 2018
• PM2.5 monthly maps for the year 2019
• PM2.5 monthly maps for the year 2020

The layers were prepared at 1-km spatial resolution. All processing was implemented using Python in combination with GDAL and GRASS GIS.

All layers are available publicly as Cloud-Optimized GeoTIFFs for download and via the ODS-Europe Viewer. For more information contact: Saleem Ibrahim (CVUT Prague).

Climate change indicators

The objective of our reprocessing efforts has been to aggregate selected ERA5 Land hourly datasets (period: 2000–2020, DOI) to daily data while increasing the resolution from the native ERA5 resolution of 0.1 degree to 30 arc-sec by using auxiliary data (CHELSA, Karger et. al, 2020, DOI). The original ERA5 data are available from the Copernicus Climate Data Store (CDS). We have processed air temperature 2 meter above ground, surface temperature and precipitation. All data were first aggregated in time to obtain for temperatures daily minimum, mean, and maximum and for precipitation the daily sum. Spatial resolution was enhanced to 30 arc seconds with the help of CHELSA climatology. For each day we used the corresponding monthly long-term average of CHELSA. The aim was to use the fine spatial detail of CHELSA and at the same time preserve the general regional pattern and fine temporal detail of ERA5 Land. The steps included aggregation and enhancement, specifically:

ERA5 Land temperature data:

1. spatially aggregating CHELSA to the resolution of ERA5 Land,
2. calculate difference of ERA5 Land – aggregated CHELSA,
3. interpolate differences with a Gaussian filter to 30 arc seconds,
4. add the interpolated differences to CHELSA,

ERA5 Land precipitation data:

1. spatially aggregating CHELSA to the resolution of ERA5 Land,
2. calculate proportion of ERA5 Land / aggregated CHELSA,
3. interpolate proportion with a Gaussian filter to 30 arc seconds,
4. multiply the interpolated proportions with CHELSA,

Using proportions ensures that areas without precipitation remain areas without precipitation. Only if there was actual precipitation in a given area, precipitation was redistributed according to the spatial detail of CHELSA. The resulting dataset is now available via the ODS-Europe viewer (see below). For more information contact: Markus Metz (Mundialis).

Fire disturbance and flood events

Fire disturbance indicator includes 2 products in this release: (1) 109 fire activations mapped by the Copernicus Emergency Rapid Mapping  Service: individual downloaded, rasterized and mosaicked per year and season, resampled at 30-m and (2) MODIS MCD64A1 collection 6 Burned Area Monthly aggregated by season, mosaicked for the Geo-harmonizer region (window 8 and 22) and resampled to 250m. Thus, the product based on EMS covers the period 2012–2020 and the one based on MCD64A1: winter 2000–2020. 

The flood events indicator includes one product in this release: (123) flood activations mapped by the Copernicus Emergency Rapid Mapping  Service: individual downloaded, rasterized and mosaicked per year and season, resampled at 30m. The temporal extent is 2012–2020. For more information contact: Codrina M. Ilie (TerraSigna).

Actual and Potential Natural vegetation

Predictions of distribution (probability of occurrence) for 10 target forest species from the European Forest Atlas were produced for the period 2000–2020 and at 120-m spatial resolution. Predictions include actual distribution of forest tree species and their potential distribution for the most recent year (2019). Users can put maps of actual vs potential distribution next to each other and visualize gaps in data. Some artifacts in predictions are currently being filtered out (due to missing values in some input layers) and final maps will be made publicly available via the ODSE-viewer in the coming weeks. Fore more information contact: Carmelo Bonannella (OpenGeoHub).