Land Cover Map 2021 (LCM2021) is a suite of geospatial land cover datasets (raster and polygon) describing the UK land surface in 2021. These were produced at the UK Centre for Ecology & Hydrology by classifying satellite images from 2021. Land cover maps describe the physical material on the surface of the country. For example grassland, woodland, rivers & lakes or man-made structures such as roads and buildingsThis is a 10 m Classified Pixel dataset, classified to create a single mosaic of national cover. Provenance and quality:UKCEH’s automated land cover classification algorithms generated the 10m classified pixels. Training data were automatically selected from stable land covers over the interval of 2017 to 2019. A Random Forest classifier used these to classify four composite images representing per season median surface reflectance. Seasonal images were integrated with context layers (e.g., height, aspect, slope, coastal proximity, urban proximity and so forth) to reduce confusion among classes with similar spectra.Land cover was validated by organising the pixel classification into a land parcel framework (the LCM2021 Classified Land Parcels product). The classified land parcels were compared to known land cover producing confusion matrix to determine overall and per class accuracy.View full metadata information and download the data at catalogue.ceh.ac.uk

Introduction and Rationale: Due to our increasing understanding of the role the surrounding landscape plays in ecological processes, a detailed characterization of land cover, including both agricultural and natural habitats, is ever more important for both researchers and conservation practitioners. Unfortunately, in the United States, different types of land cover data are split across thematic datasets that emphasize agricultural or natural vegetation, but not both. To address this data gap and reduce duplicative efforts in geospatial processing, we merged two major datasets, the LANDFIRE National Vegetation Classification (NVC) and USDA-NASS Cropland Data Layer (CDL), to produce an integrated land cover map. Our workflow leveraged strengths of the NVC and the CDL to produce detailed rasters comprising both agricultural and natural land-cover classes. We generated these maps for each year from 2012-2021 for the conterminous United States, quantified agreement between input layers and accuracy of our merged product, and published the complete workflow necessary to update these data. In our validation analyses, we found that approximately 5.5% of NVC agricultural pixels conflicted with the CDL, but we resolved a majority of these conflicts based on surrounding agricultural land, leaving only 0.6% of agricultural pixels unresolved in our merged product. Contents: Spatial data

Attribute table for merged rasters

Technical validation data

Number and proportion of mismatched pixels Number and proportion of unresolved pixels Producer's and User's accuracy values and coverage of reference data Resources in this dataset:Resource Title: Attribute table for merged rasters. File Name: CombinedRasterAttributeTable_CDLNVC.csvResource Description: Raster attribute table for merged raster product. Class names and recommended color map were taken from USDA-NASS Cropland Data Layer and LANDFIRE National Vegetation Classification. Class values are also identical to source data, except classes from the CDL are now negative values to avoid overlapping NVC values. Resource Title: Number and proportion of mismatched pixels. File Name: pixel_mismatch_byyear_bycounty.csvResource Description: Number and proportion of pixels that were mismatched between the Cropland Data Layer and National Vegetation Classification, per year from 2012-2021, per county in the conterminous United States.Resource Title: Number and proportion of unresolved pixels. File Name: unresolved_conflict_byyear_bycounty.csvResource Description: Number and proportion of unresolved pixels in the final merged rasters, per year from 2012-2021, per county in the conterminous United States. Unresolved pixels are a result of mismatched pixels that we could not resolve based on surrounding agricultural land (no agriculture with 90m radius).Resource Title: Producer's and User's accuracy values and coverage of reference data. File Name: accuracy_datacoverage_byyear_bycounty.csvResource Description: Producer's and User's accuracy values and coverage of reference data, per year from 2012-2021, per county in the conterminous United States. We defined coverage of reference data as the proportional area of land cover classes that were included in the reference data published by USDA-NASS and LANDFIRE for the Cropland Data Layer and National Vegetation Classification, respectively. CDL and NVC classes with reference data also had published accuracy statistics. Resource Title: Data Dictionary. File Name: Data_Dictionary_RasterMerge.csv

The statewide dataset contains a combination of land cover mapping from 2016 aerial imagery and land use derived from standardized assessor parcel information for Massachusetts. The data layer is the result of a cooperative project between MassGIS and the National Oceanic and Atmospheric Administration’s (NOAA) Office of Coastal Management (OCM). Funding was provided by the Mass. Executive Office of Energy and Environmental Affairs.

This land cover/land use dataset does not conform to the classification schemes or polygon delineation of previous land use data from MassGIS (1951-1999; 2005).In this map service layer hosted at MassGIS' ArcGIS Server, all impervious polygons are symbolized by their generalized use code; all non-impervious land cover polygons are symbolized by their land cover category. The idea behind this method is to use both cover and use codes to provide a truer picture of how land is being used: parcel use codes may indicate allowed or assessed, not actual use; land cover alone (especially impervious) does not indicate actual use.

See the full datalayer description for more details.This map service is best displayed at large (zoomed in) scales. Also available are a Feature Service and a Tile Service (cache). The tile cache will display very quickly in in ArcGIS Online, ArcGIS Desktop, and other applications that can consume tile services.

This dataset consists of the 1km raster, percentage target class version of the Land Cover Map 2015 (LCM2015) for Great Britain. The 1km percentage product provides the percentage cover for each of 21 land cover classes for 1km x 1km pixels. This product contains one band per target habitat class (producing a 21 band image). The 21 target classes are based on the Joint Nature Conservation Committee (JNCC) Broad Habitats, which encompass the entire range of UK habitats. This dataset is derived from the vector version of the Land Cover Map, which contains individual parcels of land cover and is the highest available spatial resolution. LCM2015 is a land cover map of the UK which was produced at the Centre for Ecology & Hydrology by classifying satellite images from 2014 and 2015 into 21 Broad Habitat-based classes. LCM2015 consists of a range of raster and vector products and users should familiarise themselves with the full range (see related records, the CEH web site and the LCM2015 Dataset documentation) to select the product most suited to their needs. LCM2015 was produced at the Centre for Ecology & Hydrology by classifying satellite images from 2014 and 2015 into 21 Broad Habitat-based classes. It is one of a series of land cover maps, produced by UKCEH since 1990. They include versions in 1990, 2000, 2007, 2015, 2017, 2018 and 2019.

This web map service (WMS) is the 25m raster version of the Land Cover Map 2015 (LCM2015) for Great Britain and Northern Ireland. It shows the target habitat class with the highest percentage cover in each 25m x 25m pixel. The 21 target classes are based on the Joint Nature Conservation Committee (JNCC) Broad Habitats, which encompass the entire range of UK habitats.The 25m raster web map service is the most detailed of the LCM2015 raster products, both thematically and spatially, and it is derived from the LCM2015 vector product. For LCM2015 per-pixel classifications were conducted, using a random forest classification algorithm. The resultant classifications were then mosaicked together, with the best classifications taking priority. This produced a per-pixel classification of the UK, which was then 'imported' into the spatial framework, recording a number of attributes, including the majority class per polygon which is the Land Cover class for each polygon.Find out more about Land Cover Map 2015 at ceh.ac.uk.LCM2015 is available for download to Catchment Based Approach (CaBA) Partnerships in the desktop GIS data package. Please contact your CaBA catchment host for further information.

[Metadata] Description: Agricultural Land Use Maps (ALUM) for islands of Kauai, Oahu, Maui, Molokai, Lanai and Hawaii as of 1978-1980. Sources: State Department of Agriculture; Hawaii Statewide GIS Program, Office of Planning. Note: August, 2018 - Corrected one incorrect record, removed coded value attribute domain.For more information on data sources and methodologies used, please refer to complete metadata at https://files.hawaii.gov/dbedt/op/gis/data/alum.pdf or contact Hawaii Statewide GIS Program, Office of Planning and Sustainable Development, State of Hawaii; PO Box 2359, Honolulu, HI 96804; (808) 587-2846; email: gis@hawaii.gov; Website: https://planning.hawaii.gov/gis.

This dataset consists of the 25m raster version of the Land Cover Map 1990 (LCM1990) for Great Britain. The 25m raster product consists of three bands: Band 1 - raster representation of the majority (dominant) class per polygon for 21 target classes; Band 2 - mean per polygon probability as reported by the Random Forest classifier (see supporting information); Band 3 - percentage of the polygon covered by the majority class. The 21 target classes are based on the Joint Nature Conservation Committee (JNCC) Broad Habitats, which encompass the entire range of UK habitats. This dataset is derived from the vector version of the Land Cover Map, which contains individual parcels of land cover and is the highest available spatial resolution. The 25m raster is the most detailed of the LCM1990 raster products both thematically and spatially, and it is used to derive the 1km products. LCM1990 is a land cover map of the UK which was produced at the UK Centre for Ecology & Hydrology by classifying satellite images (mainly from 1989 and 1990) into 21 Broad Habitat-based classes. It is the first in a series of land cover maps for the UK, which also includes maps for 2000, 2007, 2015, 2017, 2018 and 2019. LCM1990 consists of a range of raster and vector products and users should familiarise themselves with the full range (see related records, the UKCEH web site and the LCM1990 Dataset documentation) to select the product most suited to their needs. This work was supported by the Natural Environment Research Council award number NE/R016429/1 as part of the UK-SCAPE programme delivering National Capability.

Extensive land use and geographic data at the tax lot level in GIS format (ESRI Shapefile). Contains more than seventy fields derived from data maintained by city agencies, merged with tax lot features from the Department of Finance’s Digital Tax Map, clipped to the shoreline. All previously released versions of this data are available at BYTES of the BIG APPLE- Archive

The CNES Land Cover Map (Occupation des Sols, OSO) produces land classification for Metropolitan France at 10 m spatial resolution based on Sentinel-2 L2A data within the Theia Land Cover CES framework. Maps for 2021, 2020, 2019, and 2018 use a 23-categories nomenclature. For earlier maps in 2017 and 2016, a fully compatible 17-classes nomenclature is employed.

MethodThis dataset includes a detailed example for using our method (described in paper linked to below) to digitize historical land-use maps in R.MapsWe also release all of the Swedish land-use maps that we digitized for this project. This includes the Economic Map of Sweden (Ekonomiska kartan) over Sweden's 15 southernmost counties (7069 25 km2 sheets), plus 11 sheets of the District Economic Map (Häradsekonomiska kartan - but see http://bolin.su.se/data/Cousins-2015 for more accurate manual digitization).SvenskaHär kan du ladda ner 7069 Ekonomiska kartblad som vi digitaliserade över södra Sverige. En kort beskrivning av metoden publicerades i tidningen Kart & Bildteknik (se länk nedan).--UpdatesVersion 2: The digitized Economic Maps have been resampled so that they are all at a 1m resolution. In the original version they were all very close to 1m but not exactly the same, which made mosaicking difficult. This should be easier now. We now also link to the published paper in Methods in Ecology and Evolution.For more information, please see the readme file. For help or collaboration, please contact alistair.auffret@natgeo.su.se. If you use the data here in your work or research, please cite the publication appropriately.

Human use of the land has a large effect on the structure of terrestrial ecosystems and the dynamics of biogeochemical cycles. For this reason, terrestrial ecosystem and biogeochemistry models require moderate resolution information on land use in order to make realistic predictions. Few such datasets currently exist.

This collection consists of output from models that estimate the spatial pattern of land use in four land-use categories by relating a high-resolution land-cover dataset to state-level census data on land use. The models have been parameterized using a goodness-of-fit measure.

The land cover product used was from the IGBP DISCover global product, derived from 1 km AVHRR imagery, with 16 land cover classes (Belward et al., 1999). Land-use data at state-level resolution came from the USDA's Major Land Uses database (USDA, 1996), aggregated into the four general land-use categories described below.

The model was used to generate maps of land use in 1992 for the conterminous U.S. at 0.5 degree spatial resolution. Two different parameterization schemes were used to spatially interpolate land use from land cover, based on the state-level land use census data: 1) a National Parameterization, and 2) a Regional Parameterization.

For the National Parameterization, a single parameterization relating aggregate land cover and state-level land use. For the Regional Parameterization, a separate parameterization was used for each of seven different regions. The seven regions include: Northeast, Southeast, East North-central, West North-central, Southern Plains, Mountain, and Pacific. These regions are substantially different in terms of land use and land cover. In both cases, the results are a nationally gridded map at 0.5 degrees of land use categories for cropland, pasture/range, forest, and other land use; the other land use category is also further spilt into three additional subcategories (forested, non-forested, non-vegetated).

This project is currently being extended to other regions of the globe, and for other time periods, where both land use census data and image-derived land cover data are available.

Available Datasets:

1) US Land Use - 1992 National Parameterization 2) US Land Use - 1992 Regional Parameterization

Each dataset has 4 major land use categories and 3 subcategories of the Other major land use category.

This map was created to be used in the CBF website map gallery as updated satellite imagery content for the Chesapeake Bay watershed.This map includes the Chesapeake Bay watershed boundary, state boundaries that intersect the watershed boundary, and NLCD 2019 Land Cover data as well as a imagery background. This will be shared as a web application on the CBF website within the map gallery subpage.

Under various scenarios, land use changes in Belgium are simulated at 10-meter resolution. Three SSP-RCP scenarios were used to model the land use trends in the present (2020) and the year 2050 at the national level in Belgium. Key inputs to the model include regional land use demand, quantification of the suitability of grid cells for different land use types, and a reference land cover map. The 10 meter-resolution baseline land use map of Belgium was sourced from the European Space Agency (ESA) WorldCover for the reference year 2020. The classification systems ESA is different from LUH2. To make these datasets comparable for land use simulations, we performed reclassification based on the guidelines provided by Pérez-Hoyos et al. (2012); Dong et al. (2018); Liao et al. (2020) to unify the land use classes, except water, into six general categories: 1) urban, 2) cropland, 3) pasture, 4) forestry, 5) bare/sparse vegetation, and 6) undefined.

Minnesota's original public land survey plat maps were created between 1848 and 1907 during the first government land survey of the state by the U.S. Surveyor General's Office. This collection of more than 3,600 maps includes later General Land Office (GLO) and Bureau of Land Management maps up through 2001. Scanned images of the maps are available in several digital formats and most have been georeferenced.

The survey plat maps, and the accompanying survey field notes, serve as the fundamental legal records for real estate in Minnesota; all property titles and descriptions stem from them. They also are an essential resource for surveyors and provide a record of the state's physical geography prior to European settlement. Finally, they testify to many years of hard work by the surveying community, often under very challenging conditions.

The deteriorating physical condition of the older maps (drawn on paper, linen, and other similar materials) and the need to provide wider public access to the maps, made handling the original records increasingly impractical. To meet this challenge, the Office of the Secretary of State (SOS), the State Archives of the Minnesota Historical Society (MHS), the Minnesota Department of Transportation (MnDOT), MnGeo and the Minnesota Association of County Surveyors collaborated in a digitization project which produced high quality (800 dpi), 24-bit color images of the maps in standard TIFF, JPEG and PDF formats - nearly 1.5 terabytes of data. Funding was provided by MnDOT.

In 2010-11, most of the JPEG plat map images were georeferenced. The intent was to locate the plat images to coincide with statewide geographic data without appreciably altering (warping) the image. This increases the value of the images in mapping software where they can be used as a background layer.

This dataset, produced by Impact Observatory, Microsoft, and Esri, displays a global map of land use and land cover (LULC) derived from ESA Sentinel-2 imagery at 10 meter resolution for the years 2017 - 2023. Each map is a composite of LULC predictions for 9 classes throughout the year in order to generate a representative snapshot of each year. This dataset was generated by Impact Observatory, which used billions of human-labeled pixels (curated by the National Geographic Society) to train a deep learning model for land classification. Each global map was produced by applying this model to the Sentinel-2 annual scene collections from the Mircosoft Planetary Computer. Each of the maps has an assessed average accuracy of over 75%. These maps have been improved from Impact Observatory’s previous release and provide a relative reduction in the amount of anomalous change between classes, particularly between “Bare” and any of the vegetative classes “Trees,” “Crops,” “Flooded Vegetation,” and “Rangeland”. This updated time series of annual global maps is also re-aligned to match the ESA UTM tiling grid for Sentinel-2 imagery. Data can be accessed directly from the Registry of Open Data on AWS, from the STAC 1.0.0 endpoint, or from the IO Store for a specific Area of Interest (AOI).

Use this global model layer when performing analysis across continents. This layer displays a global land cover map and model for the year 2050 at a pixel resolution of 300m. ESA CCI land cover from the years 2010 and 2018 were used to create this prediction.Variable mapped: Projected land cover in 2050.Data Projection: Cylindrical Equal AreaMosaic Projection: Cylindrical Equal AreaExtent: Global Cell Size: 300mSource Type: ThematicVisible Scale: 1:50,000 and smallerSource: Clark UniversityPublication date: April 2021What you can do with this layer?This layer may be added to online maps and compared with the ESA CCI Land Cover from any year from 1992 to 2018. To do this, add Global Land Cover 1992-2018 to your map and choose the processing template (image display) from that layer called “Simplified Renderer.” This layer can also be used in analysis in ecological planning to find specific areas that may need to be set aside before they are converted to human use.Links to the six Clark University land cover 2050 layers in ArcGIS Living Atlas of the World:There are three scales (country, regional, and world) for the land cover and vulnerability models. They’re all slightly different since the country model can be more fine-tuned to the drivers in that particular area. Regional (continental) and global have more spatially consistent model weights. Which should you use? If you’re analyzing one country or want to make accurate comparisons between countries, use the country level. If mapping larger patterns, use the global or regional extent (depending on your area of interest). Land Cover 2050 - GlobalLand Cover 2050 - RegionalLand Cover 2050 - CountryLand Cover Vulnerability to Change 2050 GlobalLand Cover Vulnerability to Change 2050 RegionalLand Cover Vulnerability to Change 2050 CountryWhat these layers model (and what they don’t model)The model focuses on human-based land cover changes and projects the extent of these changes to the year 2050. It seeks to find where agricultural and urban land cover will cover the planet in that year, and what areas are most vulnerable to change due to the expansion of the human footprint. It does not predict changes to other land cover types such as forests or other natural vegetation during that time period unless it is replaced by agriculture or urban land cover. It also doesn’t predict sea level rise unless the model detected a pattern in changes in bodies of water between 2010 and 2018. A few 300m pixels might have changed due to sea level rise during that timeframe, but not many.The model predicts land cover changes based upon patterns it found in the period 2010-2018. But it cannot predict future land use. This is partly because current land use is not necessarily a model input. In this model, land set aside as a result of political decisions, for example military bases or nature reserves, may be found to be filled in with urban or agricultural areas in 2050. This is because the model is blind to the political decisions that affect land use.Quantitative Variables used to create ModelsBiomassCrop SuitabilityDistance to AirportsDistance to Cropland 2010Distance to Primary RoadsDistance to RailroadsDistance to Secondary RoadsDistance to Settled AreasDistance to Urban 2010ElevationGDPHuman Influence IndexPopulation DensityPrecipitationRegions SlopeTemperatureQualitative Variables used to create ModelsBiomesEcoregionsIrrigated CropsProtected AreasProvincesRainfed CropsSoil ClassificationSoil DepthSoil DrainageSoil pHSoil TextureWere small countries modeled?Clark University modeled some small countries that had a few transitions. Only five countries were modeled with this procedure: Bhutan, North Macedonia, Palau, Singapore and Vanuatu.As a rule of thumb, the MLP neural network in the Land Change Modeler requires at least 100 pixels of change for model calibration. Several countries experienced less than 100 pixels of change between 2010 & 2018 and therefore required an alternate modeling methodology. These countries are Bhutan, North Macedonia, Palau, Singapore and Vanuatu. To overcome the lack of samples, these select countries were resampled from 300 meters to 150 meters, effectively multiplying the number of pixels by four. As a result, we were able to empirically model countries which originally had as few as 25 pixels of change.Once a selected country was resampled to 150 meter resolution, three transition potential images were calibrated and averaged to produce one final transition potential image per transition. Clark Labs chose to create averaged transition potential images to limit artifacts of model overfitting. Though each model contained at least 100 samples of "change", this is still relatively little for a neural network-based model and could lead to anomalous outcomes. The averaged transition potentials were used to extrapolate change and produce a final hard prediction and risk map of natural land cover conversion to Cropland and Artificial Surfaces in 2050.39 Small Countries Not ModeledThere were 39 countries that were not modeled because the transitions, if any, from natural to anthropogenic were very small. In this case the land cover for 2050 for these countries are the same as the 2018 maps and their vulnerability was given a value of 0. Here were the countries not modeled:AndorraAntigua and BarbudaBarbadosCape VerdeComorosCook IslandsDjiboutiDominicaFaroe IslandsFrench GuyanaFrench PolynesiaGibraltarGrenadaGuamGuyanaIcelandJan MayenKiribatiLiechtensteinLuxembourgMaldivesMaltaMarshall IslandsMicronesia, Federated States ofMoldovaMonacoNauruSaint Kitts and NevisSaint LuciaSaint Vincent and the GrenadinesSamoaSan MarinoSeychellesSurinameSvalbardThe BahamasTongaTuvaluVatican CityIndex to land cover values in this dataset:The Clark University Land Cover 2050 projections display a ten-class land cover generalized from ESA Climate Change Initiative Land Cover. 1 Mostly Cropland2 Grassland, Scrub, or Shrub3 Mostly Deciduous Forest4 Mostly Needleleaf/Evergreen Forest5 Sparse Vegetation6 Bare Area7 Swampy or Often Flooded Vegetation8 Artificial Surface or Urban Area9 Surface Water10 Permanent Snow and Ice

China’s high-resolution hybrid cropland maps (CCropLand30) by integrating state-of-the-art remote sensing LULC products (GlobeLand30, GLAD, CLUD, CLCD, and CACD) with the recently released county-level cropland area data from the latest national land survey (NLDS). CCropLand30 has a higher pixel-scale accuracy than the input maps, and exhibit better spatial agreement with the NLDS data. The dataset can provide great support for cropland monitoring, management, and various research fields such as water resources, agriculture, and climate change. Five hydbrid maps circa the years 2000, 2005, 2010, 2015, and 2020, were provided in the dataset.

This is a web map service (WMS) for the 10-metre Land Cover Map 2023. The map presents the and surface classified into 21 UKCEH land cover classes, based upon Biodiversity Action Plan broad habitats.UKCEH’s automated land cover algorithms classify 10 m pixels across the whole of UK. Training data were automatically selected from stable land covers over the interval of 2020 to 2022. A Random Forest classifier used these to classify four composite images representing per season median surface reflectance. Seasonal images were integrated with context layers (e.g., height, aspect, slope, coastal proximity, urban proximity and so forth) to reduce confusion among classes with similar spectra.Land cover was validated by organising the 10 m pixel classification into a land parcel framework (the LCM2023 classified land parcels product). The classified land parcels were compared to known land cover producing a confusion matrix to determine overall and per class accuracy.

This layer displays a global map of land use/land cover (LULC) derived from ESA Sentinel-2 imagery at 10m resolution. Each year is generated with Impact Observatory’s deep learning AI land classification model, trained using billions of human-labeled image pixels from the National Geographic Society. The global maps are produced by applying this model to the Sentinel-2 Level-2A image collection on Microsoft’s Planetary Computer, processing over 400,000 Earth observations per year.The algorithm generates LULC predictions for nine classes, described in detail below. The year 2017 has a land cover class assigned for every pixel, but its class is based upon fewer images than the other years. The years 2018-2023 are based upon a more complete set of imagery. For this reason, the year 2017 may have less accurate land cover class assignments than the years 2018-2023.Variable mapped: Land use/land cover in 2017, 2018, 2019, 2020, 2021, 2022, 2023Source Data Coordinate System: Universal Transverse Mercator (UTM) WGS84Service Coordinate System: Web Mercator Auxiliary Sphere WGS84 (EPSG:3857)Extent: GlobalSource imagery: Sentinel-2 L2ACell Size: 10-metersType: ThematicAttribution: Esri, Impact ObservatoryWhat can you do with this layer?Global land use/land cover maps provide information on conservation planning, food security, and hydrologic modeling, among other things. This dataset can be used to visualize land use/land cover anywhere on Earth. This layer can also be used in analyses that require land use/land cover input. For example, the Zonal toolset allows a user to understand the composition of a specified area by reporting the total estimates for each of the classes. NOTE: Land use focus does not provide the spatial detail of a land cover map. As such, for the built area classification, yards, parks, and groves will appear as built area rather than trees or rangeland classes.Class definitionsValueNameDescription1WaterAreas where water was predominantly present throughout the year; may not cover areas with sporadic or ephemeral water; contains little to no sparse vegetation, no rock outcrop nor built up features like docks; examples: rivers, ponds, lakes, oceans, flooded salt plains.2TreesAny significant clustering of tall (~15 feet or higher) dense vegetation, typically with a closed or dense canopy; examples: wooded vegetation, clusters of dense tall vegetation within savannas, plantations, swamp or mangroves (dense/tall vegetation with ephemeral water or canopy too thick to detect water underneath).4Flooded vegetationAreas of any type of vegetation with obvious intermixing of water throughout a majority of the year; seasonally flooded area that is a mix of grass/shrub/trees/bare ground; examples: flooded mangroves, emergent vegetation, rice paddies and other heavily irrigated and inundated agriculture.5CropsHuman planted/plotted cereals, grasses, and crops not at tree height; examples: corn, wheat, soy, fallow plots of structured land.7Built AreaHuman made structures; major road and rail networks; large homogenous impervious surfaces including parking structures, office buildings and residential housing; examples: houses, dense villages / towns / cities, paved roads, asphalt.8Bare groundAreas of rock or soil with very sparse to no vegetation for the entire year; large areas of sand and deserts with no to little vegetation; examples: exposed rock or soil, desert and sand dunes, dry salt flats/pans, dried lake beds, mines.9Snow/IceLarge homogenous areas of permanent snow or ice, typically only in mountain areas or highest latitudes; examples: glaciers, permanent snowpack, snow fields.10CloudsNo land cover information due to persistent cloud cover.11RangelandOpen areas covered in homogenous grasses with little to no taller vegetation; wild cereals and grasses with no obvious human plotting (i.e., not a plotted field); examples: natural meadows and fields with sparse to no tree cover, open savanna with few to no trees, parks/golf courses/lawns, pastures. Mix of small clusters of plants or single plants dispersed on a landscape that shows exposed soil or rock; scrub-filled clearings within dense forests that are clearly not taller than trees; examples: moderate to sparse cover of bushes, shrubs and tufts of grass, savannas with very sparse grasses, trees or other plants.Classification ProcessThese maps include Version 003 of the global Sentinel-2 land use/land cover data product. It is produced by a deep learning model trained using over five billion hand-labeled Sentinel-2 pixels, sampled from over 20,000 sites distributed across all major biomes of the world.The underlying deep learning model uses 6-bands of Sentinel-2 L2A surface reflectance data: visible blue, green, red, near infrared, and two shortwave infrared bands. To create the final map, the model is run on multiple dates of imagery throughout the year, and the outputs are composited into a final representative map for each year.The input Sentinel-2 L2A data was accessed via Microsoft’s Planetary Computer and scaled using Microsoft Azure Batch.CitationKarra, Kontgis, et al. “Global land use/land cover with Sentinel-2 and deep learning.” IGARSS 2021-2021 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2021.AcknowledgementsTraining data for this project makes use of the National Geographic Society Dynamic World training dataset, produced for the Dynamic World Project by National Geographic Society in partnership with Google and the World Resources Institute.

We present a land use management map for Europe. This map is subject to updates, the updates are described in the pdf description. The land use management map is based on the land cover (base map) which has nine different land covers for Europe. The land use management map further divides these land covers into 20 land use management classes based on different inputs. The map is made to be used as a baseline of land use in Europe for land use modelling.