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 2018-2024 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 2018-2024. Key Properties Variable mapped: Land use/land cover in 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024Source 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 ObservatoryAnalysis: Optimized for analysisClass Definitions: ValueNameDescription1WaterAreas 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.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.Usage Information and Best PracticesProcessing TemplatesThis layer includes a number of preconfigured processing templates (raster function templates) to provide on-the-fly data rendering and class isolation for visualization and analysis. Each processing template includes labels and descriptions to characterize the intended usage. This may include for visualization, for analysis, or for both visualization and analysis. VisualizationThe default rendering on this layer displays all classes.There are a number of on-the-fly renderings/processing templates designed specifically for data visualization.By default, the most recent year is displayed. To discover and isolate specific years for visualization in Map Viewer, try using the Image Collection Explorer. AnalysisIn order to leverage the optimization for analysis, the capability must be enabled by your ArcGIS organization administrator. More information on enabling this feature can be found in the ‘Regional data hosting’ section of this help doc.Optimized for analysis means this layer does not have size constraints for analysis and it is recommended for multisource analysis with other layers optimized for analysis. See this group for a complete list of imagery layers optimized for analysis.Prior to running analysis, users should always provide some form of data selection with either a layer filter (e.g. for a specific date range, cloud cover percent, mission, etc.) or by selecting specific images. To discover and isolate specific images for analysis in Map Viewer, try using the Image Collection Explorer.Zonal Statistics is a common tool used for understanding the composition of a specified area by reporting the total estimates for each of the classes. GeneralIf you are new to Sentinel-2 LULC, the Sentinel-2 Land Cover Explorer provides a good introductory user experience for working with this imagery layer. For more information, see this Quick Start Guide.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. 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.

This web map displays the land use/land cover (LULC) timeseries layer derived from ESA Sentinel-2 imagery at 10m resolution. The visualization uses blend modes and is best used in the new Map Viewer. The time slider can be used to advance through the five years of data from 2017-2021. There are also a series of bookmarks for the locations below:Urban growth examplesOuagadougouCairo/GizaDubai, UAEKaty, Texas, USALoudoun County, VirginiaInfrastructureIstanbul International Airport, TurkeyGrand Ethiopian Renaissance Dam, EthiopiaDeforestationBorder of Acre and Rondonia states, BrazilHarz Mountains, GermanyWetlands lossPantanal, BrazilParana river, ArgentinaVegetation changing after fireNorthern California: Paradise, Redding, Clear Lake, Santa Rosa, Mendocino National ForestKangaroo Island, AustraliaVictoria and NSW, AustraliaYakutia, RussiaHurricane ImpactAbaco Island, BahamasRecent Lava FlowHawaii IslandSurface MiningBrown Coal, Cottbus, GermanyLand ReclamationMarkermeer, NetherlandsEconomic DevelopmentNorth vs South Korea

About the dataLand use land cover (LULC) maps are an increasingly important tool for decision-makers in many industry sectors and developing nations around the world. The information provided by these maps helps inform policy and land management decisions by better understanding and quantifying the impacts of earth processes and human activity.ArcGIS Living Atlas of the World provides a detailed, accurate, and timely LULC map of the world. The data is the result of a three-way collaboration among Esri, Impact Observatory, and Microsoft. For more information about the data, see Sentinel-2 10m Land Use/Land Cover Time Series.About the appOne of the foremost capabilities of this app is the dynamic change analysis. The app provides dynamic visual and statistical change by comparing annual slices of the Sentinel-2 10m Land Use/Land Cover data as you explore the map.Overview of capabilities:Visual change analysis with either 'Step Mode' or 'Swipe Mode'Dynamic statistical change analysis by year, map extent, and classFilter by selected land cover classRegional class statistics summarized by administrative boundariesImagery mode for visual investigation and validation of land coverSelect imagery renderings (e.g. SWIR to visualize forest burn scars)Data download for offline use

The 10m Annual Land Use Land Cover (LULC) map is produced Impact Observatory, Microsoft, and Esri collaboratively. The data collection is derived from ESA Sentinel-2 imagery at 10m resolution globaly using Impact Ovservatory's state of the art deep learning AI land classification model which is trained by billions of human-labeled image pixels. There are 9 LULC classes generated by the algorithm, including Built, Crops, Trees, Water, Rangeland, Flooded Vegetation, Snow/Ice, Bare Ground, and Clouds.

Land cover describes the surface of the earth. Land cover maps are useful in urban planning, resource management, change detection, agriculture, and a variety of other applications in which information related to earth surface is required. Land cover classification is a complex exercise and is hard to capture using traditional means. Deep learning models are highly capable of learning these complex semantics, giving superior results.Using the modelFollow the guide to use the model. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS.Fine-tuning the modelThis model can be fine-tuned using the Train Deep Learning Model tool. Follow the guide to fine-tune this model.InputRaster, mosaic dataset, or image service. (Preferred cell size is 10 meters.)Note: This model is trained to work on Sentinel-2 Imagery datasets which are in WGS 1984 Web Mercator (auxiliary sphere) coordinate system (WKID 3857).OutputClassified raster with the same classes as in Corine Land Cover (CLC) 2018.Applicable geographiesThis model is expected to work well in Europe and the United States.Model architectureThis model uses the UNet model architecture implemented in ArcGIS API for Python.Accuracy metricsThis model has an overall accuracy of 82.41% with Level-1C imagery and 84.0% with Level-2A imagery, for CLC class level 2 classification (15 classes). The table below summarizes the precision, recall and F1-score of the model on the validation dataset.ClassLevel-2A ImageryLevel-1C ImageryPrecisionRecallF1 ScorePrecisionRecallF1 ScoreUrban fabric0.810.830.820.820.840.83Industrial, commercial and transport units0.740.650.690.730.660.7Mine, dump and construction sites0.630.520.570.690.550.61Artificial, non-agricultural vegetated areas0.700.460.550.670.470.55Arable land0.860.900.880.860.890.87Permanent crops0.760.730.740.750.710.73Pastures0.750.710.730.740.710.73Heterogeneous agricultural areas0.610.560.580.620.510.56Forests0.880.930.900.880.920.9Scrub and/or herbaceous vegetation associations0.740.690.720.730.670.7Open spaces with little or no vegetation0.870.840.850.850.820.84Inland wetlands0.810.780.800.820.770.79Maritime wetlands0.740.760.750.870.890.88Inland waters0.940.920.930.940.910.92Marine waters0.980.990.980.970.980.98This model has an overall accuracy of 90.79% with Level-2A imagery for CLC class level 1 classification (5 classes). The table below summarizes the precision, recall and F1-score of the model on the validation dataset.ClassPrecisionRecallF1 ScoreArtificial surfaces0.850.810.83Agricultural areas0.900.910.91Forest and semi natural areas0.910.920.92Wetlands0.770.700.73Water bodies0.960.970.96Training dataThis model has been trained on the Corine Land Cover (CLC) 2018 with the same Sentinel 2 scenes that were used to produce the database. Scene IDs for the imagery were available in the metadata of the dataset.Sample resultsHere are a few results from the model. To view more, see this story.

The Sentinel-2 10m Land Use/Land Cover Time Series displays a global map of land use/land cover (LULC) derived from ESA Sentinel-2 imagery at 10m resolution.The World Imagery (Firefly) map is designed to be used as a neutral imagery basemap, with de-saturated colors, that is useful for overlaying other brightly styled layers.

Sentinel2GlobalLULC is a deep learning-ready dataset of RGB images from the Sentinel-2 satellites designed for global land use and land cover (LULC) mapping. Sentinel2GlobalLULC v2.1 contains 194,877 images in GeoTiff and JPEG format corresponding to 29 broad LULC classes. Each image has 224 x 224 pixels at 10 m spatial resolution and was produced by assigning the 25th percentile of all available observations in the Sentinel-2 collection between June 2015 and October 2020 in order to remove atmospheric effects (i.e., clouds, aerosols, shadows, snow, etc.). A spatial purity value was assigned to each image based on the consensus across 15 different global LULC products available in Google Earth Engine (GEE).

Our dataset is structured into 3 main zip-compressed folders, an Excel file with a dictionary for class names and descriptive statistics per LULC class, and a python script to convert RGB GeoTiff images into JPEG format. The first folder called "Sentinel2LULC_GeoTiff.zip" contains 29 zip-compressed subfolders where each one corresponds to a specific LULC class with hundreds to thousands of GeoTiff Sentinel-2 RGB images. The second folder called "Sentinel2LULC_JPEG.zip" contains 29 zip-compressed subfolders with a JPEG formatted version of the same images provided in the first main folder. The third folder called "Sentinel2LULC_CSV.zip" includes 29 zip-compressed CSV files with as many rows as provided images and with 12 columns containing the following metadata (this same metadata is provided in the image filenames):

• Land Cover Class ID: is the identification number of each LULC class
• Land Cover Class Short Name: is the short name of each LULC class
• Image ID: is the identification number of each image within its corresponding LULC class
• Pixel purity Value: is the spatial purity of each pixel for its corresponding LULC class calculated as the spatial consensus across up to 15 land-cover products
• GHM Value: is the spatial average of the Global Human Modification index (gHM) for each image
• Latitude: is the latitude of the center point of each image
• Longitude: is the longitude of the center point of each image
• Country Code: is the Alpha-2 country code of each image as described in the ISO 3166 international standard. To understand the country codes, we recommend the user to visit the following website where they present the Alpha-2 code for each country as described in the ISO 3166 international standard:https: //www.iban.com/country-codes
• Administrative Department Level1: is the administrative level 1 name to which each image belongs
• Administrative Department Level2: is the administrative level 2 name to which each image belongs
• Locality: is the name of the locality to which each image belongs
• Number of S2 images : is the number of found instances in the corresponding Sentinel-2 image collection between June 2015 and October 2020, when compositing and exporting its corresponding image tile

For seven LULC classes, we could not export from GEE all images that fulfilled a spatial purity of 100% since there were millions of them. In this case, we exported a stratified random sample of 14,000 images and provided an additional CSV file with the images actually contained in our dataset. That is, for these seven LULC classes, we provide these 2 CSV files:

• A CSV file that contains all exported images for this class
• A CSV file that contains all images available for this class at spatial purity of 100%, both the ones exported and the ones not exported, in case the user wants to export them. These CSV filenames end with "including_non_downloaded_images".

To clearly state the geographical coverage of images available in this dataset, we included in the version v2.1, a compressed folder called "Geographic_Representativeness.zip". This zip-compressed folder contains a csv file for each LULC class that provides the complete list of countries represented in that class. Each csv file has two columns, the first one gives the country code and the second one gives the number of images provided in that country for that LULC class. In addition to these 29 csv files, we provided another csv file that maps each ISO Alpha-2 country code to its original full country name.

© Sentinel2GlobalLULC Dataset by Yassir Benhammou, Domingo Alcaraz-Segura, Emilio Guirado, Rohaifa Khaldi, Boujemâa Achchab, Francisco Herrera & Siham Tabik is marked with Attribution 4.0 International (CC-BY 4.0)

Important Note: This item is in mature support as of February 2023 and will be retired in December 2025. A new version of this item is available for your use. Esri recommends updating your maps and apps to use the new version. This layer displays change in pixels of the Sentinel-2 10m Land Use/Land Cover product developed by Esri, Impact Observatory, and Microsoft. Available years to compare with 2021 are 2018, 2019 and 2020. By default, the layer shows all comparisons together, in effect showing what changed 2018-2021. But the layer may be changed to show one of three specific pairs of years, 2018-2021, 2019-2021, or 2020-2021.Showing just one pair of years in ArcGIS Online Map ViewerTo show just one pair of years in ArcGIS Online Map viewer, create a filter. 1. Click the filter button. 2. Next, click add expression. 3. In the expression dialogue, specify a pair of years with the ProductName attribute. Use the following example in your expression dialogue to show only places that changed between 2020 and 2021:ProductNameis2020-2021By default, places that do not change appear as a transparent symbol in ArcGIS Pro. But in ArcGIS Online Map Viewer, a transparent symbol may need to be set for these places after a filter is chosen. To do this:4. Click the styles button. 5. Under unique values click style options. 6. Click the symbol next to No Change at the bottom of the legend. 7. Click the slider next to "enable fill" to turn the symbol off.Showing just one pair of years in ArcGIS ProTo show just one pair of years in ArcGIS Pro, choose one of the layer's processing templates to single out a particular pair of years. The processing template applies a definition query that works in ArcGIS Pro. 1. To choose a processing template, right click the layer in the table of contents for ArcGIS Pro and choose properties. 2. In the dialogue that comes up, choose the tab that says processing templates. 3. On the right where it says processing template, choose the pair of years you would like to display. The processing template will stay applied for any analysis you may want to perform as well.How the change layer was created, combining LULC classes from two yearsImpact Observatory, Esri, and Microsoft used artificial intelligence to classify the world in 10 Land Use/Land Cover (LULC) classes for the years 2017-2021. Mosaics serve the following sets of change rasters in a single global layer: Change between 2018 and 2021Change between 2019 and 2021Change between 2020 and 2021To make this change layer, Esri used an arithmetic operation combining the cells from a source year and 2021 to make a change index value. ((from year * 16) + to year) In the example of the change between 2020 and 2021, the from year (2020) was multiplied by 16, then added to the to year (2021). Then the combined number is served as an index in an 8 bit unsigned mosaic with an attribute table which describes what changed or did not change in that timeframe. Variable mapped: Change in land cover between 2018, 2019, or 2020 and 2021 Data Projection: Universal Transverse Mercator (UTM)Mosaic Projection: WGS84Extent: GlobalSource imagery: Sentinel-2Cell Size: 10m (0.00008983152098239751 degrees)Type: ThematicSource: Esri Inc.Publication date: January 2022What can you do with this layer?Global LULC maps provide information on conservation planning, food security, and hydrologic modeling, among other things. This dataset can be used to visualize land cover anywhere on Earth. This layer can also be used in analyses that require land cover input. For example, the Zonal Statistics tools allow a user to understand the composition of a specified area by reporting the total estimates for each of the classes. Land Cover processingThis map was produced by a deep learning model trained using over 5 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 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. Processing platformSentinel-2 L2A/B data was accessed via Microsoft’s Planetary Computer and scaled using Microsoft Azure Batch.Class definitions1. WaterAreas 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.2. TreesAny significant clustering of tall (~15-m 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).4. Flooded 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.5. CropsHuman planted/plotted cereals, grasses, and crops not at tree height; examples: corn, wheat, soy, fallow plots of structured land.7. Built 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.8. Bare 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.9. Snow/IceLarge homogenous areas of permanent snow or ice, typically only in mountain areas or highest latitudes; examples: glaciers, permanent snowpack, snow fields. 10. CloudsNo land cover information due to persistent cloud cover.11. Rangeland Open 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.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.For questions please email environment@esri.com

The "Sen-2 LULC Dataset" is a collection of 2,13,750+ pre-processed 10 m resolution images representing 7 distinct classes of Land Use Land Cover. The 7 classes are water, Dense forest, Sparse forest, Barren land, Built up, Agriculture land and Fallow land. Multiple classes are present in the single image of the dataset. The Sentinel-2 images of Central India are taken from USGS (United States Geological Survey) EXPLORER (https://earthexplorer.usgs.gov/) with cloud clover percentage ranging from 0 to 0.5. The images are combination of bands 3, 4 and 5 constituting the red, green and blue bands with spectral resolution of 10m. The images are taken within the months of March and April 2022. The images used in the dataset belongs to Sentinel-2 Level-2A product (https://sentinels.copernicus.eu/web/sentinel/user-guides/sentinel-2-msi/product-types/level-2a#:~:text=The%20Level%2D2A%20product%20provides,(UTM%2FWGS84%20projection).). The dataset contains equal number of mask images. The dataset contains 6 folders with train, test and validate images and train, test and validate masks. This dataset can be used for Land Use Land Cover Classification (LULC) of Indian region to build the deep learning models. This dataset is beneficial for LULC classification research.

EuroSAT is a land use and land cover classification dataset. The dataset is based on Sentinel-2 satellite imagery covering 13 spectral bands and consists of 10 LULC classes with a total of 27,000 labeled and geo-referenced images. The dataset is associated with the publications "Introducing EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification" and "EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification".

EuroSAT_RGB.zip contains the RGB version of the dataset, which includes the optical R, G and B frequency bands encoded as JPEG images.

EuroSAT_MS.zip contains the multi-spectral version of the EuroSAT dataset, which includes all 13 Sentinel-2 bands in the original value range.

The SENTINEL-2 Normalized Difference Vegetation Index (NDVI) is a proxy to quantify the vegetation amount. It is defined as NDVI=(NIR-Red)/(NIR+Red) where NIR corresponds to the reflectance in the near infrared band , and Red to the reflectance in the red band. It is closely related to FAPAR and is little scale dependant.

ESA WorldCover 10 m 2021 v200

The European Space Agency (ESA) WorldCover 10 m 2021 product provides a global land cover map for 2021 at 10 m resolution based on Sentinel-1 and Sentinel-2 data. The WorldCover product comes with 11 land cover classes, aligned with UN-FAO's Land Cover Classification System, and has been generated in the framework of the ESA WorldCover project.

The ESA WorldCover 10m 2021 v200 product updates the existing ESA WorldCover 10m 2020 v100 product to 2021 but is produced using an improved algorithm version (v200) compared to the 2020 map. Consequently, since the WorldCover maps for 2020 and 2021 were generated with different algorithm versions (v100 and v200, respectively), changes between the maps should be treated with caution, as they include both real changes in land cover and changes due to the algorithms used.

The WorldCover 2021 v200 product is developed by a consortium lead by VITO Remote Sensing together with partners Brockmann Consult, Gamma Remote Sensing AG, IIASA and Wageningen University

Click here to view the maps

More information about the land cover maps

Product User Manual & Product Validation Report

ESA WorldCover 10 m 2020 v100

The European Space Agency (ESA) WorldCover 10 m 2020 product provides a global land cover map for 2020 at 10 m resolution based on Sentinel-1 and Sentinel-2 data. The WorldCover product comes with 11 land cover classes, aligned with UN-FAO's Land Cover Classification System, and has been generated in the framework of the ESA WorldCover project.

The WorldCover product is developed by a consortium lead by VITO Remote Sensing together with partners Brockmann Consult, CS SI, Gamma Remote Sensing AG, IIASA and Wageningen University

Click here to view the maps

More information about the land cover maps

Product User Manual & Product Validation Report

To evaluate land use and land cover (LULC) maps an independent and representative test dataset is required. Here, a test dataset was generated via stratified random sampling approach across all areas in Fiji not used to generate training data (i.e. all Tikinas which did not contain a training data point were valid for sampling to generate the test dataset). Following equation 13 in Olofsson et al. (2014), the sample size of the test dataset was 834. This was based on a desired standard error of the overall accuracy score of 0.01 and a user's accuracy of 0.75 for all classes. The strata for sampling test samples were the eight LULC classes: water, mangrove, bare soil, urban, agriculture, grassland, shrubland, and trees.

There are different strategies for allocating samples to strata for evaluating LULC maps, as discussed by Olofsson et al. (2014). Equal allocation of samples to strata ensures coverage of rarely occurring classes and minimise the standard error of estimators of user's accuracy. However, equal allocation does not optimise the standard error of the estimator of overall accuracy. Proportional allocation of samples to strata, based on the proportion of the strata in the overall dataset, can result in rarely occurring classes being underrepresented in the test dataset. Optimal allocation of samples to strata is challenging to implement when there are multiple evaluation objectives. Olofsson et al. (2014) recommend a "simple" allocation procedure where 50 to 100 samples are allocated to rare classes and proportional allocation is used to allocate samples to the remaining majority classes. The number of samples to allocate to rare classes can be determined by iterating over different allocations and computing estimated standard errors for performance metrics. Here, the 2021 all-Fiji LULC map, minus the Tikinas used for generating training samples, was used to estimate the proportional areal coverage of each LULC class. The LULC map from 2021 was used to permit comparison with other LULC products with a 2021 layer, notably the ESA WorldCover 10m v200 2021 product.

The 2021 LULC map was dominated by the tree class (74\% of the area classified) and the remaining classes had less than 10\% coverage each. Therefore, a "simple" allocation of 100 samples to the seven minority classes and an allocation of 133 samples to the tree class was used. This ensured all the minority classes had sufficient coverage in the test set while balancing the requirement to minimise standard errors for the estimate of overall accuracy. The allocated number of test dataset points were randomly sampled within each strata and were manually labelled using 2021 annual median RGB composites from Sentinel-2 and Planet NICFI and high-resolution Google Satellite Basemaps.

Data format

The Fiji LULC test data is available in GeoJSON format in the file fiji-lulc-test-data.geojson. Each point feature has two attributes: ref_class (the LULC class manually labelled and quality checked) and strata (the strata the sampled point belongs to derived from the 2021 all-Fiji LULC map). The following integers correspond to the ref_class and strata labels:

1. water
2. mangrove
3. bare earth / rock
4. urban / impervious
5. agriculture
6. grassland
7. shrubland
8. tree

Use

When evaluating LULC maps using test data derived from a stratified sample, the nature of the stratified sampling needs to be accounted for when estimating performance metrics such as overall accuracy, user's accuracy, and producer's accuracy. This is particulary so if the strata do not match the map classes (i.e. when comparing different LULC products). Stehman (2014) provide formulas for estimating performance metrics and their standard errors when using test data with a stratified sampling structure.

To support LULC accuracy assessment a Python package has been developed which provides implementations of Stehman's (2014) formulas. The package can be installed via:

pip install lulc-validation

with documentation and examples here.

In order to compute performance metrics accounting for the stratified nature of the sample the total number of points / pixels available to be sampled in each strata must be known. For this dataset that is:

1. 1779768,
2. 3549325,
3. 541204,
4. 687659,
5. 14279258,
6. 15115599,
7. 4972515,
8. 116131948

Acknowledgements

This dataset was generated with support from a Climate Change AI Innovation Grant.

Overview

A geospatial dataset of point geometries with a land use / land cover label and several remote-sensing derived predictor variables that can be used to train and test a land use / land cover classifier.

This dataset was generated with support from a Climate Change AI Innovation Grant and the Australian Centre for International Agricultural Research.

Each of the point geometries was assigned one of the following class labels:

1. water
2. mangrove
3. bare soil / rock
4. urban / impervious
5. cropland / agriculture
6. grassland
7. shrubland
8. trees

The class property associated with each POINT feature stores the point's class label.

Class definitions

The cropland / agriculture class is defined as any location where agricultural activities associated with cropping or livestock management were visible in high-resolution images. Land that is recently fallow, but where evidence of cropping or grazing activities is present, would be labelled as cropland. Grassland is defined as any low vegetation (e.g. below knee height) without a bush, shrub, or woody structure. Scrubland is defined as any vegetation that is below head height, does not form a closed canopy, and has a clearly visible bush, shrub, or woody structure. Trees are defined as any vegetation greater than head height forming a clear canopy.

Methods

Image interpretation and labelling points with a land cover class was undertaken within a custom Google Earth Engine application. Within a region of interest, a year’s worth of Sentinel-2 images was clustered into 15 classes using a k-means algorithm. A stratified random sample of points was generated for manual labelling using clusters as strata. Ground truth datasets ere generated in the Ba, Magodro, Rewa, Sigatoka, RakiRaki, Sigatoka, Suva, Suva (urban), Lautoka (urban), Noco, Vuya, Nadi, and Labasa regions.

To support image interpretation and labelling a point’s land cover using high-resolution images (Google satellite basemaps), ancillary datasets were used (e.g. Planet and Sentinel-2 images) in conjunction with field verification.

Two quality-checks were applied to the labelled land cover points. First, each point was manually screened and quality checked to ensure consistency in class labels. Second, using Planet NICFI basemaps and Sentinel-2 RGB composites 2019, 2020, and 2021, each of the labelled land cover points was screened for a change in land cover event occurring at any point during those three years. If a change in land cover was observed, the point was dropped from the dataset.

For each labelled point in 2019, 2020, and 2021 features were extracted comprising annual median cloud free spectral reflectance across Sentinel-2 wavebands, monthly NDVI composites, and annual median NDVI, NDBI, NDWI, and GCVI bands and elevation, slope, and aspect bands.

This resulted in a dataset of 13,914 labelled points across three years: 2019, 2020, and 2021. The difference in the number of points across years is due to cloud cover preventing features being generated in some years

Feature definitions:

• B_* - median annual cloud free spectral reflectance for Sentinel-2 wavebands
• ndvi - median annual cloud free NDVI computed from Sentinel-2
• gcvi - median annual cloud free GCVI computed from Sentinel-2
• ndwi - median annual cloud free NDWI computed from Sentinel-2
• ndbi - median annual cloud free NDBI computed from Sentinel-2
• ndvi_* - median monthly cloud free NDVI computed from Sentinel-2
• elevation - elevation computed from SRTM
• aspect - aspect computed from SRTM
• slope - slope computed from SRTM

Sentinel-2 Level-1C imagery with on-the-fly renderings for visualization. This imagery layer pulls directly from theSentinel-2 on AWScollection and is updated daily with new imagery.Sentinel-2 imagery can be applied across a number of industries, scientific disciplines, and management practices. Some applications include, but are not limited to, land cover and environmental monitoring, climate change, deforestation, disaster and emergency management, national security, plant health and precision agriculture, forest monitoring, watershed analysis and runoff predictions, land-use planning, tracking urban expansion, highlighting burned areas and estimating fire severity. Geographic Coverage GlobalContinental land masses from65.4° South to 72.1° North, with these special guidelines:All coastal waters up to 20 km from the shoreAll islands greater than 100 km2All EU islandsAll closed seas (e.g. Caspian Sea)The Mediterranean Sea Temporal Coverage This layer includes a rolling collection of Sentinel-2 imagery acquired within the past 14 months. This layer is updated daily with new imagery. The revisit time for each point on Earth is every 5 days. The number of images available will vary depending on location. Product Level This service provides Level-1C Top of Atmosphere imagery.Alternatively,Sentinel-2 Level-2A is also available. Image Selection/Filtering The most recent and cloud free images are displayed by default. Any image available within the past 14 months can be displayed via custom filtering. Filtering can be done based on attributes such as Acquisition Date, Estimated Cloud Cover, and Tile ID. Tile_ID is computed as [year][month][day]T[hours][minutes][seconds]_[UTMcode][latitudeband][square]_[sequence].More… Visual Rendering Default rendering is Natural Color (bands 4,3,2) with Dynamic Range Adjustment (DRA). The DRA version of each layer enables visualization of the full dynamic range of the images. Rendering (or display) of band combinations and calculated indices is done on-the-fly from the source images via Raster Functions. Various pre-defined Raster Functions can be selected or custom functions created. Available renderings include: Agriculture with DRA,Bathymetric with DRA,Color-Infrared with DRA,Natural Color with DRA,Short-wave Infrared with DRA,Geology with DRA,NDMI Colorized,Normalized Difference Built-Up Index (NDBI),NDWI Raw,NDWI - with VRE Raw,NDVI – with VRE Raw (NDRE),NDVI - VRE only Raw,NDVI Raw,Normalized Burn Ratio,NDVI Colormap. Multispectral Bands BandDescriptionWavelength (µm)Resolution (m)1Coastal aerosol0.433 - 0.453602Blue0.458 - 0.523103Green0.543 - 0.578104Red0.650 - 0.680105Vegetation Red Edge0.698 - 0.713206Vegetation Red Edge0.733 - 0.748207Vegetation Red Edge0.773 - 0.793208NIR0.785 - 0.900108ANarrow NIR0.855 - 0.875209Water vapour0.935 - 0.9556010SWIR – Cirrus1.365 - 1.3856011SWIR-11.565 - 1.6552012SWIR-22.100 - 2.28020Additional Notes Overviews exist with a spatial resolution of 150m and are updated every quarter based on the best and latest imagery available at that time.To work with source images at all scales, the ‘Lock Raster’ functionality is available. NOTE: ‘Lock Raster’ should only be used on the layer for short periods of time, as the imagery and associated record Object IDs may change daily.This ArcGIS Server dynamic imagery layer can be used in Web Maps and ArcGIS Desktop as well as Web and Mobile applications using the REST based Image services API.Images can be exported up to a maximum of 4,000 columns x 4,000 rows per request.Data SourceSentinel-2 imagery is the result of close collaboration between the (European Space Agency) ESA, the European Commission and USGS. Data is hosted by the Amazon Web Services as part of theirRegistry of Open Data. Users can access the imagery fromSentinel-2 on AWS, or alternatively accessEarthExploreror theCopernicus Data Space Ecosystemto download the scenes.For information on Sentinel-2 imagery, seeSentinel-2.

The Dynamic World Training Data is a dataset of over 5 billion pixels of human-labeled ESA Sentinel-2 satellite image, distributed over 24000 tiles collected from all over the world. The dataset is designed to train and validate automated land use and land cover mapping algorithms. The 10m resolution 5.1km-by-5.1km tiles are densely labeled using a ten category classification schema indicating general land use land cover categories. The dataset was created between 2019-08-01 and 2020-02-28, using satellite imagery observations from 2019, with approximately 10% of observations extending back to 2017 in very cloudy regions of the world. This dataset is a component of the National Geographic Society - Google - World Resources Institute Dynamic World project. […]

The dataset contains maps of the main classes of agricultural land use (dominant crop types and other land use types) in Germany, which have been produced annually at the Thünen Institute beginning with the year 2017 on the basis of satellite data. The maps cover the entire open landscape, i.e., the agriculturally used area (UAA) and e.g., uncultivated areas. The map was derived from time series of Sentinel-1, Sentinel-2, Landsat 8 and additional environmental data. Map production is based on the methods described in Blickensdörfer et al. (2022).

All optical satellite data were managed, pre-processed and structured in an analysis-ready data (ARD) cube using the open-source software FORCE - Framework for Operational Radiometric Correction for Environmental monitoring (Frantz, D., 2019), in which SAR and environmental data were integrated.

The map extent covers all areas in Germany that are defined as agricultural land, grassland, small woody features, heathland, peatland or unvegetated areas according to ATKIS Basis-DLM (Geobasisdaten: © GeoBasis-DE / BKG, 2020).

Version v201:Post-processing of the maps included a sieve filter as well as a ruleset for the reduction of non-plausible areas using the Basis-DLM and the digital terrain model of Germany (Geobasisdaten: © GeoBasis-DE / BKG, 2015). The final post-processing step comprises the aggregation of the gridded data to homogeneous objects (fields) based on the approach that is described in Tetteh et al. (2021) and Tetteh et al. (2023).

The maps are available in FlatGeobuf format, which makes downloading the full dataset optional. All data can directly be accessed in QGIS, R, Python or any supported software of your choice using the provided URL to the datasets (right click on the respective data set --> “copy link address”). By doing so the entire map area or only the regions of interest can be accessed. QGIS legend files for data visualization can be downloaded separately.

Class-specific accuracies for each year are proveded in the respective tables. We provide this dataset "as is" without any warranty regarding the accuracy or completeness and exclude all liability.

Mailing list

If you do not want to miss the latest updates, please enroll to our mailing list.

References:Blickensdörfer, L., Schwieder, M., Pflugmacher, D., Nendel, C., Erasmi, S., & Hostert, P. (2022). Mapping of crop types and crop sequences with combined time series of Sentinel-1, Sentinel-2 and Landsat 8 data for Germany. Remote Sensing of Environment, 269, 112831.

BKG, Bundesamt für Kartographie und Geodäsie (2015). Digitales Geländemodell Gitterweite 10 m. DGM10. https://sg.geodatenzentrum.de/web_public/gdz/dokumentation/deu/dgm10.pdf (last accessed: 28. April 2022).

BKG, Bundesamt für Kartographie und Geodäsie (2020). Digitales Basis-Landschaftsmodell. https://sg.geodatenzentrum.de/web_public/gdz/dokumentation/deu/basis-dlm.pdf (last accessed: 28. April 2022).

Frantz, D. (2019). FORCE—Landsat + Sentinel-2 Analysis Ready Data and Beyond. Remote Sensing, 11, 1124.

Tetteh, G.O., Gocht, A., Erasmi, S., Schwieder, M., & Conrad, C. (2021). Evaluation of Sentinel-1 and Sentinel-2 Feature Sets for Delineating Agricultural Fields in Heterogeneous Landscapes. IEEE Access, 9, 116702-116719.

Tetteh, G.O., Schwieder, M., Erasmi, S., Conrad, C., & Gocht, A. (2023). Comparison of an Optimised Multiresolution Segmentation Approach with Deep Neural Networks for Delineating Agricultural Fields from Sentinel-2 Images. PFG – Journal of Photogrammetry, Remote Sensing and Geoinformation Science

National-scale crop type maps for Germany from combined time series of Sentinel-1, Sentinel-2 and Landsat data (2017 to 2021) © 2024 by Schwieder, Marcel; Tetteh, Gideon Okpoti; Blickensdörfer, Lukas; Gocht, Alexander; Erasmi, Stefan; licensed under CC BY 4.0.

Funding was provided by the German Federal Ministry of Food and Agriculture as part of the joint project “Monitoring der biologischen Vielfalt in Agrarlandschaften” (MonViA, Monitoring of biodiversity in agricultural landscapes).

Training dataset for urban green land cover and land use detection for Sentinel-2 satellite images. Samples are pixel-wise labelled scenes over the city of Prague, including bigger parks and smaller vegetation patches within high-density urban areas.

Contains four classes:

• 0: Non-vegetated pixels
• 1: Low recreational vegetation
• 2: High recreational vegetation
• 3: Non-recreational vegetation

ben-ge/ERA-5: BigEarthNet Extended with Geographical and Environmental Data/Environmental Data

M. Mommert, N. Kesseli, J. Hanna, L. Scheibenreif, D. Borth, B. Demir, "ben-ge: Extending BigEarthNet with Geographical and Environmental Data", IEEE International Geoscience and Remote Sensing Symposium, Pasadena, USA, 2023.

ben-ge is a multimodal dataset for Earth observation (https://github.com/HSG-AIML/ben-ge) that serves as an extension to the BigEarthNet dataset. ben-ge complements the Sentinel-1/2 data contained in BigEarthNet by providing additional data modalities:

* elevation data extracted from the Copernicus Digital Elevation Model GLO-30;
* land-use/land-cover data extracted from ESA Worldcover;
* climate zone information extracted from Beck et al. 2018;
* environmental data concurrent with the Sentinel-1/2 observations from the ERA-5 global reanalysis;
* a seasonal encoding.

This archive contains the environmental data of ben-ge, which were extracted from the ERA-5 global reanalysis.

Data

Weather data at the time of observation (temperature at 2 m above the ground, relative humidity, wind vectors at 10 m above the ground) are extracted from the ERA-5 global reanalysis (https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels) for the pressure level at the mean elevation of the observed scene and the time of observation (separately queried for Sentinel-1/2 observations).

Environmental data are available in the file ben-ge_era-5.csv. For each patch, identified through the Sentinel-2 patch_id or the corresponding Sentinel-1 patch id patch_id_s1, the file contains the following parameters:
* atmpressure_level: atmospheric pressure level at which parameters have been queried [mbar]
* temperature_s2: temperature 2m above ground at the time of the Sentinel-2 observation [K]
* temperature_s1: temperature 2m above ground at the time of the Sentinel-1 observation [K]
* wind-u_s2: eastward component of the wind, at a height of 10 meters above the surface of the Earth at the time of the Sentinel-2 observation [m/s]
* wind-u_s1: eastward component of the wind, at a height of 10 meters above the surface of the Earth at the time of the Sentinel-1 observation [m/s]
* wind-v_s2: northward component of the wind, at a height of 10 meters above the surface of the Earth at the time of the Sentinel-2 observation [m/s]
* wind-v_s1: northward component of the wind, at a height of 10 meters above the surface of the Earth at the time of the Sentinel-2 observation [m/s]
* relhumidity_s2: relative humidity at the time of the Sentinel-2 observation [%]
* relhumidity_s1: relative humidity at the time of the Sentinel-1 observation [%]

as extracted from the ERA-5 global reanalysis (https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels) for the patch location. Please see the corresponding documentation for details.

Relevant meta data for the ben-ge dataset are compiled in the file ben-ge_meta.csv. This file resides on the root level of this archive and contains the following data for each patch:
* patch_id: the Sentinel-2 patch id, which plays a central role for cross-referencing different data modalities for individual patches;
* patch_id_s1: the Sentinel-1 patch id for this specific patch;
* timestamp_s2: the timestamp for the Sentinel-2 observation;
* timestamp_s1: the timestamp for the Sentinel-1 observation;
* season_s2: the seasonal encoding (see below) for the time of the Sentinel-2 observation;
* season_s1: the seasonal encoding (see below) for the time of the Sentinel-1 observation;
* lon: longitude (WGS-84) of the center of the patch [degrees];
* lat: latitude (WGS-84) of the center of the patch [degrees];
* climatezone: integer value indicating the climate zone based on Beck et al. 2018 (see below for details).

File and directory structure

This archive contains the following directory and file structure:

|
|--- README (this file)
|--- ben-ge_meta.csv (ben-ge meta data)
|--- ben-ge_era-5.csv (ben-ge environmental data)

To properly conserve the file and directory structure of the ben-ge dataset, please place this archive file on the root level of the ben-ge dataset and then unpack it. Once unpacked, ben-ge/era-5 requires 80 MB of space.

Other data modalities from ben-ge (as well as Sentinel-1/2 data as provided by BigEarthNet, https://bigearth.net/#downloads), may be added as required. For reference, the recommended structure for the full dataset looks as follows:

|
|--- ben-ge_meta.csv (ben-ge meta data)
|--- ben-ge_era-5.csv (ben-ge environmental data)
|--- ben-ge_esaworldcover.csv (patch-wise ben-ge land-use/land-cover data)
|--- dem/ (digital elevation model data)
| |--- S2A_MSIL2A_20171208T093351_3_82_dem.tif
| ...
|--- esaworldcover/ (land-use/land-cover data)
| |--- S2B_MSIL2A_20170914T93030_26_83_esaworldcover.tif
| ...
|--- sentinel-1/ (Sentinel-1 SAR data)
| |--- S1A_IW_GRDH_1SDV_20180219T063851_29UPV_70_43/
| |--- S1A_IW_GRDH_1SDV_20180219T063851_29UPV_70_43_labels_metadata.json (BigEarthNet label file)
| |--- S1A_IW_GRDH_1SDV_20180219T063851_29UPV_70_43_VH.tif (BigEarthNet/Sentinel-1 VH polarization data)
| |--- S1A_IW_GRDH_1SDV_20180219T063851_29UPV_70_43_VV.tif (BigEarthNet/Sentinel-1 VV polarization data)
| ...
|--- sentinel-2/ (Sentinel-2 multispectral data)
| |--- S2B_MSIL2A_20170818T112109_31_83/
| |--- S2B_MSIL2A_20170818T112109_31_83_B01.tif (BigEarthNet/Sentinel-2 Band 1 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B02.tif (BigEarthNet/Sentinel-2 Band 2 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B03.tif (BigEarthNet/Sentinel-2 Band 3 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B04.tif (BigEarthNet/Sentinel-2 Band 4 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B05.tif (BigEarthNet/Sentinel-2 Band 5 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B06.tif (BigEarthNet/Sentinel-2 Band 6 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B07.tif (BigEarthNet/Sentinel-2 Band 7 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B08.tif (BigEarthNet/Sentinel-2 Band 8 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B09.tif (BigEarthNet/Sentinel-2 Band 9 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B11.tif (BigEarthNet/Sentinel-2 Band 11 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B12.tif (BigEarthNet/Sentinel-2 Band 12 data)
| |--- S2B_MSIL2A_20170818T112109_31_83_B8A.tif (BigEarthNet/Sentinel-2 Band 8A data)
| |--- S2B_MSIL2A_20170818T112109_31_83_labels_metadata.json (BigEarthNet label file)
...

More Information

For more information, please refer to https://github.com/HSG-AIML/ben-ge.

Citing ben-ge

If you use data contained in this archive, please cite the following paper:

M. Mommert, N. Kesseli, J. Hanna, L. Scheibenreif, D. Borth, B. Demir, "ben-ge: Extending BigEarthNet with Geographical and Environmental Data", IEEE International Geoscience and Remote Sensing Symposium, Pasadena, USA, 2023.