Find research datasets worth reusing

opencc-by-4.0Oct 2024View details →

Navigating deep learning strategies for large-area land cover mapping using very-high-resolution imagery in Senegal: Validation Data

Rapid advances in deep learning for land cover classification of trees, shrubs and very small agricultural fields using very high-resolution satellite data (< 2 m), has tremendous potential for resolving current challenges in quantifying land cover change in sub-Saharan African (SSA), due to growing demand for food resources. We conducted experiments with different training strategies for scaling up UNet convolutional neural network models for regional land cover mapping with multispectral WorldView (WV)-2 and –3, imagery in three distinct regions of Senegal which has complex seasonal wet/dry conditions and cropland-savanna mosaics.  The validation exercise of this research consisted in validating more than 70,000 km2 across Senegal. The infrastructure was setup in the NASA SMCE system with a total of twelve George Mason University (GMU) students participating as operators. These operators validated more than 59 WV-2 and -3 images, each consisting of 200 stratified points in 5,000 x 5,000-pixel images. This effort resulted in a total of ~35,000 aggregated observations that are available through the eo-validation API for public consumption. Each validation point from this dataset has three individual observations.

opencc-by-4.0Sep 2024View details →

Dataset: Mapping saltmarsh communities in South Portugal using high spatiotemporal resolution satellite imagery

<div> <div> <div> This repository containts the datasets from the article "Mapping saltmarsh communities in South Portugal using high spatiotemporal resolution satellite imagery" (Submitted). The dataset was used in a workflow used to create saltmarsh maps for the Algarve region (South Portugal), focused on the 4 main costal systems of the region: Alvor, Arade, Ria Formosa and Guadiana.   For a description of the methodology see the article [link] and Github repo [link].   <h1>Repository content</h1> <h2>1. system-masks.zip</h2> Contains 4 <code>geojson</code>files with a polygon which delimits the areas included in the study. The files are named after the respective systems that they delimit. Any region outside of these polygons were not used in the analysis. CRS - EPSG:4326 <h2>2. manual-clean-up-masks.gpkg</h2> Polygons which were manually created to mask out (exclude) pixels which were classified as saltmarsh, but are clearly not. File contains a single layer with 52 polygons and one variable. Variables: <ul> <li>system [string] - Which system the polygon delimits</li> </ul> <h2>3. saltmarsh-training-data.gpkg</h2> Data used for supervised model training. Each row represents one quadrat, and each column contains either quadrat identifiers, target classes, or predictor classes. File contains a single layer with 2448 points and 18 variables. Variables: <ul> <li>water_system [string] - Study system in which the quadrat was sampled</li> <li>transect [string] - Name of transect in which the quadrat was sampled</li> <li>quad_id [integer] - Unique identifier per quadrat</li> <li>cluster [integer] - Vegetation cluster identified via hierarchical clustering. They are nested within <code>water_system</code>, and the same number within different systems will not correspond to the same vegetation type.</li> <li>marsh_type [string] - Functional groupings of saltmarsh vegetation (low, middle or high), created by grouping <code>cluster</code> based on niche of the defined clusters.</li> <li>train [boolean] - Was quadrat used in the train (TRUE) or test (FALSE) stage of model training?</li> <li>ndvi [numerical] - Normalized Difference Vegetation Index, calculated from the satellite image mosaic as (nir – red) / (nir + red).</li> <li>ndwi_high [numerical] - Normalized Difference Water Index estimated from images at high tide, calculated as (green – nir) / (green + nir)</li> <li>ndwi_low [numerical] - Normalized Difference Water Index estimated from images at low tide, calculated as (green – nir) / (green + nir)</li> <li>subtime [numerical] - Fraction of time that a cell is estimated to be submerged in water over one year.</li> <li>coastal_blue [numerical] - Surface reflectance values at 443 nm.</li> <li>blue [numerical] - Surface reflectance values at 490 nm.</li> <li>green_i [numerical] - Surface reflectance values at 531 nm.</li> <li>green [numerical] - Surface reflectance values at 565 nm.</li> <li>yellow [numerical] - Surface reflectance values at 610 nm.</li> <li>red [numerical] - Surface reflectance values at 665 nm.</li> <li>rededge [numerical] - Surface reflectance values at 705 nm.</li> <li>nir [numerical] - Surface reflectance values at 865 nm.</li> </ul> <h2>4. saltmarsh-transect-metadata.csv</h2> Comma-delimited file with information about vegetation sampling transects. Each row represents one transect. File contains 6 variables. Variables: <ul> <li>water_system [string] - Study system in which the transect was sampled</li> <li>transect_set [string] - Which set of transects was this transect sampled in? Set A was performed in 2019, set B in 2023.</li> <li>site [string] - Name of the site within the study system. This was used exclusively to plan transects.</li> <li>transect [string] - Name of transect in which the quadrat was sampled</li> <li>date [date yyyy-mm-dd] - Date of transect sampling.</li> <li>notes [string] - Notes taken during transect sampling and which might be relevant to understand data issues.</li> </ul> <h2>5. saltmarsh-vegetation-quadrats.gpkg</h2> Data used for to create vegetation clusters (<code>cluster</code>) and saltmarsh community types (<code>marsh_type</code>). The later was used as the target class in the supervised model training. Each row represents one quadrat, and each column contains either quadrat identifiers, or presence/absence of species. File contains a single layer with 2448 points and 32 variables. Variables: <ul> <li>water_system [string] - Study system in which the transect was sampled</li> <li>transect [string] - Name of transect in which the quadrat was sampled</li> <li>transect_set [string] - Which set of transects was this transect sampled in? Set A was performed in 2019, set B in 2023.</li> <li>quad_id [integer] - Unique identifier per quadrat</li> <li>distance_from_water [integer] - Distance from start of quadrat, which was the point closes to the water where saltmarsh was found for that transect.</li> <li>cluster [integer] - Vegetation cluster identified via hierarchical clustering. They are nested within <code>water_system</code>, and the same number within different systems will not correspond to the same vegetation type.</li> <li>marsh_type [string] - Functional groupings of saltmarsh vegetation (low, middle or high), created by grouping <code>cluster</code> based on niche of the defined clusters.</li> <li>Arthrocaulon.macrostachyum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Tripolium.pannonicum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Atriplex.halimus [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Cistanche.phelypaea [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Atriplex.portulacoides [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Limbarda.crithmoides [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Juncus.effusus [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Limoniastrum.monopetalum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Myriolimon.ferulaceum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Limonium.vulgare [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Phragmites.australis [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Polygonum.maritimum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Puccinellia.maritima [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Salicornia.procumbens [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Salicornia.europaea [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Caroxylon.vermiculatum [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Salicornia.fruticosa [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Salicornia.perennis [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Bolboschoenus.maritimus [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Sporobolus.maritimus [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Spergularia.bocconei [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Suaeda.vera [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Triglochin.maritima [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> <li>Sporobolus.montevidensis [boolean] - Presence (1) or absence (0) of the species with the variable name.</li> </ul> <h2>6. predicted-map.tif</h2> Geotiff file with a single layer for predicted saltmarsh community. Values are:   - no data - Not saltmarsh   - 1 - Low saltmarsh   - 2 - Middle saltmarsh   - 3 - High saltmarsh CRS - EPSG:32629 </div> </div> </div>

opencc-by-4.0Jul 2021View details →

Dataset for: Importance of satellite observations for high-resolution mapping of near-surface NO2 by machine learning

Dataset for: Importance of satellite observations for high-resolution mapping of near-surface NO2 by machine learning This dataset is uploaded as a part of the article by Kim et al. (2021). The dataset is the hourly maps of near-surface nitrogen dioxide (NO2) concentrations at 100 m resolution for an Alpine domain (Switzerland and northern Italy, 6-12 °E, 42-48 °N). The dataset is provided per day (24 hours) in a netcdf (*.nc ~550MB).  In this work, we have generated NO2 hourly maps for Feb. 2019 to May 2020 and, here, we upload for March 2019 only (~16 GB). If you need data for another period of time, please contact Gerrit Kuhlmann (gerrit.kuhlmann@empa.ch) or Minsu Kim (minsu.kim@empa.ch).

opencc-by-4.0Mar 2023View details →

Comparison of high-resolution global canopy height maps and their applicability to biodiversity modelling - dataset

This repository was created to provide datasets related with an article comparing high-resolution global canopy height maps and exploring their applicability to biodiversity modeling in temperate biomes. EBR stands for Entlebuch Biosphere Reserve, MRF stands for Mount Richmond Forest and TAW stands for Trinity Alps Wilderness. The original airborne laser scanning point clouds used for the generation of the canopy height models were sourced from the LINZ Data Service and OpenTopography, and licensed for reuse under the CC BY 4.0 licence (<a href="https://mcas-proxyweb.mcas.ms/certificate-checker?login=false&originalUrl=https%3A%2F%2Fdoi.org.mcas.ms%2F10.5069%2FG97D2SB0%3FMcasTsid%3D20893&McasCSRF=cf3ae9aed6f2016d3ceedda452d422646f4f8e5a5e6270380b370aad4964323a">https://doi.org/10.5069/G97D2SB0</a>); Federal Office of Topography swisstopo (<a href="https://mcas-proxyweb.mcas.ms/certificate-checker?login=false&originalUrl=https%3A%2F%2Fwww.swisstopo.admin.ch.mcas.ms%2Fen%2Fgeodata%2Fheight%2Fsurface3d.html%3FMcasTsid%3D20893&McasCSRF=cf3ae9aed6f2016d3ceedda452d422646f4f8e5a5e6270380b370aad4964323a">https://www.swisstopo.admin.ch/en/geodata/height/surface3d.html</a>); and U.S. Geological Survey (<a href="https://mcas-proxyweb.mcas.ms/certificate-checker?login=false&originalUrl=https%3A%2F%2Fapps.nationalmap.gov.mcas.ms%2Fdownloader%2F%3FMcasTsid%3D20893&McasCSRF=cf3ae9aed6f2016d3ceedda452d422646f4f8e5a5e6270380b370aad4964323a">https://apps.nationalmap.gov/downloader/</a>). The Global Forest Canopy Height Map - GFCH (Potapov et al. 2021; https://glad.umd.edu/dataset/gedi) and the high-resolution canopy height model of the Earth - HRCH (Lang et al. 2022, https://langnico.github.io/globalcanopyheight/) are provided free of charge, without restriction of use under Creative Commons Attribution 4.0 International License. Publications, models, and data products that make use of these datasets must include proper acknowledgement. P. Potapov, X. Li, A. Hernandez-Serna, A. Tyukavina, M.C. Hansen, A. Kommareddy, A. Pickens, S. Turubanova, H. Tang, C.E. Silva, J. Armston, R. Dubayah, J. B. Blair, M. Hofton (2021) Mapping and monitoring global forest canopy height through integration of GEDI and Landsat data. Remote Sensing of Environment, 112165. <a href="https://doi.org/10.1016/j.rse.2020.112165">https://doi.org/10.1016/j.rse.2020.112165</a> Lang, N., Jetz, W., Schindler, K., & Wegner, J. D. (2022). A high-resolution canopy height model of the Earth. arXiv preprint arXiv:2204.08322. R scripts related with this datasets are available at Github (https://github.com/lukasgabor/Comparison-of-high-resolution-global-canopy-height-maps-and-their-applicability; <a href="https://doi.org/10.5281/zenodo.7332716">DOI: 10.5281/zenodo.7332716</a>) In the previous version (1.0) the average was calculated for the canopy height. In this version (1.1), the maximum height is calculated for the canopy height.

edi48/100

High resolution shrub cover raster maps of the Jornada Basin LTER, including JER and CDRRC (2011)

This data package contains two raster shrub cover maps of the Jornada Experimental Range (JER) and Chihuahuan Desert Rangeland Research Center (CDRRC) in southern New Mexico, USA. The maps are derived from 1m resolution National Agriculture Imagery Program (NAIP) aerial photos acquired in 2011. A one-meter resolution raster file lists cover as non-shrub or non-shrub. A hectare resolution raster file contains fractional shrub cover values (0-1) calculated from the one-meter raster.

openCC0Jun 2020View details →

OpenNeuro44/100

Robust functional mapping of layer-selective responses in human lateral geniculate nucleus with high-resolution 7T fMRI

Open the record for dataset details and reuse information.

openCC0Jan 2020View details →

opencc-by-4.0Sep 2020View details →

ACS_Bayelva_class: 302 high-resolution snow cover maps covering the 2012-2017 snowmelt seasons in the Bayelva catchment (Svalbard, Norway)

The ACS_Bayelva_class dataset contains 302 high-resolution binary snow cover images that were obtained by classifying orthrorectified photographs of a 1.77 km^2 area of interest in the Bayelva catchment. This latest version (2.0) of the dataset includes the orthorectified photographs that were used to classify the binary snow cover images. The catchment is close to Ny-Ålesund, the northernmost permanent civilian settlement in the world and a major hub for polar research, in the Norwegian high-Arctic Svalbard archipelago. The imagery has a (roughly) daily temporal resolution and a ground sampling distance (pixel spacing) of 0.5 m. The dataset spans 6 snowmelt seasons, covering the months May-August for the period 2012-2017. The orthophotos were obtained by processing oblique time-lapse photographs taken by a terrestrial automatic camera system (ACS) mounted at 562 m a.s.l. near the summit of Scheteligfjellet (719 m a.s.l.) a few kilometers west of Ny-Ålesund. The orthophotos were manually classified into binary snow cover images (0=no snow, 1=snow) by iteratively selecting a (visually) optimal threshold on the intensity in the blue-band for each image. More details are provided in the study of Aalstad et al. (2020) [a copy is available in this repository] where this dataset was created. The ACS was maintained by scientists from the group of Sebastian Westermann at the Section for Physical Geography and Hydrology in the Department of Geosciences at the University of Oslo, Oslo, Norway.

opencc-by-4.0Jul 2022View details →

High-resolution maps of material stock and population in Germany from 1985 to 2018

Global societal material stocks such as buildings and infrastructure accumulated rapidly within recent decades, along with population growth. Material stocks constitute the physical basis of most socio-economic activities and services, such as mobility, housing, health, or education. The dynamics of stock growth, and its relation to the population that demands those services, is an essential indicator for long-term societal resource use and patterns of emissions. The creation of societal material stock creates path dependencies for future resource use, with an important impact on how the transformation towards sustainable societies can succeed. This dataset features detailed maps of material stock and population for Germany on a 30m grid. The data is based on recent maps of material stock and building volume (compare to Haberl et al. 2021, doi: 10.1021/acs.est.0c05642), recent and historic census data, and a time series of Landsat TM, ETM+, and OLI Earth Observation data. Temporal extent The data contains annual maps from 1985 to 2018. Data format and units Per German federal state, the data come in tiles of 30x30km. The projection is EPSG:3035. The images are compressed GeoTiff files (*.tif). There is a mosaic in GDAL Virtual format (*.vrt), which can readily be opened in most Geographic Information Systems. Please consider the generation of image pyramids before using *.vrt files. All image data has 34 bands, where band 1 is data for 1985, and band 34 is data for 2018. The dataset features <ul> <li>population (Scaled by 100 to reduce data storage size. Divide by 100 to get people per cell)</li> <li>mass (in tons) of … <ul> <li>total material stock <ul> <li>… material stock in buildings <ul> <li>… in commercial and industrial buildings</li> <li>… in multi-family residential buildings</li> <li>… in single-family residential buildings</li> <li>… in high-rise buildings</li> <li>… in lightweight buildings</li> </ul> </li> <li>… material stock in road infrastructure</li> <li>… material stock in rail infrastructure</li> <li>… material stock in other infrastructure</li> </ul> </li> </ul> </li> </ul> Material stock in high-rise and lightweight buildings is not featured in the corresponding publication due to its overall negligible amount. It is, however, included here for completeness. Further information For further information, please see the publication or contact Franz Schug (fschug@wisc.edu). Visit our website to learn more about our project MAT_STOCKS - Understanding the Role of Material Stock Patterns for the Transformation to a Sustainable Society. Corresponding publication Schug, F., Frantz, D., Wiedenhofer, D., Virág, D., Haberl, H., van der Linden, S., Hostert, P. (in rev.): High-resolution mapping of 33 years of material stock and population growth in Germany. Journal of Industrial Ecology Funding This research was funded by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (MAT_STOCKS, grant agreement No 741950).

opencc-by-4.0Nov 2014View details →

Development of a global inundation map at high spatial resolution from topographic downscaling of coarse-scale remote sensing data

Overview: The Global Inundation Extent from Multi-Satellites (GIEMS; Prigent et al. 2007, Papa et al. 2010) downscaled at 15 arc-second (GIEMS-D15; Fluet-Chouinard et al. 2015) was produced through the downscaling of the GIEMS database (natively at 0.25°).  The downscaling procedure predicts the location of surface water cover with an inundation ranking surface generated by bagged decision trees. The decision trees were trained on binary presence/absence of wetland in the GLC2000 global land cover map (Bartholomé & Belward 2005) and used 13 topographic and hydrographic predictors derived from the SRTM-derived HydroSHEDS database (Lehner, Verdin & Jarvis 2008). The downscaling technique to three temporal aggregation of the GIEMS dataset representing three states of land surface inundation extents: mean annual minimum (MAMin; total area, 6.5 × 106 km2), mean annual maximum (MAMax; 12.1 × 106 km2), and long-term maximum (LTMax; 17.3 × 106 km2). The area of MAMin and MAMax from GIEMS were supplemented with the minimum area value from lakes, river and reservoirs from GLWD (Lehner & Döll 2004; classes 1,2,3). LTMax was corrected as the mean area from 3-year rolling maximum from GIEMS and the total wetland area from GLWD (classes 1-12). The accuracy of GIEMS-D15 reflects distribution errors introduced by the downscaling process as well as errors from the original satellite estimates. Yet, a comparison against independent regional wetland maps showed adequate agreement over large floodplains and wetlands. GIEMS-D15 offers a higher resolution delineation of inundated areas than originally offered by GIEMS, allowing for the assessment of global freshwater resources and the study of large floodplain and wetland ecosystems. Projection: WGS84 (EPSG:4326) Geographic extent: <ul> <li>Longitude: -180° to 180°</li> <li>Latitude: -56° to 84°</li> </ul> Spatial resolution: 15 arc-second (500m at equator) Legend (for discrete pixel values): <ul> <li>0 = Upland</li> <li>1 = Mean Annual Minimum (MAMin)</li> <li>2 = Mean Annual Maximum (MAMax)</li> <li>3 = Long Term Maximum (LTMax)</li> </ul>

openmit-licenseJun 2018View details →

Dakar very-high resolution land cover map

This land cover map of Dakar (Senegal) was created from a Pléiades very-high resolution imagery with a spatial resolution of 0.5 meter. The methodology followed a open-source semi-automated framework [1] that rely on <a href="https://grass.osgeo.org/">GRASS GIS</a> using a local unsupervised optimization approach for the segmentation part [2-3]. Description of the files: <ul> <li>"Landcover.zip" : The direct output from the supervised classification using the Random Forest classifier.</li> <li>"Landcover_Postclassif_Level8_Splitbuildings.zip" : Post-processed version of the previous map ("Landcover"), with reduced misclassifications from the original classification (rule-based used to reclassify the errors, with a focus on built-up classes).</li> <li>"Landcover_Postclassif_Level8_modalfilter3.zip" : Smoothed version of the previous product (modal filter with window 3x3 applied on the "Landcover_Postclassif_Level8_Splitbuildings"). </li> <li>"Landcover_Postclassif_Level9_Shadowsback.zip" : Corresponds to the "level8_Splitbuildings" with shadows coming from the original classification.</li> <li>"Dakar_legend_colors.txt" : Text file providing the correspondance between the value of the pixels and the legend labels and a proposition of color to be used.</li> </ul>   References: [1] Grippa, Taïs, Moritz Lennert, Benjamin Beaumont, Sabine Vanhuysse, Nathalie Stephenne, and Eléonore Wolff. 2017. “An Open-Source Semi-Automated Processing Chain for Urban Object-Based Classification.” Remote Sensing 9 (4): 358. <a href="https://doi.org/10.3390/rs9040358">https://doi.org/10.3390/rs9040358</a>. [2] Grippa, Tais, Stefanos Georganos, Sabine G. Vanhuysse, Moritz Lennert, and Eléonore Wolff. 2017. “A Local Segmentation Parameter Optimization Approach for Mapping Heterogeneous Urban Environments Using VHR Imagery.” In Proceedings Volume 10431, Remote Sensing Technologies and Applications in Urban Environments II., edited by Wieke Heldens, Nektarios Chrysoulakis, Thilo Erbertseder, and Ying Zhang, 20. SPIE. <a href="https://doi.org/10.1117/12.2278422">https://doi.org/10.1117/12.2278422</a>. [3] Georganos, Stefanos, Taïs Grippa, Moritz Lennert, Sabine Vanhuysse, and Eleonore Wolff. 2017. “SPUSPO: Spatially Partitioned Unsupervised Segmentation Parameter Optimization for Efficiently Segmenting Large Heterogeneous Areas.” In Proceedings of the 2017 Conference on Big Data from Space (BiDS’17).   Founding:  This dataset was produced in the frame of two research project : MAUPP (<a href="http://maupp.ulb.ac.be">http://maupp.ulb.ac.be</a>) and REACT (<a href="http://react.ulb.be">http://react.ulb.be</a>), funded by the Belgian Federal Science Policy Office (<a href="http://eo.belspo.be/About/Stereo3.aspx">BELSPO</a>).

openmit-licenseJun 2018View details →

Ouagadougou very-high resolution land cover map

This land cover map of Ouagadougou (Burkina Faso) was created from a WorldView3 very-high resolution imagery with a spatial resolution of 0.5 meter. The methodology followed a open-source semi-automated framework [1] that rely on <a href="https://grass.osgeo.org/">GRASS GIS</a> using a local unsupervised optimization approach for the segmentation part [2-3]. Description of the files: <ul> <li>"Landcover.zip" : The direct output from the supervised classification using the Random Forest classifier.</li> <li>"Landcover_Postclassif_Level5_Splitbuildings.zip" : Post-processed version of the previous map ("Landcover"), with reduced misclassifications from the original classification (rule-based used to reclassify the errors, with a focus on built-up classes).</li> <li>"Landcover_Postclassif_Level5_modalfilter3.zip" : Smoothed version of the previous product (modal filter with window 3x3 applied on the "Landcover_Postclassif_Level5_Splitbuildings"). </li> <li>"Landcover_Postclassif_Level6_Shadowsback.zip" : Corresponds to the "level5_Splitbuildings" with shadows coming from the original classification.</li> <li>"Ouaga_legend_colors.txt" : Text file providing the correspondance between the value of the pixels and the legend labels and a proposition of color to be used.</li> </ul>   References: [1] Grippa, Taïs, Moritz Lennert, Benjamin Beaumont, Sabine Vanhuysse, Nathalie Stephenne, and Eléonore Wolff. 2017. “An Open-Source Semi-Automated Processing Chain for Urban Object-Based Classification.” Remote Sensing 9 (4): 358. <a href="https://doi.org/10.3390/rs9040358">https://doi.org/10.3390/rs9040358</a>. [2] Grippa, Tais, Stefanos Georganos, Sabine G. Vanhuysse, Moritz Lennert, and Eléonore Wolff. 2017. “A Local Segmentation Parameter Optimization Approach for Mapping Heterogeneous Urban Environments Using VHR Imagery.” In Proceedings Volume 10431, Remote Sensing Technologies and Applications in Urban Environments II., edited by Wieke Heldens, Nektarios Chrysoulakis, Thilo Erbertseder, and Ying Zhang, 20. SPIE. <a href="https://doi.org/10.1117/12.2278422">https://doi.org/10.1117/12.2278422</a>. [3] Georganos, Stefanos, Taïs Grippa, Moritz Lennert, Sabine Vanhuysse, and Eleonore Wolff. 2017. “SPUSPO: Spatially Partitioned Unsupervised Segmentation Parameter Optimization for Efficiently Segmenting Large Heterogeneous Areas.” In Proceedings of the 2017 Conference on Big Data from Space (BiDS’17).   Founding:  This dataset was produced in the frame of two research project : MAUPP (<a href="http://maupp.ulb.ac.be">http://maupp.ulb.ac.be</a>) and REACT (<a href="http://react.ulb.be">http://react.ulb.be</a>), funded by the Belgian Federal Science Policy Office (<a href="http://eo.belspo.be/About/Stereo3.aspx">BELSPO</a>).

opencc-by-4.0Nov 2022View details →

High resolution cropland agreement map (30 m) circa 2020

Accurate and precise measurements of global cropland extent are needed for monitoring the sustainability of agriculture at all scales. Recent advancement in remote sensing and land cover mapping methods have greatly increased the ability to estimate cropland area distribution and trends. Here the FAO presents a map of cropland agreement produced by consolidating information at pixel level from six high-resolutions maps for circa 2020. The following six high resolution layers were used: ESRI 10 meter LU/LC, FROM-GLC, GLAD, GLC-FCS30, Globeland30 and Worldcover. Two bands are included in the dataset: <ol> <li>Simple agreement (values between 1 and 6)</li> <li>Detailed agreement (values between 1 and 63)</li> </ol> The map, developed in the Google Earth Engine platform, combines the 6 land cover/cropland layers to show their cropland agreement on pixel level at a spatial resolution of 30 meters. The simple agreement has pixel values that range from 1 (only 1 dataset classifies as cropland) to 6 (all datasets agree on presence of cropland). Pixels with a value of 0 indicate pixels where all datasets agree on absence of cropland. The second band includes a detailed agreement, showing which combination of the 6 datasets classify a pixel as cropland. The overview table (DetailedAgreement_LookupTable.xlsx) shows what the pixel values of this detailed agreement (from 1 to 63) correspond to. The dataset has been uploaded in 16 tiles, in the preview below and in the file "ACroplandAgreement_30m_Tiles.png" the extent of each tile can be found. For more information on FAO statistics on land cover and land use: FAO. 2022. Land use statistics and indicators. Global, regional and country trends, 2000–2020. FAOSTAT Analytical Brief, no. 48. Rome. <a href="https://doi.org/10.4060/cc0963en">https://doi.org/10.4060/cc0963en</a> FAO. 2021. Land cover statistics. Global, regional and country trends, 2000–2019. FAOSTAT Analytical Brief Series No. 37. Rome.

opencc-by-4.0Feb 2023View details →

Predictive high-resolution mapping of sea floor rock cover for the UK and Ireland

Predicted seabed rock cover using the machine learning algorithm Catboost and marine environmental predictors. Supporting data for T4.1 of Horizon 2020 project FutureMARES.

opencc-by-4.0Oct 2023View details →

High-resolution maps of rubber and rubber-related deforestation for Southeast Asia

This dataset contains maps of rubber plantations in 2021, and maps of rubber-related deforestation between 1993-2016 for Southeast Asia. The rubber maps have a 10 m pixel size, and the deforestation maps have a 30 m pixel size. The dataset and the methods for generating it are described in Wang et al. 2023. High-resolution maps show that rubber causes substantial deforestation. Nature. Please note that an update of this dataset will follow in September 2025.

opencc-by-4.0Dec 2019View details →

High resolution global industrial and smallholder oil palm map for 2019

The dataset contains 634 100x100 km tiles, covering areas where oil palm plantations were detected. The file 'grid.shp' contains the grid that covers the potential distribution of oil palm. The file 'grid_withOP.shp' shows the 100x100 grid squares with presence of oil palm plantations. The classified images (‘oil_palm_map’ folder, in geotiff format) are the output of the convolutional neural network based on Sentinel-1 and Sentinel-2 half-year composites. The images have a spatial resolution of 10 meters and contain three classes: [1] Industrial closed-canopy oil palm plantations, [2] Smallholder closed-canopy oil palm plantations, and [3] other land covers/uses that are not closed canopy oil palm. The file ‘Validation_points_GlobalOilPalmLayer_2019.shp’ includes the 13,495 points that were used to validate the product. Each point includes the attribute ‘Class’, which is the labelled class assigned by visual interpretation, and the attribute ‘predClass, which reflects the predicted class by the convolutional neural network. The ‘Class’ and ‘predClass’ values are the same as the raster files: [1] Industrial closed-canopy oil palm plantations, [2] Smallholder closed-canopy oil palm plantations, and [3] other land covers/uses that are not closed canopy oil palm. See article for additional information: Descals, Adrià, et al. "High-resolution global map of smallholder and industrial closed-canopy oil palm plantations." Earth System Science Data 13.3 (2021): 1211-1231.   Changelog v1: - The analysis was extended to Sri Lanka, South India, and countries in Eastern Africa where oil palm can potentially grow. - The validation dataset only includes the points drawn by simple random sampling and stratified random sampling in the grid cells where the IUCN industrial layer detected oil palm. - The 'Class' and 'predClass' values in the validation dataset were reclassified with the same values as the raster images: [1] Industrial plantations, [2] Smallholder plantations, and [3] Other land covers/uses.

dryad40/100

Data from: Solanum pennellii (LA5240) backcross inbred lines (BILs) for high resolution mapping in tomato

Wild species are an invaluable source of new traits for crop improvement. Over the years the tomato community bred cultivated lines that carry introgressions from different species of the tomato tribe to facilitate trait discovery and mapping. The next phase in such projects is to find the genes that drive the identified phenotypes. This can be achieved by genotyping a few thousand individuals resulting in fine-mapping that can potentially identify the causative gene. To couple trait discovery and fine mapping we are presenting large, recombination-rich, Backcross Inbred Line (BIL) populations involving an unexplored accession of the wild, green-fruited species Solanum pennellii (LA5240; the Lost Accession) with two modern tomato inbreds: LEA, determinate, and TOP indeterminate. The LEA and TOP BILs are in BC2F6-8 generation and include 1,400 and 500 lines respectively. The BILs were genotyped with ~5,000 SPET markers, showing that in the euchromatic regions there was one recombinant every 17-18 Kb while in the heterochromatin a recombinant every 600-700 Kb (TOP and LEA respectively). To gain perspective on the topography of recombination we compared five independent members of the self-pruning gene family with their respective neighboring genes; based on PCR markers, in all cases we found recombinants. Further mapping analysis of two known morphological mutations that segregated in the BILs (Self-pruning and Hair), showed that the maximal delimited intervals were 73 Kb and 210 Kb respectively and included the known causative genes. The LOST_BILs provide a solid framework to study traits derived from a tolerant wild tomato.

opencc-zeroJan 2024View details →

opencc-by-4.0Sep 2023View details →

Estimation of forest height and biomass from open-access multi-sensor satellite imagery and GEDI Lidar data: high-resolution maps of metropolitan France

Maps of forest height, aboveground biomass (AGB)* and volume (VOL)* at 10 m spatial resolution for the year 2020 on France.  * AGB and Volume maps are available on request. The methodology and validation of the maps are presented here: https://hal.science/hal-04249151 Please cite : David Morin, Milena Planells, Stéphane Mermoz, Florian Mouret. Estimation of forest height and biomass from open-access multi-sensor satellite imagery and GEDI Lidar data: high-resolution maps of metropolitan France. 2023. hal-04249151

opencc-by-4.0Apr 2024View details →

The Adult Adansonia digitata L. (baobab tree) distribution map derived from very high resolution satelite imagery for 2010s across the Sahel at 1km resolution

The baobab tree (Adansonia digitata L.) is an integral part of rural livelihoods throughout the African continent. However, the combined effects of climate change and increasing global demand for baobab products are currently exerting pressure on the sustainable utilization of these resources. Here we employ sub-meter resolution satellite imagery to identify nearly 3 million baobab trees in the Sahel, a dryland region of 1.5 million km2. This achievement is considered an essential step towards improving valuable woody species' management and monitoring system. The map's overall underestimate bias is 0.27. To prevent mismanagement of this specific tree species, we aggregated every single adult baobab tree map to 1 × 1 km grids. We also classified the baobab trees using the tree crown diameters( small: 3-9m; medium 9m-13m; large: >13m).  The baobab tree count map is also available for these three different size classes.

opencc-by-4.0May 2024View details →

High-Resolution Canopy Fuel Maps Based on GEDI: A Foundation for Wildfire Modeling in Germany

Open access publication under review. Visit <a href="https://ee-forestfuels-ger.projects.earthengine.app/view/gedi-fuels">this Earth Engine app</a> to explore the data interactively.   Abstract: Forest fuels are essential for wildfire behavior modeling and risk assessments but difficult to quantify accurately. An increase in fire frequency in recent years, particularly in regions traditionally not prone to fire, such as central Europe, has increased demands for large-scale remote sensing fuel information. This study develops a methodology for mapping canopy fuels over large areas (Germany) at high spatial resolution, exclusively relying on open remote sensing data. We propose a two-step approach where we first use measurements from NASA’s GEDI instrument to estimate canopy fuel variables at the footprint level, before predicting high-resolution raster maps. Instead of using field measurements, we generate (GEDI-) footprint-level estimates for Canopy (Base) Height (CH, CBH), Cover (CC), Bulk Density (CBD), and Fuel Load (CFL) by segmenting airborne LiDAR point clouds and processing tree-level metrics with allometric crown biomass models. To predict footprint-level canopy fuels we fit and tune Random Forest models, which are cross-validated using k-fold Nearest Neighbor Distance Matching. Predictions at >1.6 M GEDI footprints and biophysical raster covariates are combined with a Universal Kriging method to produce countrywide maps at 20-meter resolution. Agreement (RMSE/R²) with validation data (from the same population) was strong for footprint-level predictions and moderate for map predictions. A validation with estimates based on National Forest Inventory data revealed low to modest agreement. Better accuracy was achieved for variables related to height (CH, CBH) rather than to cover or biomass (CBD, CFL). Error analysis pointed towards a mixture of biases in model predictions and validation data, as well as underestimation of model prediction standard errors. Contributing factors may be simplification through allometric equations and spatial and temporal mismatch of data inputs. The proposed workflow has the potential to support regions where wildfire is an emerging issue, and fuel and field information is scarce or unavailable.   Data: This repository contains modeling data, model objects (R), and predicted maps. The TIFF-files each have six bands, which includes (1) the final Universal Kriging result, (2) the linear model prediction (3) the prediction of residual Kriging, (4) the Kriging variance, (5) the linear model prediction standard error, and (6) Universal Kriging standard error.   Disclaimer: Maps in this repository are predicted using canopy fuel estimates from GEDI measurements. These are limited the region between 51.6° North and South. Map predictions exceeding this range should be considered an extrapolation of the model to an unknown biophysical domain. Error maps (6) can aid in utilizing our canopy fuel maps.