This data set, a collaboration between the GLAD (Global Land Analysis & Discovery) lab at the University of Maryland, Google, USGS, and NASA, measures areas of tree cover loss across all global land (except Antarctica and other Arctic islands) at approximately 30 × 30 meter resolution. The data were generated using multispectral satellite imagery from the Landsat 5 thematic mapper (TM), the Landsat 7 thematic mapper plus (ETM+), and the Landsat 8 Operational Land Imager (OLI) sensors. Over 1 million satellite images were processed and analyzed, including over 600,000 Landsat 7 images for the 2000-2012 interval, and more than 400,000 Landsat 5, 7, and 8 images for updates for the 2011-2022 interval. The clear land surface observations in the satellite images were assembled and a supervised learning algorithm was applied to identify per pixel tree cover loss.In this data set, “tree cover” is defined as all vegetation greater than 5 meters in height, and may take the form of natural forests or plantations across a range of canopy densities. Tree cover loss is defined as “stand replacement disturbance,” or the complete removal of tree cover canopy at the Landsat pixel scale. Tree cover loss may be the result of human activities, including forestry practices such as timber harvesting or deforestation (the conversion of natural forest to other land uses), as well as natural causes such as disease or storm damage. Fire is another widespread cause of tree cover loss, and can be either natural or human-induced.This data set has been updated five times since its creation, and now includes loss up to 2022 (Version 1.10). The analysis method has been modified in numerous ways, including new data for the target year, re-processed data for previous years (2011 and 2012 for the Version 1.1 update, 2012 and 2013 for the Version 1.2 update, and 2014 for the Version 1.3 update), and improved modelling and calibration. These modifications improve change detection for 2011-2022, including better detection of boreal loss due to fire, smallholder rotation agriculture in tropical forests, selective losing, and short cycle plantations. Eventually, a future “Version 2.0” will include reprocessing for 2000-2010 data, but in the meantime integrated use of the original data and Version 1.7 should be performed with caution. Read more about the Version 1.7 update here.When zoomed out (< zoom level 13), pixels of loss are shaded according to the density of loss at the 30 x 30 meter scale. Pixels with darker shading represent areas with a higher concentration of tree cover loss, whereas pixels with lighter shading indicate a lower concentration of tree cover loss. There is no variation in pixel shading when the data is at full resolution (≥ zoom level 13).The tree cover canopy density of the displayed data varies according to the selection - use the legend on the map to change the minimum tree cover canopy density threshold.

The forest cover data for the year 2000 is part of the WWF Deforestation Fronts Report. WCS for data download (Working with Web Coverage Service (WCS)): https://maps.globilportal.org/server/services/Forests/Global_Forest_Cover_2000/ImageServer/WCSServer?SERVICE=WCS&REQUEST=GetCapabilitiesResolution: A 250m x 250m (6.25ha) spatial resolution was selected for this analysis.Canopy threshold: When using data depicting percent tree cover, a threshold value of 25% was adopted.Multiple available remote sensing datasets were assessed in order to establish the likely extent of forests. After analysing the quality of various remote sensing products, it became evident that in order to obtain a global assessment of forest cover loss, no single approach/data source would work everywhere. This is because each available dataset adopts different definitions of forest, uses different thresholds of tree canopy cover to define forest, and comprises different timeframes. Those discrepancies lead to different estimates of forest cover. In addition, each remote sensing product has its own limitations in terms of area of coverage and timeframes of analysis. In order to address these limitations, an all-available data approach was used as a way to undertake the forest loss assessment by having all the datasets compensating one another. Using this approach, each dataset can complement the other datasets, thus contributing to achieve higher accuracy in classification.A majority vote based on the consensus theoretic classification method was developed to inform the condition of each location in terms of current forest presence.The global forest cover map of 2000 was derived by adding the global forest map for 2018 and the global deforested area map during the period from 2000 to 2017. Thereby, the forest cover map of 2000 is equal to the addition of the present (2018) forest cover map and the forest loss areas from 2000 to 2017. The rationales are that the main purpose of this study is to determine forest cover loss and due to the advancement of technology, the forest detection closer to the present is more accurate than that in the past.

This dataset includes input data used in the following article: Vieilledent G., C. Grinand, F. A. Rakotomalala, R. Ranaivosoa, J.-R. Rakotoarijaona, T. F. Allnutt, and F. Achard. 2018. Combining global tree cover loss data with historical national forest-cover maps to look at six decades of deforestation and forest fragmentation in Madagascar. Biological Conservation. 222: 189-197. [doi:10.1016/j.biocon.2018.04.008]. bioRxiv: 147827. Results from this article have been updated for the periods 2010-2015 and 2015-2017.

The amount of tree cover around forest remnants was calculated using 30-m-resolution global maps of forest extent and change from 2000 through 2023, version 1.11 (Hansen et al. 2013). Each pixel value represents the percentage of canopy closure for all vegetation taller than 5 m in height for the year 2000, which was classified as either forest (≥ 50% tree cover) or non-forest (Hasui et al. 2024). Therefore, such binarization is used loosely, as tree cover can comprise both natural forests and tree plantations (e.g., monocultures of oil palm, rubber, or eucalypt) (Tropek et al. 2014). Despite this, tree cover mostly represents natural forests, and tree plantations are a high-quality matrix type for forest species, given the structural similarity between the habitat (forest) and the matrix (tree plantation) (Prevedello & Vieira 2010). When bird surveys were conducted before 2002, we used the tree cover map for the year 2000. For surveys conducted from 2002 to 2017, we updated the tree cover map by subtracting the cumulative forest loss from 2001 up to the year prior to each survey’s first year (Moulatlet et al. 2021).

The World Resources Institute and Google DeepMind created a global map of the dominant drivers of tree cover loss from 2001 to 2022 at 1 km spatial resolution. We used a deep learning model to classify seven driver classes: permanent agriculture, hard commodities, shifting cultivation, forest management, wildfires, settlements & infrastructure, and other natural disturbances. As part of the study, we collected a set of training samples through interpretation of very high resolution satellite imagery and developed a single world-wide customized residual convolutional neural network model (ResNet) using satellite data (Landsat and Sentinel-2) and ancillary biophysical and population data. In addition, we collected a stratified random sample of validation plots through interpretation of very high resolution satellite imagery to estimate the accuracy of the final classification map.

In this repository, we make the following available:

• The training and validation data, available in two separate files. Both datasets were collected by a team of image interpreters and assessed for quality by two additional interpreters. Note that while for the validation data the quality of the primary driver was rigorously assessed, the secondary driver wasn’t subject to the same level of quality control.

• The global drivers of forest loss raster (drivers_forest_loss_1km.tif), including the discrete classification and probabilities for each class.

Creator & Contact

Created by World Resources Institute and Google DeepMind

Contact: Michelle Sims: Michelle.Sims@wri.org

Definitions

A driver is defined as the direct cause of tree cover loss, and can include both temporary disturbances (natural or anthropogenic) or permanent loss of tree cover due to a change to a non-forest land use (e.g., deforestation). The dominant driver is defined as the direct driver that caused the majority of tree cover loss within each 1 km cell over the time period.

Classes are defined as follows:

The Tree Cover Dynamics (TCD) Conterminous United States (CONUS) dataset is a suite of 30 m wall-to-wall products, derived from USGS Landsat-4, Landsat-5 and Landsat-7 Collection 1 Analysis Ready Data (ARD), defining for each year: (i) the estimated percent tree cover (PTC), (ii) if tree cover loss is detected, the estimated percent tree cover decrease from the previous year (ΔPTC), (iii) if tree cover loss is detected, the Landsat acquisition dates bounding the tree cover loss event (i.e., the last valid observation before the loss, and the first valid observation after the loss) and (iv) a forest status thematic map (three thematic classes: stable forest, stable non-forest, forest cover loss). The products are available for every year from 1985 to 2019. The dataset is provided as georeferenced GeoTIFF images, defined in the CONUS Albers Equal-Area Conic map projection at 30m resolution.

The forest cover data for the year 2018 is part of the WWF Deforestation Fronts Report. WCS for data download (Working with Web Coverage Service (WCS)): https://maps.globilportal.org/server/services/Forests/Global_Forest_Cover_2018/ImageServer/WCSServer?SERVICE=WCS&REQUEST=GetCapabilities Resolution: A 250m x 250m (6.25ha) spatial resolution was selected for this analysis.Canopy threshold: When using data depicting percent tree cover, a threshold value of 25% was adopted.Multiple available remote sensing datasets were assessed in order to establish the likely extent of forests. After analysing the quality of various remote sensing products, it became evident that in order to obtain a global assessment of forest cover loss, no single approach/data source would work everywhere. This is because each available dataset adopts different definitions of forest, uses different thresholds of tree canopy cover to define forest, and comprises different timeframes. Those discrepancies lead to different estimates of forest cover. In addition, each remote sensing product has its own limitations in terms of area of coverage and timeframes of analysis. In order to address these limitations, an all-available data approach was used as a way to undertake the forest loss assessment by having all the datasets compensating one another. Using this approach, each dataset can complement the other datasets, thus contributing to achieve higher accuracy in classification.A majority vote based on the consensus theoretic classification method was developed to inform the condition of each location in terms of current forest presence.The 2018 forest cover map was produced by updating a 2015 map. This was done by subtracting the deforestation data produced for the 2015-2017 period. The total deforested areas from 2000 to 2017 were masked out from the present (2018) forest map.

This data layer references data from a high-resolution tree canopy change-detection layer for Seattle, Washington. Tree canopy change was mapped by using remotely sensed data from two time periods (2016 and 2021). Tree canopy was assigned to three classes: 1) no change, 2) gain, and 3) loss. No change represents tree canopy that remained the same from one time period to the next. Gain represents tree canopy that increased or was newly added, from one time period to the next. Loss represents the tree canopy that was removed from one time period to the next. Mapping was carried out using an approach that integrated automated feature extraction with manual edits. Care was taken to ensure that changes to the tree canopy were due to actual change in the land cover as opposed to differences in the remotely sensed data stemming from lighting conditions or image parallax. Direct comparison was possible because land-cover maps from both time periods were created using object-based image analysis (OBIA) and included similar source datasets (LiDAR-derived surface models, multispectral imagery, and thematic GIS inputs). OBIA systems work by grouping pixels into meaningful objects based on their spectral and spatial properties, while taking into account boundaries imposed by existing vector datasets. Within the OBIA environment a rule-based expert system was designed to effectively mimic the process of manual image analysis by incorporating the elements of image interpretation (color/tone, texture, pattern, location, size, and shape) into the classification process. A series of morphological procedures were employed to ensure that the end product is both accurate and cartographically pleasing. No accuracy assessment was conducted, but the dataset was subjected to manual review and correction.

University of Vermont Spatial Analysis Laboratory in collaboration with City of Seattle.

This dataset consists of City of Seattle SDOT Urban Forestry Management Units which cover the following tree canopy categories:

• Existing tree canopy percent
• Possible tree canopy - vegetation percent
• Relative percent change
• Absolute percent change

For more information, please see the 2021 Tree Canopy Assessment.

This data layer references data from a high-resolution tree canopy change-detection layer for Seattle, Washington. Tree canopy change was mapped by using remotely sensed data from two time periods (2016 and 2021). Tree canopy was assigned to three classes: 1) no change, 2) gain, and 3) loss. No change represents tree canopy that remained the same from one time period to the next. Gain represents tree canopy that increased or was newly added, from one time period to the next. Loss represents the tree canopy that was removed from one time period to the next. Mapping was carried out using an approach that integrated automated feature extraction with manual edits. Care was taken to ensure that changes to the tree canopy were due to actual change in the land cover as opposed to differences in the remotely sensed data stemming from lighting conditions or image parallax. Direct comparison was possible because land-cover maps from both time periods were created using object-based image analysis (OBIA) and included similar source datasets (LiDAR-derived surface models, multispectral imagery, and thematic GIS inputs). OBIA systems work by grouping pixels into meaningful objects based on their spectral and spatial properties, while taking into account boundaries imposed by existing vector datasets. Within the OBIA environment a rule-based expert system was designed to effectively mimic the process of manual image analysis by incorporating the elements of image interpretation (color/tone, texture, pattern, location, size, and shape) into the classification process. A series of morphological procedures were employed to ensure that the end product is both accurate and cartographically pleasing. No accuracy assessment was conducted, but the dataset was subjected to manual review and correction.University of Vermont Spatial Analysis LaboratoryThis dataset consists of hexagons 50-acres in area, or several city blocks. The dataset covers the following tree canopy categories:Existing tree canopy percentPossible tree canopy - vegetation percentRelative percent changeAbsolute percent changeAverage maximum afternoon temperature (F)Tree canopy percentage & average afternoon temperature (F)For more information, please see the 2021 Tree Canopy Assessment.

A 6-in resolution tree canopy change (2010 - 2017) dataset derived from the 2017 Light Detection and Ranging (LiDAR) data capture. This dataset represents a "top-down" mapping perspective and all tree polygons are classed as: (1) No Change, (2) Gain, (3) Loss. No change indicates that this portion of the canopy has undergone no modifications during the time period. Gain indicates that new tree canopy has appeared during the time period. Loss indicates that this portion of the tree canopy was removed during the time period. To learn more about this dataset, visit the interactive "Understanding the 2017 New York City LiDAR Capture" Story Map -- https://maps.nyc.gov/lidar/2017/ Please see the following link for additional documentation on this dataset -- https://github.com/CityOfNewYork/nyc-geo-metadata/blob/master/Metadata/Metadata_TreeCanopyChange.md

The Tree Cover Loss indicator shows year-by-year tree cover loss, defined as stand level replacement of vegetation greater than 5 meters, within the selected area. The tree cover loss data set is a collaboration of the University of Maryland, Google, USGS, and NASA, and uses Landsat satellite images to map annual tree cover loss at a 30 × 30 meter resolution. Note that “tree cover loss” is not the same as “deforestation” – tree cover loss includes change in both natural and planted forest, and does not need to be human caused. The data from 2011 onward were produced with an updated methodology that may capture additional loss. Comparisons between the original 2001-2010 data and future years should be performed with caution.

AG.LND.FRLS.HA. Shows year-by-year tree cover loss, defined as stand level replacement of vegetation greater than 5 meters, within the selected area. The tree cover loss data set is a collaboration of the University of Maryland, Google, USGS, and NASA, and uses Landsat satellite images to map annual tree cover loss at a 30 × 30 meter resolution. Note that “tree cover loss” is not the same as “deforestation” – tree cover loss includes change in both natural and planted forest, and does not need to be human caused. The data from 2011 onward were produced with an updated methodology that may capture additional loss. Comparisons between the original 2001-2010 data and future years should be performed with caution. The World Bank’s ESG Data Draft dataset provides information on 17 key sustainability themes spanning environmental, social, and governance categories.

Global Forest Watch (GFW) is an online platform that provides data and tools for monitoring forests. By harnessing cutting-edge technology, GFW allows anyone to access near real-time information about where and how forests are changing around the world. Data on tree cover gain and loss have been obtained from Global Forest Watch. Tree cover gain data is from the Global Land Analysis & Discovery (GLAD) lab at the University of Maryland and "measures areas of tree cover gain from the year 2000 to 2020 across the globe at 30x30 meter resolution" (Global Forest Watch). This data was aggregated to country/economy level. Tree cover loss data is a collaboration of the University of Maryland, Google, USGS, and NASA, and uses Landsat satellite images to map annual tree cover loss at a 30 × 30 meter resolution. This collection includes only a subset of indicators from the source dataset.

The USDA Forest Service (USFS) builds multiple versions of percent tree canopy cover data, in order to serve needs of multiple user communities. These datasets encompass CONUS, Coastal Alaska, Hawaii, U.S. Virgin Islands and Puerto Rico. There are three versions of data within the 2016 TCC Product Suite, which include: The initial model outputs referred to as the Analytical data; A masked version of the initial output referred to as Cartographic data; And a modified version built for the National Land Cover Database and referred to as NLCD data, which includes a canopy cover change dataset derived from subtraction of datasets for the nominal years of 2011 and 2016. The Analytical data are the initial model outputs generated in the production workflow. These data are best suited for users who will carry out their own detailed statistical and uncertainty analyses on the dataset and place lower priority on the visual appearance of the dataset for cartographic purposes. Datasets for the nominal years of 2011 and 2016 are available. The Cartographic products mask the initial model outputs to improve the visual appearance of the datasets. These data are best suited for users who prioritize visual appearance of the data for cartographic and illustrative purposes. Datasets for the nominal years of 2011 and 2016 are available. The NLCD data are the result of further processing of the masked data. The goal was to generate three coordinated components. The components are (1) a dataset for the nominal year of 2011, (2) a dataset for the nominal year of 2016, and (3) a dataset that captures the change in canopy cover between the two nominal years of 2011 and 2016. For the NLCD data, the three components meet the criterion of 2011 TCC + change in TCC = 2016 TCC. These NLCD data are best suited for users who require a coordinated three-component data stack where each pixel's values meet the criterion of 2011 TCC + change in TCC = 2016 TCC. Datasets for the nominal years of 2011 and 2016 are available, as well as a dataset that captures the change (loss or gain) in canopy cover between those two nominal years of 2011 and 2016, in areas where change was identified. These tree canopy cover data are accessible for multiple user communities, through multiple channels and platforms, as listed below: Analytical USFS Tree Canopy Cover Datasets (Download) USFS Enterprise Data Warehouse (Image Service) Cartographic USFS Tree Canopy Cover Datasets (Download) USFS Enterprise Data Warehouse (Map Service) NLCD Multi-Resolution Land Characteristics (MRLC) Consortium (Download) USFS Enterprise Data Warehouse (Image Service) The Hawaii TCC 2011 cartographic dataset is comprised of a single layer. The pixel values range from 0 to 99 percent. The background is represented by the value 255. The dataset has data gaps due to consistent clouds/shadows in the Landsat images used for modeling. These data gaps are represented by the value 110.

Summary

This repository contains spatial datasets with metadata on land cover, tree canopy change, and estimated tree points and crown polygons for New York City (NYC; New York, USA) as of 2021, made available by The Nature Conservancy, New York Cities Program and developed under contract by the University of Vermont Spatial Analysis Lab. The datasets are provided herein with high-level background and information; additional analysis, particularly on tree canopy change and distribution across NYC considering various geogrpahic units are planned for release in a forthcoming report by The Nature Conservancy. For questions about these data, contact Michael Treglia, Lead Scientist with The Nature Conservancy, New York Cities Program, at michael.treglia@tnc.org.

Datasets included here are as follows (file names in italics):

• Land cover as of 2021 (landcover_nyc_2021_6in.tif):
  • Raster dataset with six-inch (15.24 centimeter) pixel resolution, delineating land covers as: 1) tree canopy (with crowns greater than eight feet [2.44 meters] tall; 2) grass/shrub (including vegetation less than or equal to eight feet [2.44 feet] tall; 3) bare ground; 4) open water; 5) building; 6) road; 7) other impervious; and 8) railroad. This is intended to serve as an update to high-resolution land cover data for 2010 and 2017 made available by the City of New York.
• Tree canopy change during 2017-2021 (treecanopychange_nyc_2017_2021_6in.tif):
  • Raster dataset with six-inch (15.24 centimeter) pixel resolution, with pixels that were estimated tree canopy in 2017 (based on 2017 land cover data) or 2021 delineated as: 1) canopy that did not change (“no change”); 2) canopy that was gained (“gain”); 3) canopy that was lost (“loss”). This is intended to be an updated tree canopy change dataset, analogous to a canopy change dataset for 2010-2017 made available by the City of New York.
• Estimated tree points, crown polygons, and objects as of 2021 (Trees_Centroids_Crown_Objects_2021.gdb.zip):
  • The approximated locations (centroids) and approximated tree crowns as circles (shapes), and tree objects themselves based on canopy data (objects) for individual trees with crowns taller than eight feet (2.44 meters); in cases where there are trees with overlapping crowns, only the top trees are captured. These data are based on automated processing of the tree canopy class from the land cover data; additional methodological details are included in the metadata for this dataset. Given the height cutoff, that this dataset only captures the trees seen from above, and the large number of understory trees in some areas (e.g., forested natural areas), and limits in the automated processing this is not intended to be a robust census of trees in NYC, but may serve as useful for some purposes. Unlike the land cover and tree canopy change datasets, no directly comparable datasets for NYC from past years that we are aware of.

These datasets were based on object-based image analysis of a combination of 2021 Light Detection and Ranging (LiDAR; data available from the State of New York) for tree canopy and tree location/crown data in particular) along with high-resolution aerial imagery (from 2021 via the USDA National Agriculture Inventory Program and from 2022 via the New York State GIS Clearinghouse), followed by manual corrections. The general methods used to develop the land cover and tree canopy datasets are described in MacFaden et al. (2012). A per-pixel accuracy assessment of the land cover data with 1,999 points estimated an overall accuracy of 95.52% across all land cover classes, and 99.06% for tree canopy specifically (a critical focal area for this project). Iterative review of the data and subject matter expertise were contributed by from The Nature Conservancy and the NYC Department of Parks and Recreation.

While analyses of tree canopy and tree canopy change across NYC are pending, those interested can review a report that includes analyses of the most recent data (2010-2017) and a broad consideration of the NYC urban forest, The State of the Urban Forest in New York City (Treglia et al 2021).

References

MacFaden, S. W., J. P. M. O’Neil-Dunne, A. R. Royar, J. W. T. Lu, and A. G. Rundle. 2012. High-resolution tree canopy mapping for New York City using LIDAR and object-based image analysis. Journal of Applied Remote Sensing 6(1):063567.

Treglia, M.L., Acosta-Morel, M., Crabtree, D., Galbo, K., Lin-Moges, T., Van Slooten, A., & Maxwell, E.N. (2021). The State of the Urban Forest in New York City. The Nature Conservancy. doi: 10.5281/zenodo.5532876

Terms of Use

© The Nature Conservancy. This material is provided as-is, without warranty under a Creative Commons Attribution-NonCommercial-ShareAlike International 4.0 (CC BY-NC-SA 4.0) license.

• The Nature Conservancy (TNC) oversaw development of these data and reserves all rights in the data provided.
• TNC makes no guarantee of accuracy or completeness.
• Data are for informational purposes and are not suitable for legal, engineering, or surveying purposes. Data do not represent an on-the-ground survey and represent only the approximate relative location of feature boundaries.
• TNC is not obligated to update/maintain the data to reflect changing conditions.
• Commercial use is not allowed.
• Redistribution (sublicensing) is allowed, provided all accompanying metadata as well as these Terms of Use are provided, unaltered, alongside the data.
• TNC should be credited as the data source in derivative works, following the recommended citation provided herein.
• Users are advised to pay attention to the contents of this metadata document.

Recommended Citation

If using any of these datasets, please cite the work according to the following recommended citation:

The Nature Conservancy. 2024. New York City Land Cover (2021), Tree Canopy Change (2017-2021), and Estimated Tree Location and Crown Data (2021). Developed under contract by the University of Vermont Spatial Analysis Laboratory. doi: 10.5281/zenodo.14053441.

Technical Notes about the Spatial Data

All spatial data are provided in the New York State Plan Long Island Zone (US survey foot) coordinate reference system, EPSG 2263. The land cover and tree canopy change datasets are made available as raster data in Cloud Optimized GeoTIFF format (.tif), with associated metadata files as .xml files. The vector data of estimated tree locations and crown objects and shapes are made available in a zipped Esri File Geodatabase, with metadata stored within the File Geodatabase.

The Copernicus High Resolution Forest Layer Tree Cover Change Mask (TCCM) 2015-2018 raster product provides information on the change between the reference years 2015 and 2018 and consists of 4 thematic classes (unchanged areas with no tree cover / new tree cover / loss of tree cover / unchanged areas with tree cover) at 20m spatial resolution and covers EEA38 area and the United Kingdom.The Tree Cover Change Mask (TCCM) 2015-2018 is a change product based on the binary Tree Cover Masks (TCMs) of the primary status layers Dominant Leaf Type 2015 at 20m spatial resolution and Dominant Leaf Type 2018 at 10m spatial resolution. First, the 2018 product has been aggregated to 20m to enable a map-to-map comparison. The therof derived pixel-based change map is categorised into loss and gain strata and subsequently reclassified using the Reference Database for Change Calibration to improve the TCM 2018 and to detect omission and commission errors in the TCM 2015. Subsequently, a 1 pixel boundary filter has been applied in order to mitigate geometric imprecisions between the input layers 2015 and 2018, caused by the different satellite input data characteristics. Remaining change areas are filtered according to the specified Minimum Mapping Unit (MMU) of 1 ha. Finally, a manual enhancement has been performed within identified regional clusters of remaining issues, considering both: loss and gain of tree cover. The product covers the whole EEA39 area and is provided in European projection. National products might show a broken MMU due to reprojection.More information and dataset available from: Tree Cover Change Mask 2015-2018 (raster 20 m), Europe, 3-yearly — Copernicus Land Monitoring Service

This layer is a high-resolution tree canopy change-detection layer for Baltimore City, MD. It contains three tree-canopy classes for the period 2007-2015: (1) No Change; (2) Gain; and (3) Loss. It was created by extracting tree canopy from existing high-resolution land-cover maps for 2007 and 2015 and then comparing the mapped trees directly. Tree canopy that existed during both time periods was assigned to the No Change category while trees removed by development, storms, or disease were assigned to the Loss class. Trees planted during the interval were assigned to the Gain category, as were the edges of existing trees that expanded noticeably. Direct comparison was possible because both the 2007 and 2015 maps were created using object-based image analysis (OBIA) and included similar source datasets (LiDAR-derived surface models, multispectral imagery, and thematic GIS inputs). OBIA systems work by grouping pixels into meaningful objects based on their spectral and spatial properties, while taking into account boundaries imposed by existing vector datasets. Within the OBIA environment a rule-based expert system was designed to effectively mimic the process of manual image analysis by incorporating the elements of image interpretation (color/tone, texture, pattern, location, size, and shape) into the classification process. A series of morphological procedures were employed to insure that the end product is both accurate and cartographically pleasing. No accuracy assessment was conducted, but the dataset will be subjected to manual review and correction. 2006 LiDAR and 2014 LiDAR data was also used to assist in tree canopy change.

The USDA Forest Service (USFS) builds multiple versions of percent tree canopy cover data, in order to serve needs of multiple user communities. These datasets encompass CONUS, Coastal Alaska, Hawaii, U.S. Virgin Islands and Puerto Rico. There are three versions of data within the 2016 TCC Product Suite, which include: The initial model outputs referred to as the Analytical data; A masked version of the initial output referred to as Cartographic data; And a modified version built for the National Land Cover Database and referred to as NLCD data, which includes a canopy cover change dataset derived from subtraction of datasets for the nominal years of 2011 and 2016. The Analytical data are the initial model outputs generated in the production workflow. These data are best suited for users who will carry out their own detailed statistical and uncertainty analyses on the dataset and place lower priority on the visual appearance of the dataset for cartographic purposes. Datasets for the nominal years of 2011 and 2016 are available.

The Cartographic products mask the initial model outputs to improve the visual appearance of the datasets. These data are best suited for users who prioritize visual appearance of the data for cartographic and illustrative purposes. Datasets for the nominal years of 2011 and 2016 are available.

The NLCD data are the result of further processing of the masked data. The goal was to generate three coordinated components. The components are (1) a dataset for the nominal year of 2011, (2) a dataset for the nominal year of 2016, and (3) a dataset that captures the change in canopy cover between the two nominal years of 2011 and 2016. For the NLCD data, the three components meet the criterion of 2011 TCC + change in TCC = 2016 TCC. These NLCD data are best suited for users who require a coordinated three-component data stack where each pixels values meet the criterion of 2011 TCC + change in TCC = 2016 TCC. Datasets for the nominal years of 2011 and 2016 are available, as well as a dataset that captures the change (loss or gain) in canopy cover between those two nominal years of 2011 and 2016, in areas where change was identified.

These tree canopy cover data are accessible for multiple user communities, through multiple channels and platforms, as listed below: Analytical USFS Tree Canopy Cover Datasets (Download) USFS Enterprise Data Warehouse (Image Service) Cartographic USFS Tree Canopy Cover Datasets (Download) USFS Enterprise Data Warehouse (Map Service) NLCD Multi-Resolution Land Characteristics (MRLC) Consortium (Download) USFS Enterprise Data Warehouse (Image Service) The Puerto Rico and the US Virgin Islands TCC NLCD change dataset is comprised of a single layer. The pixel values range from -97 to 98 percent where negative values represent canopy loss and positive values represent canopy gain. The background is represented by the value 127 and data gaps are represented by the value 110 since this is a signed 8-bit image.This record was taken from the USDA Enterprise Data Inventory that feeds into the https://data.gov catalog. Data for this record includes the following resources: ISO-19139 metadata ArcGIS Hub Dataset ArcGIS GeoService For complete information, please visit https://data.gov.

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

Losses and gains in canopy cover of the world’s tree canopies affect carbon stocks, species habitats, water cycles, and human livelihoods. Consistent and multi-decadal global data on tree-canopy cover dynamics are needed for modelling climate scenarios, tracking progress towards restoration targets, and diverse other research, management and policy applications. However, most data only map binary ‘forest’/‘non forest’ distinctions that are regionally restricted or biassed by data gaps, and those mapping tree-canopy cover are limited to the 21st century. Here, we present an annual and global time-series of tree-canopy cover between 1992 and 2018. To develop these data, we integrated complementary products, using their respective strengths to compensate for weaknesses, and exploiting path dependencies in change processes to derive predictions into the data-sparse 1990s. Our model validation indicates we can accurately map tree-canopy cover (r2=0.95 [±0.01], RMSE=6.75% [±0.08], F1-score=0.97 [±0.0]) and our time-series agree well with national forest statistics (r2=0.94 [±0.0]).

This repository contains the Global Tree-Canopy Cover Change dataset (GTCCC), which consists of a global time-series on per-pixel tree-canopy covers estimated at a 300-m resolution between 1992 and 2018. The repository contains the following:

GTCCC_canopyDensity.tar.gz - Annual GeoTiffs on per-piel tree-canopy cover estimates (EPSG:4326)

GTCCC_uncertainty.tar.gz - Annual GeotiFFs with per-pixel estimates of the 95% confidence interval of the the predictions of each RFReg decision tree.

GTCCC_change.tar.gz - Multiple outputs describing changes in tree-canopy cover between 1992 and 2018. The contents are described in a README.txt file found within.

modelling_infrastructure.tar.gz - Infrastructure to generate the GTCCC dataset, including code, and some intermediary outputs, such as reference samples and the predictive model. The contents are described in a README.txt file found within.

The predictors used to generate the GTCCC are provided separately given large volume, and can be reached by clicking here.