A map of trees in New York City classified at the genus level from remote sensing

Summary This dataset is a classification of trees in New York City at the genus level, derived from remote sensing and a street tree database. The classification is a collection of polygons for tree crowns, stored as a geopackage file (gpkg). This dataset covers all tree canopy in the city (i.e., wall-to-wall), with canopy cover current for 2021. There are classification predictions for 1,806,811 tree crowns, with additional crowns inserted from reference street tree database data (for 1,821,184 total labeled) into the source tree crown polygon dataset (2,127,625 total crowns, the difference being unclassified due to incomplete data). The overall accuracy is 81.5% (kappa = 0.788) In the Genus_Predicted column, each crown is classified as one of 18 common, dominant deciduous tree genera present in the city, based on Nowak et al. (2018) and Treglia et al. (2021). There are many more (hundreds) of tree species in the city, and this is restricted to the genus level for the most common genera because of limitations in the separability with the XGBoost classifier. Included genera are: Acer (maple) Ailanthus (tree of heaven) Betula (birch) Fraxinus (ash) Ginkgo (ginkgo) Gleditsia (honeylocust) Liquidambar (sweetgum) Liriodendron (tuliptree) Malus (apple) Platanus (London planetree and sycamore) Prunus (cherry) Pyrus (pear) Quercus (oak) Robinia (black locust) Styphnolobium (pagoda tree) Tilia (linden) Ulmus (elm) Zelkova (zelkova) However, where a single valid street tree database point intersects with a tree crown object, we include this reference value in the Genus_Ref column (as well as Species_Ref, FullCultivar_Ref, and other information from the street tree database). Where available, we merge the genus label from the street tree database into the labeled classification, superseding any predicted classification label; this is stored in the Genus_Merged column. This includes many more genera than we were able to map with the classifier. In normal usage, we recommend using the Genus_Merged column. This dataset is derived from: NYC Parks & Recreation Forestry Tree Points street tree database, which are surveyed by expert staff at Parks, the Natural Areas Conservancy of NYC, and citizen scientists (and validated by expert staff). This is the reference database for types of trees as geospatial points, but is limited to trees along streets and in some parks. It includes genus, species, cultivar, diameter at breast health, tree crown and health quality flags, and additional ancillary information. This dataset is available at: https://data.cityofnewyork.us/Environment/Forestry-Tree-Points/hn5i-inap/about_data Tree crown polygons for 2021 from the Nature Conservancy New York Cities Program and the University of Vermont Spatial Analysis Lab. It was made with airborne lidar and high resolution aerial imagery. This dataset is available at: https://zenodo.org/records/14053441 (The Nature Conservancy, 2024). Airborne lidar collected in 2021 across all of NYC. This data was used to create the tree crown polygon data (by TNC and UVM), and it was also used to develop tree crown metrics related to tree structure. This dataset is available at: https://gis.ny.gov/lidar Satellite imagery time series from PlanetScope satellite imagery (Planet Labs, San Francisco, California; https://docs.planet.com/data/imagery/planetscope/). This imagery is gridded at 3 m spatial resolution in the visible and near infrared, with observations from a constellation of satellites acquiring imagery daily or more frequently. However, actual return interval is often several days due to clouds and other data quality filtering. Main geopackage file: nyc_class_tree_genus_polygons_v1_1.gpkg In addition to the main geopackage file, ancillary data files include: Accuracy table (confusion matrix) with separate validation tree crowns not included in the classification training: nyc_class_tree_genus_polygons_v1_1_confmat.xlsx Column description file for classification attribute table. Lidar crown metrics were largely based on selection of metrics used in Alonzo et al. (2014): nyc_class_tree_genus_polygons_v1_1_attribute_description.xlsx Geospatial projection information New York State Plan Long Island Zone (US survey foot) coordinate reference system, EPSG 2263 (The Nature Conservancy 2024). Funding information This work is supported by the McIntire-Stennis, project award number 7005302, from the U.S. Department of Agriculture’s National Institute of Food and Agriculture. This work was supported in part by the Cornell Atkinson Center for Sustainability.