A species map of 1.8 million New York trees moves canopy policy past counting

Researchers have published an interactive map identifying 1.8 million individual trees in New York City by type, using satellite imagery, ground data sets and lidar, with an overall accuracy of 82% ^[1][2][3]. It is the first "wall-to-wall" citywide map of its kind, and it covers trees on private property, in natural areas and in other unsurveyed places that together make up an estimated 65% of the city's canopy and were absent from previous maps ^[4][5].

That gap is the whole point. Senior author Daniel Katz, an assistant professor in Cornell's School of Integrative Plant Science, said the map adds 1.4 million trees beyond the street trees that had already been surveyed ^[6][7]. By subtraction, the pre-existing street tree survey accounted for roughly 400,000 trees, meaning the city's operational picture of its own forest covered about a fifth of the individual trees now identified ^[8].

The technical unlock was timing rather than resolution. The team, with first author David Miller, used the volume of images from Planet's PlanetScope satellite constellations, which circle Earth every 90 minutes, to watch leaves turn in autumn and buds break in spring, using that seasonal signature to separate species ^[9][10][11]. Katz called satellite-based tree identification a "holy grail" in his field for decades, held back because the images were hard to access and not at the scale identification requires ^[12]. Training and verification came from New York State lidar, a crown-delineation map from The Nature Conservancy and the University of Vermont, and on-the-ground data from the NYC Department of Parks and Recreation ^[13]. The study appeared in Scientific Data ^[14].

The policy relevance is immediate. New York recently set a goal of raising tree cover from 22% to 30% by 2040, an eight-point increase and a 36% relative expansion of existing canopy ^[15][16]. A count-based target says plant more; a species-level inventory makes it possible to ask which plantings actually reduce temperature. The map is part of the Cool Trees project, led by Dr. Arnab Ghosh, associate professor of medicine at Weill Cornell Medicine, which is working on the relationship between tree density and type, temperature, and heat-related health risk as heat waves in the city become more frequent and intense ^[17][18][19]. Katz described the goal as being able to say that a given planting or management decision will change temperature by some amount and change the number of people arriving at emergency rooms during a heat wave ^[20].

Two honest limits. The map does not yet say which trees cool best; the team says it will use the data to work that out, alongside applications from allergen reduction to invasive pest management ^[21]. And 82% overall accuracy, which is higher for common New York species and therefore lower for uncommon ones, implies on the order of 324,000 trees carrying the wrong label if error were spread evenly ^[3][22]. Katz himself said the approach "isn't perfect, but it's a really good start at showing you what's where" ^[23].

What to watch: whether the species-to-temperature-to-hospitalisation chain gets quantified with enough confidence for a parks department to write it into planting specifications, and whether the method transfers. The team has compiled a database of 405 medium and large U.S. cities it hopes to map with the same ubiquitous imagery ^[24]. A separate Cornell project, "The Generative Canopy," is using street-tree data to identify where shade is most needed for vulnerable residents ^[25]. If the accuracy holds outside the city that supplied the training data, the cost of knowing what a municipal forest contains drops sharply.