The US Tree Map
The quest to map every tree in the contiguous United States
Who doesn’t love trees? They are benign beasts that bring beauty to our lives. Some grow to dizzying heights, eke out a living in the most inhospitable of places, or live to be thousands of years old. Others simply shade us from harsh weather, light up in brilliant colors each autumn, and provide habitat to wildlife. When grown in large numbers, they not only support the largest timber industries globally, they can also be effective carbon-sequesters, climate-regulators, water-filtrators, and ground-stabilizers. Undoubtedly, our nation’s forests are an American heritage.
Trees say a lot about us as a society. They are managed, cultivated, and cut down as per our demand for wood resources. They have been known to influence real estate values, or reveal social and racial inequities that plague our cities. Urban neighborhoods are getting greener as a result of an increasing value in green spaces, and to many, planting more trees is a solution to the worsening climate situation. The point is: trees deeply matter to us.
Why map them [again]?
It’s been done before. Like all natural resources, mapping the distribution of trees is a tool to better manage our forests and study changes over time. Even dated maps such as the one below have their value in understanding long-term changes to forest cover.
In the era of remote sensing and machine-learning, mapping trees has become quicker, more accurate, and can be done with greater frequency. Take the National Land Cover Database (NLCD) for instance, which relies on 30-meter Landsat imagery to produce a tree canopy density map across the nation. With seven epochs of this dataset dating back to 2001, the NLCD is the longest-running land cover database available here in the U.S.
While the NLCD might suffice to observe spatial-temporal trends over larger tracts of forest, it lacks the finesse required for high-resolution, large-scale analyses. What if we want to observe the individual trees themselves rather than proxies like “density” or “probability,” yet maintain the sheer spatial and temporal scale that the NLCD succeeds in? Cue the US Tree Map.
How did we get here?
Here at EarthDefine, high-resolution and scalable geospatial data is our bread and butter. Through the use of machine-learning and deep-learning artificial neural networks, we have transformed remotely-observed spectral and LiDAR data into consumable land cover information. Hot on the heels of our own U.S. Buildings Footprints dataset, we leveraged similar Convolutional Neural Networks (CNN) to “learn” to identify tree canopy in high-resolution, multi-spectral aerial imagery.
But first, it’s important to give credit to where it’s due. While we work day in and day out to advance and aggregate high-resolution geospatial information such as trees, building footprints, or land cover, we are only able to do so because of freely available remote sensing products like NAIP imagery. The National Agriculture Imagery Program is a treasure trove of freely available high-res ortho-imagery, and it has been the data source “backbone” of much of our work. Open access to NAIP imagery has fostered innovation in a variety of sectors, our tree canopy included. We are among those in the private and public sectors that hope to see NAIP remain free and open in light of recent proposals to license these data. I digress…
It only took shy of one-million samples — snapshots of LiDAR and NAIP derived tree canopy — to generate tree canopy at 1-meter resolution across the conterminous United States. Be it a grove of Quaking Aspens in the Colorado Rockies, a weathered Juniper in the Nevada Great Basin, a stand of mangroves in coastal Florida, or an American Elm in the National Mall of our capitol — we meticulously sampled the natural variety of our country’s trees and forests. The result is a robust and stable model capable of classifying trees from NAIP imagery, regardless of the year, month, region, etc. Because the U.S. Department of Agriculture contracts the acquisition of NAIP imagery on average every two years, we will continually update the US Tree Map as newer imagery becomes available (not to mention the back-catalog of imagery that can be used for change-analyses).
Every tree counts
So, we mapped every tree across the Lower-48. All 555-million acres of them. Let’s take a look.
Pretty neat, huh? Let’s look a little closer:
How about California? Given the frequency and severity of California’s wildfires in the several years (as I write this, I can look outside my window and notice the warm haze of smoke that has drifted in from ongoing fires in Northern California — a full three states away from my home), it’s appropriate that we should have a reliable and regularly-updated map of California's trees (estimate fuel-loads, assess defensible spaces around homes, or track the magnitude of forest lost to fires). We can literally see the scars from fires of years past, etched into otherwise densely forested regions. Take, for example, the Biscuit Fire Complex that burned nearly one-half million acres of forestlands in California and Oregon in 2002 (see below).
Viewing the trees at such scales can be pretty eye-opening to the effects of natural and human-caused phenomena on the landscape. The river bottoms of the American Southeast are such an example, where rich alluvial soils are a boon for agriculture. The conspicuously deforested lands around the lower Mississippi River delineate the historical floodplain nearly spot on. However, tree canopy data goes beyond illustrating such well-understood quirks about our landscape and land-use practices. Accurately mapping trees at this scale also can be used to monitor trends in forest cover, determine where trees could be replanted to bolster erosion control, or to identify candidate properties for conservation easements.
How accurate is it?
Great question! We estimate our nationwide tree canopy to be 96.6% accurate (hey urban planners! That figure is 97.3% for TIGER designated urban areas). We randomly sampled approximately 48,000 points, and with the help of very dedicated staff, we manually decided which points landed on trees, and which did not. This way we can evaluate how well our CNN model predicted trees from the imagery by comparing the results to our own predictions.
Putting data to work
While these trees are sure nice to look at, this data is meant to work! Whether you are a data scientist, geographer, urban planner, natural resource manager, insurance analyst or cartographer — however trees or forests matter to your work, accurate and reliable data is now available. The US Tree Map accurately maps tree canopy data with 1-meter resolution for the continental United States, making it the largest archive of high-resolution tree canopy data available.
For us at EarthDefine, this isn’t the end of the road. While we are pretty happy about our new tree canopy data, we are already working to improve it. Trees and forests change in many ways and take on a multitude of appearances in aerial imagery. We want to ensure that we can accurately map all such changes— things like bark beetle infestations, wildfires, regenerating trees from recent disturbances, etc. The same goes for the variety of NAIP image-specific variations, like shadows in mountainous regions, leaf-off deciduous trees, or fall foliage. Our goal continues to be to provide the most reliable and accurate data, so end-users can do their work with confidence.
Over the next few weeks, we’ll explore how these data can be leveraged for scalable solutions and analyses in a series of briefs. We’ll dive into as many different applications as we can to show the benefits of good data. For those interested in exploring the data themselves, downloadable samples are available on our website. Stay tuned.
