The Download: Feature Articles
Brooklyn Like You've Never Seen It Before
By Victoria Lubas | November 13, 2020
Debra Laefer Leads NYU’s LiDAR Effort to Map Sunset Park
From finding lost cities to bringing the auto industry closer to the ever-elusive self-driving car, the potential applications of Light Detection and Ranging (LiDAR) seem straight out of Indiana Jones and The Jetsons. Recently, researchers from the NYU Center for Urban Science and Progress (CUSP) used aerial LiDAR technology to create highly-detailed scans of the Sunset Park neighborhood in Brooklyn.
The lead researcher on the project, NYU CUSP professor Debra Laefer, previously led a LiDAR project at the University College Dublin which documented Dublin’s streets and buildings ahead of the expected construction of the city's first metro system. TechCrunch reported that when the 2015 Dublin data was released it was the densest-ever at 335 points per square meter across 1.5 km² of the city’s center.
Laefer and her students at NYU CUSP applied her optimized flightpath strategy to create high-density scans of Sunset Park, Brooklyn including unprecedented vertical facade capture. Using refined techniques and high-end equipment, these LiDAR scans provide datasets that are applicable to a wide-range of fields and valuable to many organizations and associations in New York, from mapping flood risk to tracking sunlight and identifying the windiest city streets.
How LiDAR Works
In the most simplified sense, scanning an area with aerial LiDAR involves flying above an area and emitting eye-safe lasers that are pointed earthbound. The resulting dataset is a highly-detailed geometric representation of the built and natural environment. As Laefer explains it, the laser scanning equipment she employs “uses knowledge of the speed of light, high-end positional data of the helicopter, and the time delay between the emitted and received signal to position the resulting points to be able to generate the geometry of the surroundings.”
The resulting point cloud can be used to create a detailed reconstruction of the scanned area, which can shed light on phenomena, such as which curb heights are in accordance with Americans with Disability Act (ADA) requirements, or the effects of idling vehicles on an area's pollution.
LiDAR technology has been developing for decades and has both improved and become more affordable with each advancement. According to Laefer, it's being applied across the United States, and a national scan is already 70 percent complete. Areas of the United States with large, open plains can be scanned quickly and at a lower density, but to fully capture the intricacies of urban architecture—such as that found in New York—is best done with more fly-overs and at slower flight speed and lower altitudes.
Such high-density scans of Dublin and Sunset Park were made possible not by a change in the technology used, but in the way the technology is deployed in terms of the flight patterns. Aerial LiDAR scans are conducted with a scanning device affixed to the underside of an aircraft. While the density of a scan increases with every additional fly-over and when the scanning aircraft flies at a lower speed, Laefer and her team have further optimized the flight planning to create unprecedented completeness in the data capture using the orientation of the flight path and the amount of overlap to minimize occlusions (i.e. holes in the dataset). High-density scans, like those of Dublin and New York, are conducted 300 meters (about 1,000 feet) in the air because at that height it maximizes the resolution while still being safe to fly the helicopter.
The Difference with New York
To capture the vertical details of a city, LiDAR scanning must be conducted as low to the ground as safely possible so that sensors can capture the facades of buildings in addition to the ground and rooftops. Capturing vertical space is much more difficult than horizontal space – for every 10 points that the CUSP LiDAR team captures horizontally, only about 1 point is captured on a vertical surface. In terms of resolution, for her current scan this translates to 56 points per square meter vertically and 560 points per square meter horizontally. These scans are detailed enough to be able to tell the height of individual curbs, but capturing this much detail does mean a higher cost per square kilometer.
Laefer said that because of a natural sense of uncertainty for those unfamiliar with high density urban scans, getting municipal agreement to the first scan can be difficult. After much discussion with New York City officials, the team was granted permission to scan the Sunset Park neighborhood. This turned out to be a wonderful location because Sunset Park consists of a regular grid, relatively wide streets, and only medium-height buildings, which can be a recipe for great scans. Laefer explained that “a city like Washington DC, which has a height limit for buildings that they all pretty much reach, would also be easy to scan. When there is a mix of skyscrapers and low-rise buildings, things become more complicated because the minimum flight height is further restricted.”
When capturing LiDAR data in Brooklyn, the CUSP team worked with Virginia-based specialty contractor Tuck Mapping, a highly-esteemed mapping services company known for their excellence with LiDAR and urban flying. The CUSP team and Tuck Mapping agreed that based on the helicopter they would use and Sunset Park’s urban conditions they would have to fly slightly faster when scanning New York than they did in Dublin.
Flying faster means capturing slightly less data with each flyover, but the amazing amount of overlap in the New York City scan counteracted that problem. Laefer explained that typically, when LiDAR scanning, one would plan to have at least 10% overlap between scans, but higher overlap is always better because that means less dead spots at the end. The Dublin project had 67% overlap, but the New York City project had 77% overlap, resulting in a very complete scan.
When flying over a city with LiDAR, pilots tend to fly parallel to the street grid because it is easier to fly when lining up with the street as a guide. However, the team demonstrated in their Dublin scan that by rotating the pattern to be diagonal with the street grid they would be able to get greater coverage, even though this is a harder path for pilots to follow. To demonstrate definitively the improved coverage, they flew over Sunset Park twice: once parallel to the street grid and once diagonally. The combined data exceeds 560 points per square meter and is available for public download. The team’s previous release was only 335 points per square meter.
The Applications of this Data
The data gathered through NYU CUSP’s LiDAR is so specific and detailed that it is applicable to a wide variety of uses and fields. Previously, New York City had tried capturing data using Cyclomedia data, which employs laser scanning from a van but this produced highly uneven results. Gathering data from street-level meant the interference of vans, signage, scaffolds and other factors that blocked complete documentation of the visible built environment. Flying to collect data from overhead vastly minimizes that issue.
Also by flying over an area more than once, transient objects (e.g. cars, people, etc.) can more easily be filtered out. However, while CUSP’s LiDAR is very detailed, Laefer explained that since it is gathered “at a great height it’s not good enough to record license plate numbers or someone’s face. When that day comes it will necessitate a new set of privacy requirements for this type of data collection."
Laefer considers it “exciting to see how this kind of scan is shaping hearts and minds in the industry.” One example of the impact LiDAR data can have on urban planning is in the realm of alternative energy. Laefer explained that when it comes to capturing solar energy through embedded photovoltaics in building windows, this resolution of LiDAR data could greatly assist through the production of both shadow maps and solar potential maps to pinpoint the areas that receive the most sun and plan accordingly. These sunlight calculations could also play a role in real estate, identifying properties that receive the most sun at different times of day.
While city models have been generated in the past for application in Hollywood productions and video games, few are realistic and most use some form of generic library of objects that ultimately depict every sidewalk as being the same. In contrast the LiDAR data shows complex architectural details and can even distinguish every strip of sidewalk to show changes in width, uneven sections, as well as utility lines, lampposts, sewer grates and outdoor cafe seating. This detail and specificity means it can be used in surveying-based applications for fields like architecture and engineering and inform on the accessibility of a neighborhood's individual streets and corners.
The methods of collecting LiDAR data continue to advance, and Laefer can see a future where this quality and quantity of data collection may be gathered with a drone instead of a helicopter. Drone flight restrictions based on privacy and safety concerns currently prohibit that in urban environments in the US.
Laefer plans to stay in New York City and hopes to continue scanning other areas of the city once the Sunset Park project is complete. A possible future project may focus on coastal areas in New York and New Jersey and use that data to assess flood risk. The Sunset Park data will be used to pilot that idea through a National Science Foundation grant where the data will be used as a direct input into hydrological flooding models. In the midst of the COVID-19 pandemic, Laefer’s team is also looking at how LiDAR data may be useful to evaluate surfaces as vectors for coronavirus transmission.
- 2019 LiDAR Data Collection for Sunset Park, Brooklyn, NY (NYU Faculty Digital Archive)
- NYU Data Services