Open data.
Real places.
Know what you’re looking at.
The model combines public building records, surveyed terrain, a land boundary and an older 3D building survey. Each layer keeps its own date and limitations.
Different dates.
- Building snapshots
- Sep 21 / Oct 3, 2026
- Roof model survey
- 2014
- Terrain survey
- 2017
- Land boundary
- DCP 26b
- Street mapping
- 2022 imagery
- Street trees
- 2015 census
Follow the sources.
Buildings
NYC OTI footprints and recorded ground and roof elevations supply 142,178 unique building records across Staten Island. The original North Shore snapshot contains 4,485 buildings. The island expansion adds 137,693 unique IDs, plus seven clipped continuations of North Shore buildings. Invalid heights and non-building features are excluded and counted in the manifests.
Dataset 5zhs-2jue / North Shore snapshot Sep 21, 2026 / island acquired Oct 3, 2026
Roof geometry
The NYC 3D building model uses a 2014 survey. Matching checks accepted 129,307 models across Staten Island: 28,086 stepped and 101,221 flat. These include 4,191 original North Shore models and 125,116 island expansion models. The remaining 12,871 buildings use recorded-height extrusions.
Models preserve the surveyed roof and wall surfaces and load as you move closer. Matches require a footprint intersection-over-union of at least 60% and a roof-elevation difference no greater than 10 m. Missing, changed or ambiguous matches retain their current footprint extrusions. Flat roofs are valid survey results; pitched roofs are not invented.
DA16 + DA17 / NAVD88 / source survey 2014
Terrain & shoreline
The 2017 NYC/NOAA topobathymetric elevation model supplies ground heights. The original North Shore browser terrain samples every 2 m; its print terrain uses a shared 1 m grid. The island expansion uses an 80 m overview and 20 m terrain near the camera. NYC DCP’s water-excluded 26b borough boundary separates land and water.
North Shore piers and narrow shoreline gaps use the nearest surveyed land height. These 27,251 estimated 1 m² cells can be highlighted in the explorer. The island expansion has 986 estimated coastal samples, with surveyed elevations for about 99.74% of land samples. Estimated heights do not establish measured deck elevations.
Streets, parks & waterfront
NYC OTI planimetric mapping covers roadbeds, sidewalks, plazas, parking lots, medians, parks, curbs, rail lines, shoreline and transportation and waterfront structures across Staten Island. The expansion includes 57,690 mapped records in 292 chunks, alongside 2,593 records in the original 32 North Shore tiles. Features retain source IDs, with geometry clipped at the North Shore boundary to avoid overlap.
These layers use Staten Island imagery dated March 16, 2022. Roads, parks and waterfront surfaces appear in the overview; nearby chunks add the finer layers. Island roads, sidewalks and parks are split along the displayed terrain triangles so they follow its slopes. Bridge, pier and other structure deck elevations are not supplied by this mapping.
Historical street trees
101,394 tree positions come from the 2015 NYC Street Tree Census: 3,109 in the original North Shore data and 98,285 across the island expansion. The model retains census IDs, species and trunk diameter. Crown shape, radius and height are illustrative; current survival is unverified. This street-tree census does not include every tree in parks and forests.
Mapped roof equipment
6,167 cooling-tower and water-tank footprints match existing building footprints: 292 in the North Shore data and 5,875 in the island expansion. Their outlines can be enabled over buildings near the camera. They carry source record IDs; equipment height is not inferred. These layers are published in the 2022 collection, with equipment originally captured from 2014 imagery.
Bridges & connecting expressways
Bayonne Bridge, the current twin Goethals Bridge and Outerbridge Crossing follow mapped OpenStreetMap alignments. The connecting Staten Island, West Shore and Dr. Martin Luther King Jr. expressways, Korean War Veterans Parkway and interchange ramps retain their source way and node identifiers.
The models distinguish Bayonne's raised roadway and steel arch, Goethals' cable-stayed decks and Outerbridge's cantilever trusses. Principal spans and clearances are referenced to Port Authority publications. Structural members, road widths where unmapped, deck elevations and interchange grades are illustrative. Published clearance above water is not a surveyed NAVD88 roadway elevation. Ground-level roads follow the terrain visible in the model as detail loads.
Nine orange ferries
The fleet reuses the nine existing named vessel reconstructions, spanning the Ollis, Molinari, Barberi and Austen classes. Orange boats travel continuously between St. George and Whitehall, with randomized vessel handoffs at the Staten Island berths. All nine models remain in the scene.
Crossings are accelerated to about 2½ minutes for the animation. Harbor waypoints, berths and vessel assignments are illustrative. The animation is independent of scheduled departure times and is not live vessel tracking or a current service roster.
The departure panel uses NYC DOT’s published weekday, weekend and named-holiday timetable, checked October 2, 2026, in New York time. DOT lists an approximately 25-minute crossing. This saved timetable does not include service alerts or temporary schedule changes; check DOT before traveling.
NYC Ferry · live vessel positions
Catamarans follow the NYC Ferry GTFS-realtime vehicle-positions feed (operated by Hornblower for NYCEDC), read through this site’s cached relay every 15 seconds without an API key. Each vessel is placed from its reported latitude, longitude and bearing, projected into the model’s NAD83 New York Long Island frame; motion between reports is eased, not predicted. Vessels beyond the drawn harbor (the Rockaways, Soundview) are counted but not placed. The St. George landing position comes from the NYC Ferry GTFS stop list.
Hull, cabin and colors are illustrative. When no boat is reporting, an illustrative catamaran rests at the landing. Stale or unavailable feeds are identified in Places and the live boats are hidden. Nothing in this layer changes the printable model.
Harbor ships · AIS
Vessel positions come from AIS transponders received by aisstream.io’s network, collected by a small service on this project’s own server (tools/ais-collector) that keeps one websocket open for the New York harbor and serves the latest fix per vessel. The site’s relay reads that snapshot with a shared token every 10 seconds and passes on only position, course, speed, heading, name, type and length. Vessels unheard for ten minutes are dropped.
Hulls are simple shapes scaled to the reported length and colored by vessel class; they are illustrative. Staten Island Ferry boats are matched by transponder name to the nine reconstructed models, which then follow the live fixes instead of the timetable; NYC Ferry boats already drawn from their own feed are not drawn twice. AIS coverage is volunteer-received and has no service guarantee, so gaps are identified in Places. Nothing in this layer changes the printable model.
MTA buses · live vehicle positions
Buses follow the MTA Bus Time GTFS-realtime vehicle-positions feed, read through this site’s server relay every 20 seconds. The relay holds the developer key, keeps only vehicles on Staten Island and returns their positions, bearings and route numbers; the key never reaches the browser. Vehicles are projected into the model’s NAD83 New York Long Island frame and eased between reports. Only buses within the North Shore corridor are drawn.
Bodies are illustrative 12 m buses in the blue-and-white livery. Stale or unavailable feeds are identified in Places and the buses are hidden. Nothing in this layer changes the printable model.
Staten Island Railway · live trip updates
Modeled trains follow the MTA SIR GTFS-RT feed, obtained from MTA through this site’s cached relay and checked every 30 seconds without an API key. The railway alignment and stops come from the MTA subway GTFS snapshot retrieved October 2, 2026. The browser estimates movement between reported station events and predicted arrivals; the feed does not establish a continuous GPS location.
Car bodies, windows and doors are illustrative. Trains are drawn above the terrain, including tunnel sections, so their progress can be seen in the North Shore view. Late or unavailable feeds are identified in Places; stale trains are hidden. Neither the live layer nor its surface elevations change the printable model.
Zoning & waterfront proposal
Zoning is a dated North Shore reference layer. The zoning snapshot and waterfront proposal remain local to the North Shore. Proposal massing comes from the supplied North Shore Waterfront Plan Environmental Assessment Statement figures, CEQR 26DME018R, dated August 26, 2026. Building forms follow EDC's Sept 29 2026 scoping presentation (slides 30–33, 35), traced by hand. Plan registration, intermediate dimensions and the proposal boundary remain approximate.
Tax lots
Selecting a building adds its tax lot from the NYC Department of City Planning’s PLUTO release 26v2 (August 2026): address, land use, building class, floor area, units, floor area ratio, year built and altered, zoning districts on the lot, landmark status, FEMA floodplain flags, assessed values and civic districts. The Land use building colour and the year-built fallback use the same records. Buildings join to lots by BBL, so several footprints can share one record and condominium units appear as a billing lot. Owner names are left out of the browser payload. All 125,829 Staten Island lots are kept in 365 shards that load by tax block as needed.
Demographics
Neighborhood stories and building cards carry a line of figures from the 2019–2023 American Community Survey 5-year estimates: people, median age, foreign-born share, commute by public transit and owner-occupied share. The Census publishes these by tract, not by neighborhood, so each tract’s counts are split across the viewer’s outlines by the share of its land inside each one. That assumes people are spread evenly across a tract, which is false where housing sits beside parks, water or industry; small neighborhoods cut from large tracts carry wide margins, and the strip shows the population margin at 90% confidence. Medians are population-weighted means of tract medians. A share is left out when fewer than 100 people or households stand behind it. The three shores use the Census PUMAs directly, which follow Community Districts 1, 2 and 3. Neighborhood outlines are unofficial and do not define Census places.
Zoning headroom
The Zoning headroom building colour and the Show unused allowance layer compare what each tax lot has built with what its zoning allows, using PLUTO’s built floor area ratio and its maximum residential, commercial and community facility FAR. Remaining floor area is the allowed FAR minus the built FAR, times the lot area. By default the allowance is the higher of residential and commercial; the Builder can measure against residential only, or against any use including community facility, which often gives a much larger figure. Buildings are coloured by the share of the allowance still unused; a lot built above its allowance is tinted clay, and a lot whose allowed FAR is zero in every column (parks, utilities, some special districts) is grey rather than “nothing allowed”.
Ghost volumes rise from each recorded roof by the lot’s remaining floor area divided by the footprints drawn on that lot, in storeys of 3.2 m, so buildings sharing a lot rise by the same number of storeys. Parking and vacant lots hold most of the headroom and have little or no footprint to draw on; their figure appears in the panel only. Every figure is an upper bound: FAR is the simplest zoning control, and height limits, sky exposure planes, yards, lot coverage and parking can stop a lot short of it. Built floor area is the Department of Finance figure and may not match the footprint and height drawn here. Lots split by a district boundary report only the district with the larger share. Condominium units resolve to the billing lot, which carries the whole building.
Explore below the water.
The Water control lowers a horizontal plane through a browser-only seabed model around Staten Island. Exposed seabed has a subdued brown material. Dashed mauve lines identify mapped coverage edges; empty areas beyond those edges or within gaps are unknown, not evidence of a land connection.
We first inspected the original 2017 NYC topobathymetric DEM, dataset 9307. It has valid coastal underwater returns but large offshore gaps, so the word “topobathymetric” does not establish complete harbor coverage. Additional data were identified through NOAA’s Bathymetric Data Viewer.
The seabed uses NOAA CUDEM dataset 8483, with 1/9 arc-second output spacing (roughly 3 m), from the local 2015 and 2018 editions. These are combined elevation models with interpolated areas and mixed survey dates, not a new uniform seabed survey. The 2018 New Jersey report includes hydrographic inputs dated 1933–2015 and channel-condition surveys dated 2015–2017; bathymetry was gridded at about 10 m before resampling and filtering. Source dates cannot be assigned to every local cell from the available metadata.
Western gaps in CUDEM are supplemented by USGS CoNED New Jersey/Delaware (dataset 5040, 2015 compilation). Its 1 m output grid incorporates local NJNY bathymetry at 10 m resolution, dated 1930–2013. This is an older combined surface, not a contemporary channel survey.
Elevations are verified as NAVD88 metres before conversion into the explorer’s existing local coordinates. The original NYC heights use NAVD88 US survey feet, converted with 1200/3937 metres per foot. NOAA’s combined models already include their producers’ vertical-datum transformations. Tide-referenced survey depths such as MLLW are not inserted as though they were NAVD88 heights.
The slider’s lower limit is computed slightly below the deepest valid source elevation in the implemented area, excluding NoData and invalid outliers. It reflects available mapped coverage, not the deepest point in the entire harbor. Browser detail is coarser than the source rasters, and shoreline seam vertices meet the existing land mesh. Refer to the downloadable manifest for the precise area, source versions, processing, deepest sample and missing coverage.
This is an exploratory elevation view. Lowering the plane does not simulate drainage, trapped pools or historical coastlines. Interpolation, shoreline generalization, different source dates and dredging limit conclusions about a physical connection to New Jersey. Boats, wakes and unmasked reflections are hidden at negative levels. The reference view and downloadable printable models are unchanged.
Print revision 01.
Print files contain terrain and integral buildings with flat roof caps. Browser roof details, proposal geometry, sun, water and vertical exaggeration do not change these files.
The print reports count 4,753 building occurrences across tiles. Those are the same 4,485 unique North Shore building records used by the browser: a building crossing a tile seam is counted in each print tile it intersects. The browser assigns each record once, using a point inside its footprint. These definitions were checked against the generators and all tile building audits.
All 41 source STL hashes and mesh bounds are checked against their digital validation reports. Physical fit remains untested. Required build volume with supports or brim awaits a named slicer profile. Public file availability and distribution terms are recorded separately.
One geographic frame.
Local coordinates use metres, with X pointing east and Y pointing north. The renderer displays north along its negative Z axis. Heights use the NAVD88 vertical datum.
- Working origin
- 292600 m east / 50500 m north
- North Shore print extent
- 1600 × 3200 m
- Island browser bounds
- East −14400 to 4000 m / north −14400 to 3200 m
- Physical height rule
- Z(mm) = 8 + 0.4 × NAVD88 elevation(m)
- Physical scale
- 1:2,500 horizontally and vertically
The water plane at NAVD88 0 m is a chosen display level. Water sliders and shadows support exploration; they are not site-specific flood predictions or a certified environmental analysis.
Read the gaps, too.
Records and surveys describe different moments in time. Some recorded roofs sit below the older terrain; the original heights are retained and flagged. A roof model match is a geometric check, not confirmation of a building’s current condition.
The print geometry, 2014 roof models and proposal envelopes have different purposes. The browser keeps them switchable so you can inspect those differences. Send a correction to Mike.
Recognizable St. George.
The illustrated terminal, Empire Outlets, Borough Hall and Lighthouse Point details follow the existing building footprints, 2024 aerial imagery and public architectural references. Clocks, galleries, railings, facade accents and signs are simplified interpretations.
The bus deck has an open lower parking and bicycle level. Its columns, clearance, bicycle racks and circulation layout are illustrative, based on local context and the requested lower-level access; they are not a measured interior plan. The terminal lot is distinct from the inland courthouse garage. Future Lighthouse Point phases are not added.
The Bay Street viaduct, upper bus loop, lower parking entrance and separate angled parking exit follow OpenStreetMap contributors (ODbL), checked against 2024 aerial imagery. Bridge-layer tags distinguish levels; ramp heights, widths and supports are illustrative and blend to the scene surfaces.
Landmark sources and modeling notes, including NYC DCAS, NYCEDC, SHoP, the Landmarks Preservation Commission and the National Lighthouse Museum.
Live tide and current.
The Water panel follows observed water height from NOAA station 8518750, The Battery, using metres relative to NAVD88, the same vertical datum as the terrain. This single harbor reference is applied to a flat plane; tide height and timing can differ around Staten Island.
Current speed and direction come from NOAA station n06010, Kill Van Kull LB 14, bin 4 at a nominal 6.55 m below the surface (metadata checked October 8, 2026). Direction is where the water flows, in degrees clockwise from true north. A localized band of pixel wakes makes the measured direction visible near the meter. Stronger readings produce faster illustrative drift; pixel size, travel speed and trail length are exaggerated, and do not track real water parcels. The reading is from below the surface, rendered at the surface for legibility. This is not a surface-current map or a harbor-wide simulation. Reduced-motion settings retain stationary tapered trails; stale readings hide the effect.
Both observations use the NOAA CO-OPS Data API, polled once a minute while the page is visible. The source observations normally have a six-minute interval. Measurement times appear in New York time. Live water measurements are usually preliminary; the NOAA quality flags are checked before use. Water older than 18 minutes and current older than 30 minutes are labelled stale. Connection failures retain the last displayed height with a warning and hide the pixel wakes. Predictions never stand in for measured data.
Moving the level slider, selecting a preset or entering a Water tour switches height to manual. Enable live tide to resume. Shared manual scenes preserve their level; live scenes fetch the latest observation. Terrain and print files are unchanged. This exploratory display is not a flood forecast or a navigation instrument.
Boats, vessel wakes, floating landing details and the illustrated current flow are hidden below −0.783 m NAVD88 and restored at or above that level. This display cutoff uses The Battery mean low water (MLW), NOAA epoch 1983–2001: MLW 1.065 m minus NAVD88 1.848 m, both relative to the same station datum. It applies to live and manual levels. Local tides differ; this is a visibility rule, not a harbor-wide navigability assessment.
Flood and sea level.
Presets are approximate and unverified. Sandy and MHHW from NOAA Battery gauge 8518750; BFE from FEMA FIRM; SLR adders from NPCC 2019. Verify before relying on them.
Across the harbor.
The distant skyline includes 12,382 simplified buildings: 3,377 in Manhattan, 930 in Brooklyn, 42 on Governors Island, 4,294 in Jersey City and 3,739 around Bayonne and Kill van Kull. NYC OTI footprints and heights and NJDEP / NJGIN footprints were fetched October 3, 2026.
NYC filtering retains waterfront buildings at least 60 feet high, inland buildings at least 120 feet high, and Governors Island buildings at least 40 feet high. New Jersey records have no building heights: footprints within 400 metres of the shoreline use an explicitly estimated 12-metre height. They show the shoreline fabric; they cannot establish the actual Jersey City tower profile.
Neighboring land uses the NOAA 2017 NYC DEM and New Jersey 10ft DEM, with the NYC water-excluded borough boundary and NJDEP coastal municipality polygons. The New Jersey elevation mosaic combines regional surveys; Hudson County uses the 2014 post-Sandy collection. Heights are converted to NAVD88 metres. Terrain uses 160-metre cells near the harbor and 640-metre cells farther inland in New Jersey; it does not establish parcel-level elevations. Buildings with estimated New Jersey bases follow this ground. Land remains visible when the skyline is hidden, and low land responds to the selected water level.
Harbor terrain sources, dates, coverage and checksums.
The Verrazzano-Narrows Bridge and Statue of Liberty are illustrative geometry. Bridge tower locations follow mapped pylons; dimensions are simplified. Roofs, spires, lighting and reflections are visual context, not a validated sightline study. Automatic visibility follows the camera; Show and Hide are available in Layers. Context is hidden in Print view.
The whole island.
The browser includes terrain, buildings, surveyed roof models, streets, parks, historical street trees and 2024 aerial imagery across Staten Island. Outside the original North Shore area, an 80-metre terrain overview gives way to 20-metre terrain and nearby building and public-realm chunks. Matched 2014 roof and wall models replace basic building shapes when roof detail is enabled. Buildings without an accepted survey match retain recorded-height extrusions.
Sources are NYC OTI building footprints, acquired October 3, 2026, the NOAA 2017 NYC topobathymetric DEM, and the NYC DCP 26b water-excluded borough boundary. The expansion adds 137,693 unique building IDs, plus seven clipped continuations at the North Shore edge. Together the map contains 142,178 unique building records. Invalid source heights and non-building records are excluded and counted in the source manifest.
Terrain elevations use NAVD88 and the original coordinate origin. About 99.74% of island land samples have surveyed elevation; 986 coastal samples use the nearest valid height. Terrain resolution and source dates limit precision. The North Shore zoning and proposal layers remain local. New browser coverage does not add printable tiles or alter the existing print files.
Illustration and Google Maps.
Illustrated facades carry over the St. George project's brick, limestone, sage and slate patterns, three accepted Bay Street treatments, 24 building-specific references, and the ferry terminal. Source IDs and BINs gate building-specific matches. Other buildings receive illustrative window and wall patterns. These are visual interpretations, not measured facade surveys. Reference photographs are not redistributed.
All nine reconstructed ferry models retain their hulls, decks, cabins, railings, equipment and nameplates. Assignments, movement and wakes are simulated; the timetable is separate.
Illustrated Google tiles streams Google Maps 3D imagery when selected, with toon lighting, pixel edges, palette reduction and optional local model overlay. Google imagery has its own date, geometry and approximate vertical alignment. It is separate from the NYC/NOAA source records and print files. Google Maps and visible data-provider credits remain on the map, including in Capture mode. Custom palette images are processed in your browser.
Google requests pass through this site's server. The Google view uses the Google Maps additional terms and Google privacy policy. Map Tiles attribution and usage policies.
2024 aerial imagery.
The Aerial view uses official NYC Office of Technology and Innovation 2024 orthophotography, projected onto the terrain across Staten Island. Coverage includes an island overview, all 292 island chunks and the original 32 North Shore tiles. Images load when selected and at a resolution appropriate to the view. The photograph records its original lighting and date; the sun control does not relight it.
NYC OTI orthophotography is available under Creative Commons Attribution 4.0. Terrain projection, clipping and image sizing are adaptations. The manifests record source tiles, retrieval dates and hashes. Terrain geometry and print files are unchanged.
Guided tours.
The three tours (Before / After: 2036, Water, Sun) are scripted sequences of the same views, water levels, sun settings and proposal massing you can set by hand in the Explore, Water and Sun panels; nothing in a tour is computed differently. Each stop is stored as a scene patch in data/tours/*.json.
Numbers in the 2036 tour (three sites, seven buildings, storey counts and roof heights, 2,500 homes, about 7,150 residents, the 2036 analysis year) are read from the proposal catalog built from EDC’s September 29, 2026 presentation; the massing is illustrative, traced by hand, and the proposal page carries the reconstruction notes. Water levels in the Water tour are the panel presets (MHHW, Sandy’s peak, Sandy plus the NPCC 2080 adder) described under Flood and sea level; shadow times in the Sun tour are solar positions for the stated New York dates and times. The end-card call to action follows the proposal’s published comment window and changes to “Read the proposal” once it closes.
Neighborhood stories.
Explore includes 72 neighborhood and historic place names, with 233 headline and body alternatives supplied by the site owner on October 6, 2026. The starred choice opens first. The copy is presented as supplied and has not been independently fact-checked; the source document attributes its historical material to the neighborhood Wikipedia articles.
The mint outline is an approximate reference area, not an official neighborhood boundary. Most outlines come from Chris Whong’s NYC Neighborhood Boundaries, derived from Zillow, under CC BY-SA 4.0. The Staten Island subset is reprojected into the model’s local metres, simplified by 3 m and rounded to 0.01 m. Those adapted outlines retain the same license.
Arlington uses the PediaCities / Ontodia polygon, retained in the Hodges Ward Elliott source mirror, under the ODC Attribution License. Navigation anchors use NYC Department of City Planning neighborhood label points and referenced locality coordinates.
Fourteen entries have no separate polygon in this dataset or belong to a combined mapped area. West Brighton and West New Brighton share one outline. These reference areas can differ from residents’ understanding of a neighborhood. No terrain, building, survey or print geometry is changed.
The 233 fact locations were supplied by the site owner on October 9, 2026. They anchor the camera in Explore and the neighborhood tour. Supplied place labels, confidence and boundary metadata are retained in the downloadable catalog.
Site analytics.
This site uses Vercel Web Analytics for aggregate page-view and referrer statistics without analytics cookies or cross-site tracking.
