A GIS-based exploration of what we actually see when we walk through a city.

Overview

Cities are often defined by their skylines, but the way people experience those skylines is fundamentally different from the way they appear in postcards or drone photographs. When we move through a dense urban environment, our ground-level vantage point is constantly being obstructed by buildings, trees, elevation changes, and the built environment around us.

This, along with spending the summer in New York City, led me to wonder: Which buildings in NYC are the most visible from the places where people actually stand?

Using GIS, 3D modeling, and a city-scale viewshed analysis, I created a “visibility hierarchy” of Manhattan showing which buildings define the real, everyday skyline.

Inspiration

Before this project, I 3D-printed a raised topographic map of California using data from ArcGIS.

That print completely shifted how I thought about topography and spatial data. I had to scale the elevation 8x in order to significantly feel the terrain changes. I’m drawn to design and civil engineering, so after this experience, I naturally wanted to explore 3D buildings in ArcGIS, especially in cities known for their skylines.

Around this same time, I also made a 3D map of San Francisco, my hometown and the skyline I know best. Including that map early in this project helped ground the analysis and reminded me how familiar skylines feel from lived experience, not just from photographs. This further motivated me to examine NYC, the city with arguably the most iconic skyline in the world.

Building the 3D City Model

To study visibility, I began with combining building footprints that included height attributes and a Digital Elevation Model (DEM). This essentially served as a flat representation of the city’s physical form, every building, rooftop, and hill captured in one integrated layer.

Observation Points and Viewshed Analysis

I chose subway entrances as the primary observation points to run the viewshed analysis. These locations represent where people actually emerge into the city, instantly facing the built environment around them. A viewshed works by tracing straight lines from the observer to the surrounding surface. If the line reaches a building without being blocked by another structure or terrain feature, that building is counted as visible from that location. Running this analysis for every subway entrance generated a visibility frequency surface, which effectively measures how often each building can be seen by everyday city users.

Mapping Visibility in 2D

I summarized visibility values for each building footprint and created a 2D map showing which structures appeared most frequently across all observation points (above). This map reveals a pattern that significantly diverges from the simple height distribution. While the tallest structures concentrate in Lower Manhattan, Midtown, and Hudson Yards, visibility clusters emerged along major corridors and neighborhoods where street geometry and local elevation make certain buildings appear much more often than expected.

Extruding to 3D

To make the results easier to interpret and more visually intuitive, I extruded the footprints back into 3D buildings and colored them according to visibility score. This created a model where building height and visibility are represented simultaneously, taller forms rise from the ground, but their color intensity communicates how often they are actually seen from pedestrian viewpoints. In this model, some mid-rise buildings become visually prominent because they sit at key intersections or along long sightlines, while several tall skyscrapers fade because they are blocked by closer structures.

Comparing Height and Visibility

Placing the height-based 3D model next to the visibility-based model highlights the contrast between idealized skylines and lived experience. Buildings that dominate postcards do not always dominate actual visibility. In some cases, skyline-defining skyscrapers are hidden throughout most of Manhattan due to their deep placement within clusters of similarly tall structures. Conversely, a few buildings that are not the tallest nonetheless appear highly visible because of their position relative to major avenues or because they stand alone in neighborhoods with fewer vertical obstructions. This comparison emphasizes that a city’s visual identity is shaped by more than height—it is shaped by geometry, street layout, and the human vantage point.

Key Findings

The analysis identified Central Park Tower as the most visible building in Manhattan, appearing in the viewsheds of eighteen subway entrances. This was notable not just because it has the highest roof height in the United States (1550 ft), but because its location along Central Park’s edge grants it long, unobstructed paths for visibility. More broadly, I found that visibility does not scale linearly with height. Many mid-height buildings shape the skyline more than expected, and some iconic skyscrapers are rarely seen outside a few select vantage points.

Limitations

Although the model provides meaningful insights, it comes with limitations. Building footprint data can miss small structures that might obstruct visibility, and the viewshed tool assumes that smaller objects (trees, vehicles, scaffolding) do not block line of sight. Subway entrances capture common viewpoints but don’t fully represent every pedestrian experience, and the extrusion process assumes that buildings extend vertically without setbacks or architectural variation.