Manhattan noise-risk atlas

The Manhattan Soundscape

Uncovering Spatial-Temporal Rhythms and Urban Noise Exposure

A scrollytelling visualization of 311 noise complaints, urban sound sources, social exposure, and noise-aware movement across Manhattan.

Manhattan Building Noise

Noise Sticks to Manhattan

The story begins with a simple proposition: in Manhattan, noise does not simply disappear after a complaint is filed. It accumulates on buildings, turning ordinary footprints into an invisible layer of urban pressure.

Chapter role What / Where

Question

Where does noise stick to the city?

Evidence

3D buildings encode accumulated complaint intensity.

Insight

Noise behaves like a spatial pollution layer.

Buildings

--

Complaints

--

Max Risk

--

Accumulated Noise Index

Low51530+Peak

Bright points mark representative high-noise building centroids. Extrusion height is symbolic noise intensity, not physical building height.

Level 02 / Manhattan Temporal Rhythm

When Does Noise Wake Up?

After locating the accumulated field, the story turns to rhythm. The ridgeline asks when Manhattan becomes loud, while the right map is kept as a restrained response layer rather than the main evidence.

Chapter role When / Rhythm

Question

When does the city wake up acoustically?

Evidence

24-hour ridgelines compare category rhythms.

Insight

Different noise types follow different urban clocks.

Total Complaints

--

Peak Hour

--

Night Share

--

10PM reading: late-evening noise is treated as a temporal signature rather than a spatial hotspot.

24-hour Noise Rhythm Matrix

The vertical line is the selected hour. Story cards highlight the category that explains the rhythm.

10PM

Level 03 / Manhattan H3 Spatio-temporal Grid

From Noise to Sources

Once the rhythm is visible, this chapter asks why specific areas become noisy at specific times. H3 hexagons show the hourly field, while nightlife venues, subway stations, and Manhattan bus corridors act as evidence layers for possible sound sources.

Chapter role Why / Source mechanism

Question

Why does this place sound like this?

Evidence

H3 fields are read with nightlife, subway, and bus layers.

Insight

Noise hotspots are produced by urban activities, not random points.

Current Hour Total

--

Active Hexes

--

Top Source

--

Source Composition at Selected Hour

H3 grid + source markers are controlled by the same 24-hour slider.

10PM

Source evidence layer: the H3 grid still comes from 311 complaints, but the auxiliary layers explain possible origins. Pink points are real nightlife POIs, cyan points are subway stations, and green lines are bus corridors clipped to Manhattan only, not the full NYC bus network.

Bar / nightlife clusters

Transit / vehicle sources

Street / sidewalk sources

Construction sources

Residential conflict

Mechanical / other sources

Real nightlife venues

Subway stations

Manhattan bus corridors

Level 04 / Manhattan Soundscape Clusters

Types of Soundscape

The final analytical chapter steps back from individual hours. Instead of asking whether a place is louder, it asks what kind of soundscape it belongs to: nightlife core, residential impact zone, transit-street exposure, construction belt, or mixed-use conflict.

Chapter role Typology / Districts

Question

Are noisy places different only in intensity, or also in kind?

Evidence

Cluster colors summarize source mix and temporal behavior.

Insight

Manhattan is composed of multiple soundscape districts.

Clustered Cells

--

Soundscape Types

--

Selected Type

--

From Local Fields to Soundscape Types

Click a cluster card or a colored cell on the map. The reduced space shows how each soundscape type separates; the detailed radar profile appears only in the side panel.

Reduced Cluster Space

Socio-acoustic Burden

Who Bears the Noise Burden?

The story now leaves the full-screen map. Each dot is a Manhattan census tract, positioned by socioeconomic status and noise exposure. The question is not only where noise occurs, but who lives with it.

Chapter role Equity / Burden

Question

Do lower-income communities also experience higher complaint-based noise exposure?

Evidence

Median household income is compared against 311 noise complaints per 1,000 residents at census-tract scale.

Insight

Noise becomes a social exposure pattern, not just a spatial hotspot.

Priority Tracts

--

Median Income

--

Income × Noise r

--

Income–Noise Exposure Space

X-axis: median household income. Y-axis: 311 noise complaints per 1,000 residents. Dot size represents population.

Low income / high noisePotential socio-acoustic burden

High income / high noiseActivity-rich high exposure

Low income / low noiseLower measured exposure

High income / low noiseHigher-income quiet profile

Select a census tract

Click a dot in the quadrant chart or a tract in the small inset. The full-screen map is intentionally muted here so that the social comparison becomes the main visual argument.

Median Income

--

Noise / 1k residents

--

Population

--

Poverty Rate

--

Linked tract inset / bivariate classes

Color combines noise exposure and income: upward means higher noise; rightward means higher income.

Interactive Noise-aware Routing

Choose Two Points. Avoid the Noise.

This final chapter turns the noise field into an action layer. Click any two points on the Manhattan map. The system snaps them to the H3 risk graph and computes two routes in real time.

Chapter role Action / Decision

How to use

First click sets the origin. Second click sets the destination. The routes are then computed automatically.

Routing logic

Gold uses ordinary distance. Cyan uses Dijkstra with an additional penalty for crossing high-noise hexagons.

Map reading

The glowing field around the paths shows the routing-hour noise risk. Red/pink cells are costly zones that the quiet path tries to bypass.

1 Click origin

2 Click destination

3 Compare routes

Selected route

Waiting for two map clicks

Origin

Not selected

Destination

Not selected

Routing hour

--

Graph cells

--

Exposure reduction

--

Extra distance

--

Hotspots avoided

--

Interactive route theatre

Click the map itself. The selected origin and destination become route markers, and the two paths draw progressively through the noise field.

High noise risk

Shortest path

Lower-noise path

Clicked points

Route comparison

After two points are selected, the panel updates with distance, average risk, cumulative exposure, and hotspot crossings.

Shortest

Minimum distance route

--

Avg risk--

Hotspots--

Exposure--

Quiet

Lower-noise safety route

--

Avg risk--

Hotspots--

Exposure--

Trade-off summary

Distance

Noise exposure

This is a conceptual route over the H3 sound field, not turn-by-turn navigation. It shows how the same city can produce different movement choices when exposure is part of the route cost.

Methods & Credits

Data, Methods & Credits

This project integrates open urban data, spatial indexing, clustering, socioeconomic analysis, and interactive routing to construct a narrative visualization of Manhattan’s soundscape.

CASA0029

Group 4

Chaoyue Zhang22081693

Chenwei Fang25168446

Data Sources

311 Noise Complaints

NYC Open Data

Core dataset for identifying complaint locations, complaint categories, temporal patterns, and accumulated noise exposure.

https://nycopendata.socrata.com/Social-Services/311-Noise-Complaints/

3D Building Footprints / Building Geometry

NYC Open Data / NYC Department of Information Technology and Telecommunications

Used to construct the 3D building-based soundscape map, where accumulated complaints are mapped onto Manhattan buildings.

https://data.cityofnewyork.us/City-Government/BUILDING/5zhs-2jue/about_data

MTA Bus Routes

MTA Open Data

Used as a transit-related sound source layer to explain street-level and corridor-based noise exposure.

https://data.ny.gov/zh/Transportation/MTA-Current-Bus-Routes-Map/559g-c57t

Subway Stations

MTA Open Data

Used as transit infrastructure points to support the interpretation of noise concentration around mobility nodes.

https://data.ny.gov/Transportation/MTA-Subway-Stations/39hk-dx4f/about_data

Nightlife Points of Interest

OpenStreetMap / Overpass API

Used to represent nightlife-related noise sources such as bars, pubs, nightclubs, and selected restaurant-bar venues.

https://overpass-turbo.eu

2020 Census Tracts

NYC Open Data / U.S. Census Bureau TIGER/Line

Used as the spatial unit for linking noise exposure with population and socioeconomic indicators.

https://data.cityofnewyork.us

ACS Socioeconomic Data

U.S. Census Bureau American Community Survey 5-Year Estimates

Used to obtain median household income, total population, and poverty-related indicators for census tract-level inequality analysis.

https://www.census.gov/programs-surveys/acs

Street Centerline Network

NYC Open Data

Used to construct the street-based graph for the noise-aware routing analysis.

https://nycmaps-nyc.hub.arcgis.com

Map Tiles and Basemap

Mapbox

Used for 3D basemap rendering, interactive map navigation, and visual styling.

https://www.mapbox.com

Methods & Tools

Spatial mapping311 noise complaints were linked with building footprints, H3 grids, census tracts, and street centerlines.

Temporal aggregationComplaints were aggregated into 24-hour cycles to reveal category-specific daily rhythms.

H3 indexingPoint complaints were converted into high-resolution hexagonal units for hourly sound-field mapping.

Sound source interpretationBus routes, subway stations, and nightlife venues were added as explanatory evidence layers.

Clustering and equity analysisSoundscape types were identified through clustering, while tract-level exposure was compared with ACS socioeconomic indicators.

Noise-aware routingA Manhattan street centerline graph was used to compare shortest-distance and lower-noise routes.

Frontend frameworkPython, GeoPandas, H3, scikit-learn, Mapbox GL JS, D3.js, HTML, CSS, and JavaScript were used.

Contribution Statement

Chaoyue Zhang Led the project concept, data preprocessing workflow, spatial analysis, scrollytelling structure, frontend integration, and the noise-aware routing module.

Chenwei Fang Supported dataset collection, map-layer checking, interface testing, visual interpretation, and refinement of the story explanation.

Use of AI Tools

AI tools were used as supportive tools for code debugging, Python preprocessing, HTML/CSS refinement, Mapbox GL JS and D3.js interaction improvement, and narrative wording. Dataset selection, analytical decisions, visual interpretation, and final integration were reviewed and controlled by the group members.