Silencing the Span

Manhattan Bridge — traversal frequency, direction, overlap and who is underneath

Every B, D, N and Q crossing, by hour, by direction, by route — from the MTA's own published schedule — set against the first estimate of how many people are in the corridor to hear them. Feed . Periods follow the EU Environmental Noise Directive 2002/49/EC default day / evening / night split. Verified 5/5 Pedestrian figures are derived from four public datasets and are weaker — see the ratings on those panels.

Two crossings within 6 s are heard as one

A period mean hides a great deal. The night figure averages a 12-per-hour floor at 02:00 with the 06:00 ramp; the hourly range shows the spread, and the chart below shows the shape. Acoustic events/hr is a bracket, not a measurement — see the overlap panel for why it cannot be narrowed from this data.

One consequence of using the standard periods. The night period runs to 07:00, so the 06:00 hour sits inside it — and on a weekday that hour carries 42 crossings, more than triple the 12-per-hour floor at 02:00. Under the Lden convention the night period carries a +10 dB penalty and the evening +5 dB, on the reasoning that the same sound does more harm at those hours. That makes 06:00–07:00 the single most heavily weighted busy hour of the day, and it is invisible in any framing that starts the day at 06:00.

Period boundaries 5/5 — quoted from Directive 2002/49/EC Annex I: "the day is 12 hours, the evening four hours and the night eight hours… the default values are 07.00 to 19.00, 19.00 to 23.00 and 23.00 to 07.00 local time." The +5/+10 weightings 3/5 are the Directive's standard Lden formula, read from secondary summaries rather than the formula image in the source PDF. No penalty is applied to any number on this page — this is a count of events, not a level.

Traversals per hour

Stacked by route. Shaded bands are the three noise periods. The line is estimated distinct acoustic events under the selected overlap model — where it sits below the bars, trains are merging into single audible events.

Direction split

Northbound = toward Manhattan. Southbound = toward Brooklyn. Both cross the same four tracks over the same ground.

Route mix

Weekend composition is not the weekday mix attenuated.

Where the overlap question breaks Unverifiable from schedule

This panel exists to stop you trusting the number above it.

Every scheduled departure in this feed falls on an exact :00 or :30 second. Across weekday traversals there are exactly two distinct sub-minute values. The schedule is quantised to 30 seconds.

So the true gap between two crossings is never recorded as 1 s, or 4 s, or 12 s. It is recorded as 0 s or 30 s. Counting coincidences in the schedule measures the scheduler's rounding, not the railway.

Gaps between consecutive crossings, as written in the feed. Note the total absence of any value between 0 and 30 seconds — the entire range in which two trains would actually merge acoustically.

What can honestly be said instead

Since the schedule cannot locate the answer, the dashboard brackets it with two models that can be justified without it:

Regular assumes perfectly even spacing — the least merging physically possible, so the most events. Poisson assumes maximum disorder; the superposition of four semi-independent track streams tends toward Poisson (Palm–Khintchine), which makes it a defensible ceiling on merging and therefore a floor on events. Real operation lies between. Schedule-as-written is shown only to demonstrate that it is noise: drag the coincidence window from 1 s to 29 s and watch it refuse to move at all, then jump the instant you cross 30 s. That is the rounding, not the railway.

Does the corridor change the park finding? Partly withdrawn

An earlier version of this programme published the following:

"Peak park use and peak train frequency are out of phase. Brooklyn Bridge Park is most heavily used on weekend afternoons. That is when bridge traffic is at its lowest daytime value — 34–36 per hour on a Saturday against 57–66 per hour on a weekday afternoon."

The arithmetic is right and the conclusion is too narrow. That generalisation is withdrawn. It silently assumed one receptor population — park visitors — and then spoke as if it were the affected population.

Widen the receptor to the corridor that is actually affected, from the York St station down to the water entrance to Brooklyn Bridge Park, and there is no single phase relationship, because there is no single population:

The worst case is not the park. It is the corridor on a weekday morning, when the train rate is at its 24-hour maximum — — and the corridor is carrying commuters. The original framing could not see that case, because it was only looking at the park.

That replacement is also withdrawn Second correction

The paragraph immediately above was written before any pedestrian data existed. It is now measurable, and it is wrong.

"The worst case is not the park. It is the corridor on a weekday morning, when the train rate is at its 24-hour maximum."

That is withdrawn. The 08:00 hour does carry the most trains. It carries roughly of the people. Multiply the two and the weekday 08:00 hour scores of the daily maximum. The actual maximum is .

The programme has now made the same error twice, in opposite directions. The first framing optimised on attendance alone and concluded the concern was the park on a weekend afternoon. The correction optimised on train rate alone and concluded it was the corridor on a weekday morning. Exposure is the product, and the product peaks in between — in the early afternoon, on every day of the week.

Weekday, Saturday and Sunday all peak within an hour of each other despite carrying very different train rates, because presence varies more across the day than train rate does. Train rate is nearly flat from 07:00 to 19:00; presence changes by a factor of four inside the same window. Where people are matters more than where trains are, and only one of those two had ever been measured.

The index is relative — each day type scaled to its own maximum. No absolute person-event figure is published here, because presence is a lower bound covering only subway-delivered transients, and multiplying a lower bound by a train count would produce a number that looks authoritative and is not.

What is verified and what is not. The train rates in this panel are 5/5 — MTA's own feed, reproducible. Presence is 2/5, and that is an improvement on the 1/5 this panel carried until the denominator work below was done. It is still true that no pedestrian has been counted directly anywhere in this corridor. What has changed is that arrival rate is now measurable from fare data, so the exposure index above is a statement about shape rather than a guess. An absolute exposure figure still cannot be published, because dwell time is unmeasured and residents are not in the accumulation at all.

There is also a resolution here of an earlier correction. York St is an F train station on the Rutgers Street Tunnel and no train that crosses this bridge stops there — so it is useless as a place to measure the source. It is nonetheless the pedestrian gateway to the affected corridor. It matters as a receptor origin, not as a source.

The denominator, partly counted New

The panel above says no pedestrian count exists. That was true of direct counts and it is still true. But two MTA datasets and one city dataset can be made to answer a narrower question: how many people arrive in this corridor, and when. Figures are a typical day for the selected day type. Subway , walkway .

Arrivals and entries are different measurements, not two views of one. Entries are observed — a tap or a swipe, counted by a turnstile. Arrivals are inferred from where riders next re-enter the system. The first is the corridor emptying; the second is it filling.

Flow in, flow out, and what accumulates

Top: arrivals above the axis, entries below, by hour. Bottom: the running difference — the subway-delivered population still in the corridor — with the train rate over it. Where the two are both high is where exposure is concentrated.

The corridor fills in the morning and empties in the evening. Net flow is . That is the signature of a destination district, not a dormitory. A purely residential neighbourhood would show the opposite — residents leaving at 08:00 and returning at 18:00.

This matters for who is being exposed. The people under the bridge during the busiest train hours are disproportionately not the people who live there. They are workers and visitors, and they are the population least likely to appear at a community board meeting.

How much of this can be trusted

Over a full day, arrivals and entries at the same two stations should balance. They do, to . The two are not fully independent — the origin-destination estimate is scaled to match total system ridership — but that scaling is system-wide, so agreement at the level of two station complexes is still informative. The residual is carried through as the uncertainty on the accumulation curve.

Residents, and the part of the denominator that is still missing

Who lives here

Residential units from tax-lot records, inside three explicitly stated boxes. Units are a tax fact. The people inside them are an assumption.

Little’s Law, half solved

The number of people present in an area is L = λW — the arrival rate multiplied by how long each person stays.

This page now has λ. It does not have W. A commuter crossing from York St to an office on Water St is exposed for four minutes. A family on the Brooklyn Bridge Park lawn is exposed for three hours. Both count once in the arrival figures.

Until dwell time is measured, the accumulation curve below is the best available and it is a lower bound on transient presence, not a count of people present.

What the accumulation curve leaves out, in rough order of size.

  • Residents — between people who are present whether or not they touch a turnstile. This is almost certainly the largest single omission, and it is the population whose exposure is continuous rather than episodic.
  • Walkway arrivals — the Brooklyn Bridge discharged pedestrians toward Brooklyn on a typical day in 2019. That counter has been dead since 2019, so the current figure is unknown, and 2019 also predates the 2021 reconfiguration that moved cyclists off the promenade.
  • Every other mode — ferry at Fulton Ferry Landing, the B25 and B67 buses, Citi Bike, private car, and the Brooklyn Heights stair. None is counted here.
  • Mode-swapping — anyone who arrives by subway and leaves on foot over the bridge is counted arriving and never leaving. The small daily closure error suggests these flows roughly cancel; it does not prove it.

Ratings. Entries 5/5 — observed fare transactions, MTA's own feed. Arrivals 4/5 — MTA's own feed, but destinations are inferred from return swipes, not observed. Walkway counts 4/5 — observed and directional, but seven years stale. Residential units 5/5 — tax records. Residents 2/5 — units multiplied by an assumed occupancy and household size, reported as a range for that reason. Dwell time 1/5 — not measured by anyone, anywhere in this corridor.

Who is actually here, hour by hour New

The accumulation curve above answers “how many subway-delivered people are still in the corridor?” It cannot tell a four-minute commuter from a family that stays three hours, because it has no concept of how long anybody stays. This panel replaces it with a cohort survival model: arrivals are split into four populations, each is given a dwell distribution, and everyone is aged out of the district accordingly.

L(t) = sum over cohorts, sum over arrival hours s ≤ t, of Ac(s) · Sc(t − s). Arrivals are data. Dwell is assumption. Departures are data. The model's implied departure curve is scored against observed MTA entries, which is the only reason any of this is falsifiable.

Presence by cohort

Does it reproduce the observed departures?

Solid: departures the model implies. Dashed: hourly entries MTA actually recorded. The model was never shown this curve during construction, only during scoring.

What the fit does and does not pin down
The finding that matters here is a negative one. The total is reasonably well determined. The labels are not. Within the band of parameter sets that fit the observed departures essentially equally well, the number of non-residents present at 14:00 moves by roughly a tenth — but the split between “worker” and “visitor” inside that total swings by a factor of two or more. That is not noise in the estimate. It is a structural property: a departure curve carries no job titles, so a person who stays eight hours looks identical whether they came to work or came for the day. Any narrative built on the cohort mix rather than the cohort sum is reading detail the data does not contain.
Total presence is not an exposure curve. Total presence peaks when residents are at home, which on a weekend is the early morning. Someone asleep behind a closed window and someone standing on the waterfront path are one person each to this model, because the model has no geometry. Exposure is the integral of noise along a path; this model supplies the duration term and has no position term at all. That gap is the whole argument for the agentic population model, and it is why the non-resident line, not the total, is the one to read here.

Ratings. Arrivals and entries 5/5 and 4/5 as above. Visitor dwell distribution 4/5 for what it measures — a published visitor survey with real bins — but 2/5 as a transfer, because it was measured at Waterfront Park in Louisville, Kentucky, not in Brooklyn. Worker and transient dwell 1/5 — assumed outright. Cohort labels 1/5 — not identified by the data, see above.

What this dashboard cannot tell you

Regenerate the train data with python data-collection/build_dashboard_data.py and the pedestrian data with python data-collection/build_pedestrian_data.py. Settle the overlap question with python data-collection/bridge_realtime.py --poll 30 run for one week — that is Method 26, and it is free.