Before & after
the Key Bridge collapse.

00 · RITIS high-confidence segment speed

The nine-date historical export covers 283 named-corridor segments. Of 1,222,560 in-window minute records, 1,162,444 (95.08%) have confidence score 30; 60,116 lower-confidence records are excluded. An additional 5,094 exported boundary-time rows fall outside the half-open study windows.

Each included segment needs all three study dates per period, each with at least 192 of 240 valid minutes after quality filtering. Daily mean estimates are averaged equally over the three days; identical eligible segment pairs are then weighted by fixed source length. The baseline can change between B and C as the paired set changes. This is not an end-to-end route speed.

Units: mph. Timezone: America/New_York. Small export metadata revisions change tested corridor summaries by less than 0.004 mph. The original export and independent calculation checks remain in the SMART project; this site exposes summary data only.

01 · Published route speed

Baltimore Metropolitan Council's Key Bridge monitoring reports supply route length and weekday average travel time. This dashboard calculates the corresponding travel-time-equivalent speed in mph.

Percent speed change is 100 × (v_after / v_before − 1). The inverse relationship means a 20% increase in travel time is not a 20% decrease in speed.

02 · Team segment analysis

TTI is the average of minute-level reference speed ÷ observed speed for each segment. Three selected Wednesdays form each period; Chai describes them as incident free, which cannot be independently checked without the raw records.

Corridor summary = Σ(length × segment TTI) / Σ(length), over identical positive, nonmissing segment pairs. The archive does not include absolute speeds or minute sample counts. Mean TTI is not inverted into mean speed.

03 · Reused project materials

• TTI_vis_v3.zip: v4 geometry and six TTI fields for 25,024 directional road segments.
• Chai six-step workflow: in-city / out-of-city centerline matching, manual completion, TTI computation, attribute integration and anonymized publication.
• SMART project brief: Pulaski, Eastern and Hanover context; future ATSPM, LiDAR, ThruGreen and Video Analysis connections.

The full source inventory, notebooks and reproducible aggregation scripts are stored in the SMART project. The published bundle contains only the selected visualization data.

04 · Interpretation limits

• Before/after comparisons show association, not a causal estimate of the bridge collapse or signal interventions.
• Weather, incidents, construction, school calendars and changing reference speeds can affect results.
• BMC routes and the team's same-name corridor groups have different boundaries.
• Road geometry includes automated matches, manually traced curves and straight-line fallbacks.
• 2025 follow-up data are not evidence of bridge reopening or SMART deployment effectiveness.