Four-Minute Coverage Map That the County Will Not Accept
Cedar County Fire and EMS maintains a 12-station response plan. Coverage is mapped as 4-minute and 8-minute Service Area polygons generated from a network dataset built from the county street centerlines. The county is now disputing the map. In the Ridgeview subdivision, completed 18 months ago, the 4-minute polygon covers nothing, yet Station 7 is 1.2 km away. Elsewhere the 4-minute polygon runs straight across Mill Pond and through a fenced rail yard that has no crossing; drivers need six minutes for those trips. Station 7's point sits about 40 m east of the centerline on the southbound carriageway of a divided highway, and its service areas reach only the southbound side. The network dataset uses a single DriveTime cost with a constant 45 km/h speed on every street, and it has no turn feature class. The chief also asks why the 8-minute ring does not grow at 5 p.m. on weekdays.
1. Explain why Ridgeview shows no coverage at all and why Station 7 reaches only one side of the divided highway.
Read the worked answer
Both are connectivity questions, not speed questions. A solver can only traverse edges that are topologically connected, and only from a location that snaps to a traversable edge.
Ridgeview. New geometry that looks continuous is not connectivity. Run the network dataset's verify/build and inspect errors on the subdivision's source edges. Edges must share endpoints or vertices within the source's snapping tolerance, and where streets cross the highway on a bridge the edges need matching z-values or grade-separated connectivity, or you get either a false connection or none. The symptom (1.2 km away, zero coverage) means no path exists, not that the trip is slow. Also confirm the new edges belong to a source that actually participates in the DriveTime cost; a source with no network attributes assigned is unconstrained and is silently excluded from the analysis.
Station 7. Facility points snap to the nearest edge by default. On a divided highway the nearest edge is frequently the opposing carriageway, and with no U-turn permitted the reachable network is only that side. Fix it deliberately: set the facility's network location fields (SourceID, SourceOID, PercentAlong) and set CurbApproach to the vehicle's right side, or move the point and tighten the snap tolerance so the intended carriageway wins.
2. The Mill Pond crossing and the rail yard crossing look similar on the map. Explain why they probably are not the same defect and how you would correct each.
Read the worked answer
Rail yard: a real network error. If a centerline runs through the yard, the solver will use it, and no amount of symbology will fix the drive time. Model the closure either as an attribute on those edges plus a restriction evaluator, or as a line barrier at analysis time. Prefer the restriction: a barrier is a per-analysis setting that is easy to forget on the next map, and this coverage map is becoming an official record.
Mill Pond: likely a polygon artifact. Service Area polygons are built from the traversed edges and then generalized with a trim distance; a large trim value fills the concave space between streets that dead-end at the shoreline and produces a straight edge across water. Export the version you publish with polygon type set to Detailed and Trim Polygons off or very small, then generalize only a cartographic copy.
Separate the third possibility. A 4-minute cutoff at 45 km/h is 3 km of travel. A trip that really takes six minutes can still plot as covered, so before blaming geometry, check whether a far-side street is being reached through a legally impossible connection. Confirm each fix by running a route from the station to a known address on the far side and comparing with a real drive time.
3. The chief's 5 p.m. question and the missing turn feature class are related. Explain what the impedance must contain and how you would validate the rebuilt coverage before publishing it.
Read the worked answer
Add a traffic-aware impedance, not a blanket speed reduction. Service Area honors a time of day only when the impedance has a traffic evaluator. In the network dataset properties, build a DriveTime cost whose TravelTime evaluator reads the historical traffic Speed field by day-of-week and time of day, backed by a traffic data source with at least a year of records. Then set the analysis's Time of Day to a Tuesday 17:00 and rerun for the AM peak. If no traffic data exists, the defensible substitute is functional-class speeds plus an intersection-density delay, with the assumption printed on the map.
Turns matter. Without a turn feature class every left turn is free and illegal turns are legal, which systematically understates urban times. Add a turn feature class with global turn delay, and prohibitions if the street data carries them.
Hierarchy. Disable it for a 4-minute ring; hierarchical cutoffs distort short analyses. Keep it optional for whole-county 8-minute runs.
Validate before publishing. Run an OD Cost Matrix against 20 to 30 trips with independently observed drive times during both peaks, report mean error and worst cases, and publish the network dataset build date and traffic period with the map. A rebuilt network that nobody validated is a new map, not a corrected one.