User guide

Territory — Vaucluse boundary (84)

Always shown. Dashed outline of the department; a reference for framing. Lines in the same colour, solid or dotted, mark the outlines of the sectors at the territorial scale.

Territorial scale

Narrows the displayed network to one sector, at five scales: road agency (4), maintenance centre (11), canton (17), intercommunality (14), municipality (150).

How: pick the scale, then the unit in the list — clicking it again restores the whole department. Above twelve units a filter field appears. The map reframes on the sector, and the page URL keeps the choice (terr parameter), so it can be shared as is.

Maintenance sectors come from the operating limits of the departmental road network published by the Department on DataSud, attached to road sections rather than drawn as areas — 57 municipalities are served by more than one centre, so no official area exists to overlay. Each OSM section's sector is computed offline by proximity; the municipality and the canton are determined by inclusion in the OSM boundaries, and the intercommunality derived from the INSEE code via the État's API Géo.

The outlines of every unit of the chosen scale are drawn in dark blue: the whole division, not only the sector being filtered on. The selected sector thickens its line and takes a faint tint. Hovering an outline names it and gives its mileage; clicking it selects the unit, which saves hunting for one municipality among 150 in the list. A wide invisible band runs along each line so that it can be aimed at, but it sits below the roads: they keep priority for their own popups.

A single colour for every boundary, only the stroke pattern distinguishing them: long dashes for the department, a solid line for a constituency — canton, intercommunality, municipality — whose perimeter is exact, and a dotted line for a maintenance sector, whose perimeter is not. A road agency and a maintenance centre have no official boundary: their perimeter is rebuilt by attaching each municipality to the sector maintaining the most mileage within it. The panel states what that shortcut displaces — 2.1 % of the mileage for agencies, 5.8 % for centres. An intercommunality's outline is reassembled from the boundaries of its member municipalities, no source publishing it, and stops at the department boundary for the six that extend beyond it.

Relief. Picking a sector fades out everything around it: a veil covers the map, pierced to the exact shape of the perimeter. The sector stands out without being tinted or highlighted. The veil is translucent rather than opaque — the surroundings stay recognisable, and that is what tells you where the sector sits. It sits at the very back, so the neighbouring outlines still say what you would be clicking and the roads keep their full contrast.

The digest. Picking a sector splits the view in two: the map reframed on the sector on the left, what is known about it on the right — mileage, sections, routes, municipalities crossed, then three bar charts: the share of each network level, the share of each speed limit, and the five routes carrying the most kilometres there. Every bar carries the colour of what it measures: the network levels in their own colours, the speeds in those of the Limitations mode, and each route in the colour of the level where it carries the most kilometres — the colour it is drawn in on the map, right beside it. Nothing overlaps, so the sector reads whole beside its figures. The close button gives the width back to the map without losing the selected sector; on a narrow screen the pane becomes a bottom sheet. The digest is computed on demand from the sections the filter retains, so it describes exactly what is on screen.

The filter acts on visibility, not on styling: it combines with the Limitations mode, the hierarchy levels and the labels.

Network hierarchy

Three levels of departmental roads, each with its own colour:

• Red — regional-interest network (e.g. D900).
• Orange — territorial development network.
• Blue — local-interest network.

How: click the Network hierarchy title or each line to enable a level. Click a segment for a card (name, surface, OSM / Wikidata links).

Road-number shields. Each level becomes fully readable at its own zoom, fading in over the preceding three quarters of a level: red from zoom 10, orange from zoom 12, blue from zoom 13. Shields are anchored on the portion of the road actually on screen, so they stay visible and follow the road as you pan. Where numbers still overlap after repositioning, they merge into a clickable group badge listing the roads involved.

Cycle routes

OSM cycle itineraries (route=bicycle). Totals in kilometres: Via Rhôna (EV17), Méditerranée (EV8), Via Venaissia (V861), local routes.

How: enable via the section eye. Major cycle routes are emphasised by colour.

Roads under construction

Segments tagged construction or proposed in OSM.

How: enable the layer; the segments appear on the map. Counters in the legend.

Main towns

Sub-prefectures and large municipalities (centres + labels).

How: enable the layer for place-name markers.

Roadside areas & car-pooling

Related themes highlighting other CD84 missions along departmental roads (issue #7), sourced from OpenStreetMap:

• Car-pooling areas 🚗 (amenity=car_pooling).
• Rest areas 🌳 (highway=rest_area).
• Park-and-ride 🅿️ (amenity=parking + park_ride).
• Laybys 🅿️ (amenity=parking + parking=layby) — roadside stopping areas along departmental roads.

Parking is deliberately limited to park-and-ride and laybys to stay on roadside areas, without showing the thousands of private car parks. Objects are handled whether mapped as a node, an outline (way) or a multipolygon relation.

When the area is mapped as a surface in OSM — either a simple outline or a multipolygon relation (e.g. the Île Piot P+R in Avignon) — its footprint is drawn (light fill coordinated with the type colour) with a centred marker.

How: enable the “Roadside areas & car-pooling” layer; each type has its own swatch and counter in the legend. Click its row to hide or show that type independently. Click a marker (or the footprint) for details: type, capacity, operator, access, an OSM completeness indicator (present / missing attributes), the Wikidata link when present (otherwise flagged “no link”), and nearby street-level photos from Mapillary and Panoramax when coverage exists — click a thumbnail to open the source photo in a new tab. View / Improve links on OpenStreetMap.

Bridges layer

OSM civil-engineering structures across the department. Display adapts to zoom:

• Far zoom — clusters (number of bridges or photos).
• Zoom ≥ 12 — the structures' footprint on the map; clusters disappear.
• Zoom ≥ 16 — Panoramax / Mapillary photos near the structure.

How: enable the layer then zoom in. At far zoom, click a cluster to zoom in progressively; clicking a bridge opens the full-screen 3D view.

Bridge 3D view

Clicking a bridge opens a full-screen 3D scene where the structure is rebuilt and surrounded by its photos. Orbit with the mouse, scroll to zoom.

Model & data sources. The model is generated on the fly from the structure type (arch, aqueduct, beam, truss, suspension), material, length, width and number of spans. Information is looked up in this order:

1. OpenStreetMap first (bridge:structure, material, length, width… tags).
2. Wikidata next, if the bridge is linked to it, to fill the gaps: structure type (P31), material (P186), number of spans (P1314), length (P2043), height (P2048), width (P2049). A small info card (name, image, inception, material, dimensions) then appears, with clickable links to Wikidata and OpenStreetMap.
3. Structure name as a last resort for the arch count (e.g. "pont des 13 arches" → 13 spans).

2D map under the bridge. A basemap (IGN Plan) is laid flat beneath the model, centred on the structure and north-up: it shows rivers in blue and roads. Falls back to a plain water plane if the tiles are unavailable.

Panoramax / Mapillary photos. Each photo is placed at its real capture location (provider geolocation), with a ground marker and a mast. The image faces outward (the photographer's viewpoint); the back, facing the bridge, is coloured by source — green for Mapillary, violet for Panoramax. A colour dot marks the role of the documented element (pier, abutment, deck…), using the same colour code as the map.

Navigation. Clicking a gallery thumbnail or a panel refocuses the camera on it, shows it enlarged in the centre (the other photos fade so they don't hide it) and shows a link to the source. Click empty space to deselect. Esc or × to close. If WebGL is unavailable, a 2D fallback gallery is shown.

Road signs (stop / give way)

OpenStreetMap layer: stop (red octagon) and give way (downward triangle), pre-generated by a GitHub Actions workflow (several thousand points across Vaucluse).

What data exactly? These are the OSM nodes highway=stop and highway=give_way: they mark the traffic rule at the junction (where you must stop / give way), with a direction tag for the relevant way. They are not the physical "sign" objects (which would be traffic_sign=stop / give_way, very rarely mapped in France). So we draw realistic pictograms placed on these regulatory points: coverage is far more complete, at the cost of this semantic nuance.

How: enable the layer. When zoomed out, points are grouped into clusters (red circle with the element count); clicking a cluster zooms into it. From zoom ≥ 15, the pictograms are drawn (stop octagon, give-way triangle), only for the current view. A pictogram ringed in green has a nearby Mapillary photo: click it to view.

Direction signposts

Direction masts from OpenStreetMap (information=guidepost nodes): the bladed posts pointing to hiking and cycling destinations and points of interest (several thousand across Vaucluse, including the CD84 network).

How: enable the layer. When zoomed out, points are grouped into clusters (click to zoom in); from zoom ≥ 15, a signpost pictogram (post with blue/green blades) is drawn. The popup lists the destinations and ends with an OpenStreetMap link: voir to open the node, or compléter to open it straight in the iD editor.

Sorting by use. A hiking mast and a cycle-route mast do not serve the same reading of the map. Each mast therefore carries a use badge at the foot of its post — 🚶 hiking (hiking), 🚲 cycling (bicycle), 🚵 mountain bike, 🐴 riding — and the legend offers a filter by use with a count for each (1,575 hiking, 120 cycling, 509 with no declared use across the Vaucluse). Switching "Randonnée" off leaves only the rest on screen; a mast carrying two uses stays visible as long as one of them is active. Only the uses actually present in the data are offered.

Photos of the signpost. Many masts carry one or more photos of the sign itself in OSM (panoramax, mapillary, wikimedia_commons, image tags) — about 1,450 of the 2,170 masts in the Vaucluse. They are then shown in the popup; beyond a single photo the block becomes a carousel (‹ › arrows and a counter), since one mast gathers up to eight — one per blade, per bearing or per year. Each view is captioned with its provenance and the tag suffix where there is one (panoramax:N → "facing north", mapillary:2017 → the year). Clicking the photo opens the original at the provider. A violet dot on the mast flags this case. Without a referenced photo, we fall back to searching for a nearby Mapillary photo (green dot), which shows the surroundings rather than the sign.

Town-boundary signs

Village entries and exits from OpenStreetMap (traffic_sign=city_limit nodes, 468 across the Vaucluse, 446 of them named): the EB10 / EB20 signs marking where the speed regime changes through a built-up area. A natural companion to the Limitations mode.

How: enable the layer. When zoomed out, points are grouped into red clusters (click to zoom in); from zoom ≥ 13, each sign is drawn as the real plate — the place name on white with a red border. A plate crossed by a red diagonal is an exit from the built-up area (city_limit=end). The popup gives the direction, alternative names (including Occitan), reference and operator where available, the photos of the sign referenced in OSM, and an OpenStreetMap link (voir / compléter in the iD editor) to fix the sign straight from the map.

Road safety

Multi-year cloud of injury crashes (BAAC / ONISR, 2019 → 2024) geolocated across the Vaucluse. Two visual encodings:

• Colour = recency — recent crashes appear in a vivid/saturated orange, older ones in a dark red.
• Severity — fatal crashes leave the recency ramp: a black dot ringed in white, slightly larger, spotted at a glance whatever the year. Hospitalised injuries keep the recency colour with a black ring; slight injuries have only a thin white outline.

The legend embeds a temporal dataviz: the old → recent colour scale, a mini-histogram of crashes per year, and a year-range slider (from / to) to filter the cloud.

How: enable the layer; drag the year slider to watch the evolution over time; click a point for details (year, casualties, date, municipality, address). Note: BAAC geolocation is patchier for older years, hence a denser cloud on recent vintages.

Traffic-count stations

CD84 stations (data.gouv.fr / DataSud), classified by traffic: high (> 20k veh/day), medium, low. Each badge shows the rounded average traffic at its centre (e.g. 22k for 22,136 veh/day), readable regardless of colour.

How: enable in the legend or via the CD84 counts button on the map (both stay in sync). Click a station for a detailed card: AADT (all vehicles), HGV share and flow, and a traffic-evolution chart across every available vintage (1996–2025) with the first → last trend.

Clicking a station also highlights the matching road axis — a distinct indigo trace, separate from the round count badges — and shows the axis details (length, municipalities crossed…) in the Route information panel, without moving the view.

Weather stations

A few stations across the department (Open-Meteo). Current data: temperature, wind, humidity.

How: enable the layer; click a marker for the card. One station is also picked at random in the header.

Road events (Bison Futé)

Roadworks, congestion, ongoing crashes (DATEX II / Info Routière feed), filtered to the area.

How: enable the layer; icons on the map, popup on click.

Webcams

4 CD84 traffic webcams (Bonpas) and Mont Serein (Vision-Environnement).

How: enable the layer; click a marker to start the live video feed in the popup.

Grouped cameras. Two cameras on the same roundabout (Bonpas, site 2) sit just twelve metres apart: they overlap right up to the maximum zoom, leaving the second one unreachable. Markers that are too close are therefore merged into one, badged with the number of cameras. Its popup acts as a picker: choosing a camera starts its feed, and the ‹ N caméras button returns to the list. The group breaks apart on its own as soon as the zoom separates the points enough.

Events (OEDB)

Vaucluse events served by our static OpenEventDatabase instance: the oedb-rs repository (Rust) rebuilds an OEDB-compatible API on GitHub Pages every 3 hours, from the Bison Futé DATEX II feed filtered on Vaucluse (accidents, roadworks, jams) and manual curation (e.g. the Jeudis d'Orange night markets, category culture.market.night). Expired events are purged at every build.

How: enable the layer; click a legend row (Accidents, Roadworks, Jams, Culture, Others) to hide / show that category; click a marker for the label, dates, place and source.

Declassified sections

A departmental road can be handed back to the municipality — most town crossings have been. The section stays on the ground, its number disappears, and the road shown on the map ends up cut in two with nothing to say why. OpenStreetMap keeps the trace of it: 1,386 sections, 179 km, 92 former numbers.

How: enable the layer, then select a road. That is what the layer is for: its declassified sections take the colour of its network level while staying dashed, and the amputated axis reads whole again. On the D6, 191 sections and 15.5 km — more than the 10.6 km still classified: three disconnected pieces become one continuous axis. With no road selected, everything is apple-green dashes, as an inventory — a markedly yellow green no other layer could suggest, the project's greens all leaning towards emerald.

Two keys to read together, one of them a trap. was:ref carries the bulk of the declassifications surveyed in Vaucluse, old_ref a few more. But old_ref is also used, on 734 sections, to record the transfer of national roads to the department in 2006 — the opposite movement. Keeping only the values written as a D number separates the two cleanly, since a former national road is written N 7. A section that still carries a departmental number is dropped as well: it was renumbered, not declassified, and its course is already on the map.

These kilometres are added to nothing. They are no longer departmental: summing them would make every other figure wrong, and the sector digest as well as the network totals ignore them.

What the data is worth. A declassification recorded by contributors, with no official source alongside it and no date: the popup says so. Thirteen numbers have vanished from the departmental register entirely — the D202 has 7.7 km of former course for zero kilometres still classified — and for those the popup says so rather than offering a road to jump to.

Latest OSM changes

Recent contributions to the Vaucluse road network, following the principle of the OSM Latest Changes tool. A rolling seven days, on the order of a thousand sections created, modified or deleted depending on the week's activity.

How: enable the layer. Each section is coloured by what happened to it — green created, orange modified, red deleted (its last known geometry is then drawn). When an edit moved the geometry, the former one appears as a grey dashed line beside the new: the most direct way to see a bend straightened or a road realigned on imagery. Line weight follows the road's importance.

Filtering by road class. The legend offers one box per class — main roads (motorway, trunk, primary), secondary roads (secondary, tertiary), local access (residential, unclassified, service), tracks and active travel (track, path, footway, cycleway) and works (construction, proposed) — each with its count. Unchecking tracks and local access leaves only the structuring network, where the changes a road manager cares about happen.

The view summary. A box at the bottom right answers, for the current extent only, which roads were touched and by whom: the most edited roads (structuring axes first, then by number of changes), the age of the most recent change, and the contributors involved, with links to their OSM profiles. Clicking a road name frames the map on its changes and highlights them for a few seconds. A departmental road is grouped under its reference rather than the street names it takes through the municipalities.

Why this is not live. Overpass's augmented diff rebuilds the state of the whole extent at two dates and compares them: filtering on highway lightens what comes back, not the computation. Reckon on forty seconds for one week across the department. The query therefore runs every hour in a GitHub Actions workflow, and the browser only reads a 900 kB GeoJSON like any other layer.

Sensitive natural areas (ENS)

CD84 polygons (DataSud). Area, municipalities, manager.

How: enable the layer; click a polygon for the card.

iNaturalist observations

Observations filtered by intersection with the 84's sensitive natural areas.

How: enable the layer; click the coloured circle (enlarged click zone) for species, date, iNaturalist link. Works alongside the ENS layer shown at the same time.

Search a road

Search field (e.g. D900, D4) and a filtered list. Click a road: map centring, highlight, card in Road information.

OSM / Wikidata quality

Summary in the sidebar: Wikidata coverage, OSM relations. Detailed panel button: an on-map panel (metrics, roads without Wikidata, without relation).

See the N roads without Wikidata button: lists those roads in the sidebar, the most substantial first, with its number of OSM segments. Clicking a row follows that road on the map; opening the list shows all of them in red. Since the Wikidata identifier is carried by the OSM route relation, a road with no relation necessarily has no Wikidata: the relation has to be created first, and the list labels the opposite case separately — a relation that exists without carrying an identifier.

The percentage comes from the latest OSM snapshot, dated by the freshness badge just above it: a relation tagged after that extraction only counts at the next refresh.

Network information

Counters: number of roads, OSM segments, cumulative length, AADT min/max traffic, bridges, tunnels.

Refresh formula

Calendar by source (OSM, Wikidata, external, incubator, weather, static) and status of the last refresh. A row where one dataset has missed its turn turns amber and announces the age of the oldest.