S1: Flood Risk Assessment — Toulouse

Intermediate CLI + Map

Capabilities: filter

Use Case

A city planner needs a flood risk diagnostic along the Garonne in Toulouse: isolate the low-rise buildings (≤ 15 m) inside the 250 m corridor of the Garonne AND whose ground sits 0–15 m above water level. All in a single GISPulse pipeline.

IGN BD TOPO V3 Data

Layer	Content	Features	Source
batiments	IGN buildings (height, floors, dwellings, usage)	~31,000	data.geopf.fr — BDTOPO_V3:batiment
surfaces_eau	Hydrographic surfaces (Garonne, canals, basins)	43	data.geopf.fr — BDTOPO_V3:surface_hydrographique
cours_eau	Waterways (Garonne, canals, tributaries)	14	data.geopf.fr — BDTOPO_V3:cours_d_eau

# Download BD TOPO data for Toulouse
python examples/prepare_playground_data.py --city toulouse

# Inspect the dataset
gispulse info examples/datasets/toulouse_bdtopo.gpkg

Pipeline (4 steps)

cours_eau ──► filter_hydro (toponyme in ['la Garonne','Bras Inférieur Garonne'])   # narrow watercourses to the Garonne corridor

batiments ──► filter_in_flood_zone (intersects filter_hydro, buffer 250 m, L93)    # 250 m corridor
                │
                ▼
            filter_in_flood_altitude (altitude_minimale_sol in [134, 149] m IGN69) # ground 0-15 m above Garonne level
                │
                ▼
            filter_low_buildings (hauteur in ]0, 15] m)                            # low/mid-rise stock — most vulnerable

Metric CRS

IGN data is stored in EPSG:4326 (WGS84 degrees). For the 250 m buffer in filter_in_flood_zone we reproject to EPSG:2154 (Lambert93) — the default EPSG:3857 (Web Mercator) distorts distances by ~38 % at Toulouse's latitude (factor 1/cos(43.6°)).

Why a dedicated filter_hydro step?

BD TOPO cours_eau holds 14 lines: the Garonne and its lower branch, but also water-supply canals (Saint-Martory, Canal du Midi, Canal Latéral, Canal de Brienne) and tributaries (Hers Mort, Girou, Sausse, Riou Gras) that don't carry the same flood risk. A 250 m buffer around all of them flagged buildings west of the city near a supply canal (not floodable). The filter_hydro step narrows cours_eau to the Garonne corridor first, then filter_in_flood_zone uses that subset as its ref_layer.

Altitude: no external DTM required

BD TOPO V3 already carries per-building Z values: altitude_minimale_sol, altitude_maximale_sol, altitude_minimale_toit, altitude_maximale_toit (meters, IGN69 datum). No need to drape an external DEM (RGEALTI 1 m) over the footprints for this layer.

Garonne reference in Toulouse: ~134 m IGN69 at Pont-Neuf. The historical 1875 flood peaked around 142 m (~8 m above normal level). The filter altitude_minimale_sol BETWEEN 134 AND 149 keeps buildings whose ground sits 0–15 m above water level — the upper bound is intentionally generous to cover a centennial flood plus margin.

For a finer diagnostic (upstream/downstream gradient), replace the constant 134 with a spatial join on surfaces_eau Z, or sample a DEM raster (raster_sample capability).

Why cap building height at ≤ 15 m?

Above 15 m (~5 floors) occupants have refuge floors; low buildings (R+0 to R+3) concentrate human and material exposure during a major flood. Capping at 15 m isolates the most vulnerable cohort.

Rules

{
  "version": 2,
  "name": "flood_risk_diagnostic",
  "ref_layers": {
    "cours_eau": "cours_eau",
    "surfaces_eau": "surfaces_eau",
    "batiments": "batiments"
  },
  "steps": [
    {
      "id": "filter_hydro",
      "type": "capability",
      "capability": "filter",
      "params": {
        "expression": "toponyme in ['la Garonne', 'Bras Inférieur Garonne']",
        "crs_meters": "EPSG:2154"
      },
      "input": "cours_eau"
    },
    {
      "id": "filter_in_flood_zone",
      "type": "capability",
      "capability": "filter",
      "params": {
        "spatial_predicate": "intersects",
        "ref_layer": "filter_hydro",
        "buffer_distance": 250,
        "crs_meters": "EPSG:2154"
      }
    },
    {
      "id": "filter_in_flood_altitude",
      "type": "capability",
      "capability": "filter",
      "params": {
        "expression": "altitude_minimale_sol >= 134 and altitude_minimale_sol <= 149"
      },
      "input": "filter_in_flood_zone"
    },
    {
      "id": "filter_low_buildings",
      "type": "capability",
      "capability": "filter",
      "params": {
        "expression": "hauteur > 0 and hauteur <= 15"
      },
      "input": "filter_in_flood_altitude"
    }
  ]
}

Download

scenario-1-rules.json

Execution

gispulse run examples/datasets/toulouse_bdtopo.gpkg \
  --layer batiments \
  --rules playground/scenario-1-rules.json \
  -o output/flood_diagnostic.gpkg \
  --ref-source cours_eau:examples/datasets/toulouse_bdtopo.gpkg:cours_eau

# View result on map
gispulse serve output/flood_diagnostic.gpkg

Expected Result

Output schema

Column	Type	Source	Description
geometry	MultiPolygon	source	Building geometry
usage_1	string	source	Main usage (Residential, Industrial...)
hauteur	float	source	Building height (m) — filtered to ]0, 15] m
altitude_minimale_sol	float	source	Min ground altitude (m IGN69) — filtered to [134, 149] m
altitude_maximale_toit	float	source	Max roof altitude (m IGN69)
nombre_d_etages	int	source	Number of floors
nombre_de_logements	int	source	Number of dwellings

From the ~31,000 Toulouse buildings, the pipeline retains those matching the three criteria: (1) within 250 m of the Garonne, (2) ground 0–15 m above Garonne level, (3) low/mid-rise (≤ 15 m). This is the cohort most exposed to a major 1875-class flood. To compute the footprint, chain an area_length step on the output of filter_low_buildings.

Full interactive playground

Live 4-step pipeline (requires the demo backend):

Step by step:

1. filter_hydro (blue) — narrows cours_eau to the Garonne corridor
2. filter_in_flood_zone (orange) — buildings within 250 m, Lambert93
3. filter_in_flood_altitude (yellow) — ground between 134 and 149 m IGN69
4. filter_low_buildings (red) — height ≤ 15 m, most vulnerable cohort

Interactions:

• Building popup: usage, height, ground altitude, dwellings, floors
• Each step colours the cohort reached after its filter

Try it live

Live DemoDemo server status

GET/health

Live DemoAvailable datasets

GET/datasets

Next Steps

• S2: Commercial Buildings along Arterials — same city, 2-step filter + ref_filter pattern
• S6: Residential Real Estate — per-dwelling metrics in Versailles
• Vector capabilities — filter with ref_layer, ref_filter, buffer_distance