geo: sample seven cross-sections from the DEM

Gutters sit at -10 m south and +10..12 m north, a ~2% crown between them, and
the envelope runs 0.6-1.0 m per section. Gutter spacing independently confirms
the 20.9 m carriageway measured from the crossing. The sharpest breaks are
0.09-0.12 m/m, so 3DEP does not resolve a kerb upstand: kerb height stays a
recorded assumption rather than a derived value.

Geo/verification/2026-11-06-cross-sections.md

@@ -0,0 +1,47 @@
+# Cross-sections — crowned road, gutters, sidewalk shelves
+
+Sampled 2026-11-06 from the local 3DEP DEM (`Geo/raw/dem-3dep-1m-26911-2026-11-06.tif`) with `Geo/scripts/street_section.py`. Seven sections at 40 m spacing, offsets −18 m to +18 m from the boulevard centreline (negative = south), 1 m sampling step.
+
+## What the profile shows
+
+Every section has the same shape:
+
+- a local low at about **−10 m** (south gutter),
+- a rise to a flat-ish crown around **+2 to +8 m**,
+- a second low at about **+10 to +12 m** (north gutter),
+- then the north side climbing to the building line.
+
+Those gutter positions bracket a carriageway of roughly **20–22 m**, which independently matches the 20.9 m measured from the mid-block crossing in [`2026-11-06-street-profile.md`](2026-11-06-street-profile.md). Two unrelated measurements agreeing is the strongest evidence in the geo work so far.
+
+## Section table
+
+| Along block | Lowest point | Road crown | Cross-fall across the road | Gutter candidates |
+| ---: | ---: | ---: | ---: | --- |
+| 0.0 m | 118.78 @ −10 | ~119.41 @ 0 | ~0.6 m / 2 % | −10, +10 |
+| 40.0 m | 118.43 @ −11 | ~118.94 @ 0 | ~0.5 m / 2 % | −11, +10.5 |
+| 80.0 m | 117.84 @ −10 | ~118.33 @ 0 | ~0.5 m / 2 % | −10, +10.5–11.5 |
+| 120.0 m | 117.22 @ −10 | ~117.73 @ 0 | ~0.5 m / 2 % | −10, +11.5 |
+| 160.0 m | 116.55 @ −10 | ~117.04 @ 0 | ~0.5 m / 2 % | −10, +11.5 |
+| 200.0 m | 115.96 @ −10 | ~116.37 @ 0 | ~0.4 m / 2 % | −10, +10.5–11.5 |
+| 240.0 m | 115.63 @ −10 | ~116.02 @ 0 | ~0.4 m / 2 % | −10, +10.5–11.5 |
+
+Total envelope from the lowest south gutter to the highest north point is 0.6–1.0 m per section, depending on how far the north sidewalk has risen.
+
+## Kerbs are not in the DEM
+
+The sharpest breaks found at the expected kerb offsets are **0.09–0.12 m/m** slopes — that is a gutter transition spread over a metre, not a kerb upstand. 3DEP is bare-earth and smooths a 0.15 m vertical step at 1 m sampling, so the DEM cannot give kerb height.
+
+Consequence: kerb height and shape are an **assumption** until measured or observed in the reference photography. Write the assumption down where it is used rather than burying it in the mesh. Do not infer a kerb from the DEM's gradient.
+
+## What the demo ground mesh must do
+
+- Crown the road at roughly 2 % from each gutter to the centre, not a flat plane and not a constant tilt.
+- Step down into the gutters at about ±10.5 m from the centreline, then run the sidewalks at their own grade.
+- The north sidewalk sits higher than the road crown and keeps rising toward the building line, so the two sidewalks are not mirror images.
+- Crossing ramps must be modelled deliberately at the three crossings; nothing in the DEM supplies the local ramp geometry.
+
+## Open
+
+- Kerb height, kerb radius, and ramp slope: assumption until surveyed or measured from photographs.
+- The 200 m section peaks at +14 m rather than +18 m, which may be a tree canopy or a building edge in the DEM. Do not treat it as a road feature.
+- Drainage inlets, driveway aprons, and pavement joints are not in this data.

Geo/verification/2026-11-06-elevation.md

@@ -52,5 +52,5 @@ Window statistics: min 110.46 m, max 127.73 m, spread 17.28 m — building tops
 
 ## Next
 
-1. Sample the kerb line and the crossing ramps specifically — the grid above is a street-level average, not a kerb profile.
+1. ~~Sample the kerb line and the crossing ramps~~ — done: seven cross-sections in [`2026-11-06-cross-sections.md`](2026-11-06-cross-sections.md) locate the gutters at ±10–11 m and show a ~2 % crown. They also show the DEM does **not** resolve a kerb upstand, so kerb height stays an assumption.
 2. Build the ground mesh from the DEM along the street, not from a flat plane, and assert the 3.94 m fall survives the Blender→Unreal path.