Regenerative Microclimate Scorecard, Marseille

Vieux-Port · ~800 m study area · baseline vs. tree-canopy intervention

AT6012 infrared.city SDK 0.4.9 UTCI · Wind · Solar ~800 m site live API result

↔ Compare: Cork (Nano Nagle Place) ← Learning module

An enlarged Vieux-Port study area (~800 m, replacing the earlier ~290 m pilot) gives a far richer microclimate field for a proper case study. The same workflow runs three analyses over the block and compares a baseline against a tree-canopy intervention.

Workflow

6,347

buildings (context geometry)

2,194

trees (intervention layer)

~800 m

study area (≈24 tiles/scenario)

Marseille microclimate scorecard: baseline vs intervention across UTCI, wind, and solar — Rows: baseline (buildings only) vs. intervention (+ tree canopy). Columns: UTCI thermal comfort, wind speed, direct sun hours, Vieux-Port, Marseille.

Thermal comfort (UTCI)

Metric	Baseline	+ Trees	Change
Mean UTCI	27.3°C	27.5°C	+0.2°C
Comfortable area (9,26°C)	20.9%	17.0%	−3.9 pts
Heat-stress area (UTCI>26°C)	79.1%	83.0%	+3.9 pts
Thermal-zone diversity (Shannon)	1.03	0.94	−8.7%

A counterintuitive result, and why it matters. Over this larger, denser block the canopy intervention raised modeled heat-stress area by 3.9 points rather than lowering it. Two factors explain it, and both are good teaching points:

Coverage shift (measurement artifact). Tree canopy masks cells out of the UTCI grid (the intervention panel resolves fewer cells), so the percentages are not computed over an identical cell set , the comparison is not strictly like-for-like.
Plausible physics. In dense Mediterranean street canyons, canopy can reduce sky-view and ventilation, trapping daytime radiant heat, cooling is not guaranteed by simply adding trees.

The honest conclusion: "add trees" is not automatically regenerative; placement, density and street geometry decide the outcome, exactly the evidence-based judgement the module asks students to make.

Wind & Solar

Metric	Baseline	+ Trees
Mean wind speed	2.65 m/s	2.65 m/s
Mean direct sun hours	9.14 h	9.35 h

Method

One run_area_and_wait() call per analysis over the ~800 m Vieux-Port polygon. Buildings and weather are injected for all runs (thermal-comfort requires geometry); vegetation is injected only for the intervention. Outputs are merged NumPy grids scored by regenerative_metrics.py (UTCI bands, heat-stress %, Alberti assembly index). Weather: nearest EPW station (Marseille-Provence AP TMYx).