Skip to content

Case Studies

URBADAPT-HEAT v1.0 ships configs and data manifests for 40+ European cities. Four are demonstrated end-to-end — Rome, Athens, Lisbon, and Copenhagen — spanning a broad range of climatic, demographic, and morphological conditions. The same city-agnostic notebooks run for every city; only the city-specific YAML config, data manifest, and the CITY selector differ.

City overview

City Country Climate type Population (FUA) Hazard track
Rome Italy Hot-summer Mediterranean (Csa) ~4.3 M A (standard)
Athens Greece Hot-summer Mediterranean (Csa) ~3.7 M A (standard)
Lisbon Portugal Warm-summer Mediterranean (Csb) ~2.9 M A (standard)
Copenhagen Denmark Oceanic (Cfb) ~1.3 M B (extreme-event)

This sample covers a Mediterranean heat-exposed cluster and a northern-latitude cool-climate city (Copenhagen), where heat mortality is lower in absolute terms but concentrated in rare extreme events.

City profiles

Rome

  • Strong urban heat island in the densely built historic core and periphery.
  • High share of older building stock (pre-1975), contributing to elevated thermal SVI.
  • AC penetration ~70% (2020 baseline); moderately steep income gradient.
  • Mediterranean climate produces long heat seasons; standard daily-mean T2M track appropriate.

Athens

  • Among the most heat-stressed European capitals.
  • Very high SVI persistence parameter (k = 0.90) — strong retention of local vulnerability patterns.
  • Highest new-build vulnerability anchor (0.18) reflecting building quality in peri-urban areas.
  • Lowest thermal retrofit rate (0.8% yr⁻¹) due to slower building renovation pace.
  • EWS modelled as a counterfactual (full benefit of establishing a comprehensive HHAP from scratch) rather than marginal — Greece lacks an effective heat-health action plan at baseline (archetype: weak_threshold).

Lisbon

  • Atlantic influence moderates heat compared to Rome/Athens but extreme events remain significant.
  • Lower SVI spatial persistence (φ₂₀₅₀ = 0.65) — faster convergence toward national mean over time.
  • High thermal retrofit rate (1.2% yr⁻¹), consistent with EU renovation wave policies.

Copenhagen

  • Track-B implementation: event-exceedance hazard and event-mask EWS activation.
  • Heat mortality is lower in absolute terms but concentrated in rare extreme events.
  • Higher SVI spatial smoothing convergence (φ₂₀₅₀ = 0.60) reflecting lower long-run spatial inequality.
  • Lowest SVI persistence (k = 0.70), consistent with strong Danish social welfare state.
  • Cross-city comparisons prioritise within-city policy deltas and normalised indicators over absolute trigger-based metrics (warning-day counts are not directly comparable to Track-A cities).

Key results summary

Heat hazard trajectory

  • Annual mean T2M rises ~0.4–0.6°C per decade under the central scenario.
  • The share of FUA grid cells exceeding the 2020 P90 threshold grows from ~10% to >40% by 2050, substantially expanding the spatial footprint of intense heat exposure.

Policy cost-effectiveness

Policy Key finding
EWS Most cost-effective per avoided death; dominates at low budget levels due to negligible operational expenditure
AC Avoids the most deaths in gross terms at scale; dominated by electricity costs; income-targeted policy more equitable than uniform rollout
Trees Intermediate cost-effectiveness; equity-weighted variant avoids more deaths than uniform at identical investment; important co-benefits in cooling, biodiversity, urban amenity

Combined scenarios

  • The vegetation–AC electricity co-benefit (trees reduce AC energy demand) is an important interaction term in high-greening combined scenarios.
  • EWS benefits are slightly attenuated at high AC penetration due to the AC–EWS overlap correction.
  • All three policies lie on or near the cost-effectiveness Pareto frontier.

Equity

  • Income-targeted AC catch-up delivers disproportionately larger mortality benefits in high-SVI areas compared to uniform rollout, even at equal total user count.
  • Equity-weighted GVI allocation (street trees) consistently avoids more deaths than the standard Q3 catch-up at identical investment.

Reproducing the case study results

Open the city-agnostic template notebooks and set the city with one line near the top of NB01:

urban-heat/notebooks/city_agnostic/March2026_agnostic/template/

python os.environ["CITY"] = "Rome" # or Athens / Lisbon / Copenhagen

Then run 01 → 10 in order. Ready-made per-city copies with CITY pre-set are also provided alongside the template (Rome/, Athens/, Lisbon/, Copenhagen/, as *_<City>.ipynb). Results are written under urban-heat/outputs_variants/<variant>/<city>/.

See City Configuration for the full list of adjustable parameters and Installation & Usage for the run walkthrough.

This page is maintained in the URBADAPT-HEAT wiki and synced automatically. Edit it there, not in the website repository.