Hazard Component¶
The heat hazard is represented as daily mean 2 m air temperature (T2M) fields on the UrbClim reference grid (~100 m, EPSG:3035).
Historical data¶
| Item | Detail |
|---|---|
| Source | UrbClim via the PROVIDE/VITO API |
| Period | 2008–2017 (daily NetCDF files) |
| Variable | Daily mean 2 m air temperature (Kelvin) |
| Resolution | ~100 m, EPSG:3035 |
2020 synthetic baseline¶
The 2020 baseline is a synthetic climatological year derived from the historical archive — not a single observed year. The default construction uses the month-and-day climatological mean of the 2008–2017 period for each calendar day, producing a complete year that captures the city's characteristic seasonal cycle while suppressing interannual variability.
Alternative baseline constructions supported by the framework (selectable via city YAML config):
| Config value | Description |
|---|---|
climatology_mean (default) |
Month/day mean over 2008–2017 |
pixelwise_doy_max |
Pixelwise month/day maxima (extreme sensitivity) |
domain_peak_day |
Daily sequence from the year containing the hottest domain-mean day |
warmest_summer |
Daily sequence from the year with highest JJA domain-mean temperature |
The active mode is set via climate.t2m_baseline_mode_default in the city YAML config.
The chosen mode is recorded in output manifests for full reproducibility.
Future hazard (2030 · 2040 · 2050)¶
Future fields are generated by adding monthly mean warming deltas to the synthetic baseline.
- Deltas come from a CMIP6 GCM ensemble obtained through the PROVIDE climate service.
- Four emissions scenarios: CurPol (current policies), GS (gradual stabilisation), SP (sustainable pathway), SSP5-8.5 (high emissions).
- Each GCM contributes one representative delta (average of its 45th and 55th percentile warming).
- Three uncertainty bands: Low (P25) · Central (P50) · High (P75) across models.
- Monthly deltas are distributed uniformly to daily calendar days and added to the baseline.
- All fields are masked to the FUA boundary before storage.
CLIMADA Hazard object¶
The CLIMADA Hazard object stacks all daily T2M fields across all target years into a sparse intensity matrix:
| Attribute | Value |
|---|---|
| Type identifier | T2M |
| Intensity units | °C |
| Event frequency | 1 / N_days_year (corrected for leap years) |
| Rows | One per calendar day × target year |
| Columns | One per populated FUA grid cell |
Event frequencies are set so that frequencies sum to exactly 1 within each target year, ensuring annual mortality totals aggregate correctly across the full daily series.
The climate scenario is selected via climate.t2m_clim_scen_default; the GCM uncertainty band (low/central/high) via climate.t2m_delta_pct_band_default in the city YAML.
Hazard tracks: A (standard) and B (extreme-event)¶
The framework supports two hazard representations, selected per city via the extreme_hazard: config block and read at runtime through HAZARD_TRACK:
- Track A — standard daily-mean. The default. Every day of the target year is one CLIMADA event on the daily-mean
T2Mfield (described above). Used for Rome, Athens, Lisbon, and most configured cities. - Track B — extreme-event (heatwave). For cities where daily-mean T2M carries limited heat signal (e.g. Copenhagen). Hazard events are heatwave days exceeding a local percentile threshold of the historical distribution for a minimum run length.
Track-B event definition (Copenhagen defaults)¶
| Parameter | Config key | Default | Options |
|---|---|---|---|
| Warning/event season | event_track.season |
May 15 – Sep 30 | configurable |
| Exceedance percentile | threshold_percentile_default |
95 | 95 / 93 / 92 / 90 |
| Minimum duration | min_duration_default |
3 days | 3 / 2 |
| Intensity method | intensity_method |
exceedance |
— |
| Impact-function reference | t_ref_mode_default |
local_p90 |
local_p90 / fixed_20c |
Track B concentrates model sensitivity on the tail of the distribution while preserving the probabilistic risk chain, uses a cold-city impact-function variant (cold_city_if_variant: true), and changes the EWS activation to an event-mask fallback when the deaths-based threshold cannot be calibrated (see Adaptation Pathways). A city may run both tracks (run_standard_track / run_extreme_track); for Copenhagen the standard track is kept for comparability with policies disabled, and the extreme track is the policy-analysis track.
Input summary¶
| Input / Parameter | Source | Notes |
|---|---|---|
| Historical daily T2M | UrbClim via PROVIDE/VITO API | 2008–2017; 100 m; EPSG:3035 |
| Spatial domain | GHS-FUA boundary | Defines valid raster extent |
| Reference grid | UrbClim native grid | 100 m, EPSG:3035 |
| 2020 baseline | Month/day climatological mean | Synthetic |
| Climate warming deltas | PROVIDE CMIP6 ensemble | Monthly, per scenario and year |
| Climate scenarios | CurPol, GS, SP, SSP5-8.5 | Four CMIP6 pathways |
| Climate uncertainty | P25 / P50 / P75 across GCMs | Low / Central / High |
| Target years | 2020, 2030, 2040, 2050 | 2020 synthetic; future delta-scaled |
This page is maintained in the URBADAPT-HEAT wiki and synced automatically. Edit it there, not in the website repository.