Australia
Australia's largest city, built around a natural harbour on the Pacific coast and ringed by eucalypt bushland that regularly burns each summer.
Overall hazard score
25/100
Moderate
Equal-weighted mean of the 6 indicators below, each on a fixed absolute scale.
Each axis is that hazard's score on its own fixed 0-100 scale. Adding more cities to this site never changes these values.
Loading map…
Air quality is published as a single figure per city, so it adds the same value to every cell. The variation on this map comes from the hazards that are mapped at grid resolution.
Each hexagon is about 0.74 km². Hover or tap one for its numbers, or switch the layer above to see a single hazard.
Expected annual inundation depth from river and coastal flooding combined, population-weighted across the city. From JRC CEMS-GloFAS river maps at 90 m (seven return periods) and WRI Aqueduct coastal maps (six). Both models are undefended: they simulate water without levees, sea walls or pumping.
Measured value: 0.002194 m/year
Distance-decayed, magnitude-weighted rate of observed earthquakes around the city, from the USGS ANSS catalogue.
Measured value: 0.03601 index/year
Annual rate of tropical cyclone passages near the city, weighted by the sustained wind speed at closest approach. From NOAA IBTrACS, 1980 onwards.
Measured value: 0 events/year
Annual satellite fire-detection density in the vegetated ring around the city, from NASA FIRMS VIIRS 375 m.
Measured value: 32.07 detections/1000km²/year
Maximum temperature of the warmest month (BIO5, 1981-2010), averaged across the city. An absolute measure, comparable between climate zones. It does not account for humidity, so it understates the danger of humid tropical heat.
Measured value: 25.62 °C
Population-weighted annual mean PM2.5 concentration. A chronic hazard rather than an episodic one - it harms through years of exposure, not in a single event - but by mortality it is the largest hazard on this site.
Measured value: 12 µg/m³
Context, not hazard. None of these figures affects the score above - a sunnier or greener city is not a safer one, and mixing the two would quietly change what the number means.
Solar potential
4.17kWh/kWp per day
Photovoltaic output for a standard panel. A resource, not a hazard.
World Bank Global Solar Atlas 2.0
Mean temperature
16.4°C
Annual mean, 2010s. Shown for context; the heat score measures departure from local normal, not absolute warmth.
GHS-UCDB climate reanalysis
Greenness
0.45NDVI
Mean vegetation index across the built-up area. Higher values moderate urban heat.
Landsat via GHS-UCDB
Hospitals
3.6per 100,000 people
Mapped health facilities per head. A rough proxy for capacity to respond, and dependent on how completely the country is mapped.
healthsites.io via GHS-UCDB
Average elevation
48m
Mean ground height across the urban centre.
Copernicus GLO-90
The flood models simulate water without levees, sea walls or pumping, because global defence data does not exist at usable quality. Sydney's flood score therefore describes the exposure of the ground, not what remains after the engineering built on top of it.
The earthquake score counts shaking that has actually happened since 1970. It cannot see a fault that has been locked and silent for centuries, so it understates places like Istanbul whose danger lies in a quake that has not arrived yet.
Every indicator is mapped onto a fixed scale chosen in advance. A score of 25 means the same thing today as it will when this site covers a hundred cities.
These are city-scale estimates on a roughly one-kilometre grid. They say nothing reliable about an individual building, and nothing at all about what any person should do.
The JRC publishes its own figures for these urban centres, computed independently from several of the same datasets. Every build compares our numbers against theirs and fails if the gap is too wide to explain. The result is printed here rather than kept private, because a score you cannot audit is not worth much.
| Quantity | Ours | Reference | Gap | Status |
|---|---|---|---|---|
Urban centre area
Same polygon measured twice. Disagreement means our projection or area arithmetic is wrong.
GHS-UCDB · GC_UCA_KM2_2025
|
1,374.50 | 1,378.00 | 0% | agrees |
Population
Two different global population products (WorldPop against GHS-POP). They routinely differ by 10-20% for the same city.
GHS-POP via GHS-UCDB · GC_POP_TOT_2020
|
2,789,522.00 | 4,375,394.50 | 36% | warn |
Population in the 1-in-100-year river flood plain
Same flood model, independently aggregated. This is the check that would have caught the 90 m/1 km resolution bug.
Baugh et al. (2024) via GHS-UCDB · EX_100_SHP_2020
|
1.27 | 0.48 | 165% | agrees |
Population in the 1-in-10-year river flood plain
Same flood model, independently aggregated. This is the check that would have caught the 90 m/1 km resolution bug.
Baugh et al. (2024) via GHS-UCDB · EX_010_SHP_2020
|
0.90 | 0.17 | 429% | agrees |
Population exposed to coastal flood
Investigated 2026-08-22: this gap is a model difference, not an aggregation bug. Aqueduct's historical coastal layer puts almost no water over Tokyo Bay (101 pixels above 1 cm in the whole city window), while UCDB's two coastal scenarios track the low-elevation coastal zone closely (CF2 8.9% against 8.2% of population below 5 m; CF1 21.0% against 17.2% below 10 m). Aqueduct appears to understate enclosed bays. Aqueduct is kept because it ships rasters and UCDB publishes only city totals, but a better coastal source is an open task.
Vousdoukas et al. (2018) via GHS-UCDB · EX_CF2_SHP_2020
|
0.00 | 0.01 | 100% | warn |
Cyclone wind exposure
Not the same quantity, so no ratio is computed: we count storm passages within 200 km, UCDB models whether category-1 wind actually reaches the built-up area. Tokyo reads 24 near-passages against 0% population in category-1 wind, which is consistent - typhoons pass close often, and hurricane-force wind rarely lands on the city. Shown side by side so a qualitative contradiction would be visible.
GHS-UCDB cyclone wind exposure · EX_100_S1P_2020
|
0.00 | 0.00 | n/a | agrees |
Units differ by row; each is given in the description. A gap of 10-20% is normal - the two sides use different population grids and different aggregation - and the tolerances are set accordingly.
Generated 2026-08-22T11:44:15+00:00 from the underlying data.
This site is free, asks before it tracks anything, and is built entirely on public data. Hosting and the data pipeline cost money. If it was useful to you, a small contribution keeps it online and pays for adding the next city.