Tool 09 · Map · 1000+ cities

Global Temperature Map — Calibrated Annual Mean Temperatures for 1000+ Cities

An interactive world temperature map covering 1000+ cities across 190 countries. From Doha's 27°C tropical heat to Reykjavik's 5°C cold — every annual figure is the CHELSA v2.1 climatological mean, calibrated against 219 WMO reference stations and reported with documented ±1–2°C accuracy. The Temperature layer also has a monthly Daytime high · 1km mode for a sharper, present-day read.

Warmest capital-class
27°C
Doha — annual mean
Coldest in dataset
5°C
Reykjavik — annual mean
Accuracy
±1–2°C
CHELSA v2.1 · 219 WMO stations · 1981–2010

How global temperature is calibrated

Most "world temperature map" results online return either a raw satellite layer with no station tie-in, or a Wikipedia table whose rows mix decades, station heights, and shelter types. GeoRank publishes one consistent dataset and one consistent reference period. We use CHELSA v2.1 monthly mean temperature (tas) — a terrain-informed downscaling built at ~1 km native resolution and processed to a 0.05° (~5.5 km) grid for the map — then average the twelve months of the 1981–2010 climatological normals to get one annual mean per cell.

CHELSA downscales coarse ERA5 reanalysis using dynamically-calculated lapse rates rather than one fixed global rate, so it resolves valley floors, coastal towns, and mountain ridges that a raw ~28 km reanalysis cell would flatten into one number — typical accuracy is ±1–2°C against ground stations in well-instrumented regions. We cross-check our output against 219 WMO reference stations and the national met service archives (NOAA in the US, the Met Office in the UK, DWD in Germany) to set the published accuracy bounds. The standard caveats — altitude, coastal land-sea contrast, urban heat islands — are documented in the methodology and surfaced in the layer passport below.

🌡
Temperature — annual mean
Calibrated
●●●●
SourceCHELSA v2.1 monthly mean temp (tas)
Reference period1981–2010 climatology
Native grid~1 km terrain-informed downscaling
Reporting grid0.05° (~5.5 km) map grid
Coverage1000+ cities · 190 countries
Accuracy±1–2°C typical · ±3°C mountain/coast
Urban heat islands are not modelled — values represent the mean of the 0.05° grid cell, not the downtown core. The 56-station cross-check sample is drawn from a 219-station WMO reference set documented in the methodology.
Full methodology →

Daytime high · 1km — a second read on the same map

The Temperature layer now has a second mode. Flip the toggle from the annual mean to Daytime high · 1km and every pin switches to the monthly average daily maximum — the number that tells you what a June afternoon actually feels like, not what the whole year averages out to. It's built from CHELSA v2.1 tasmax climatology (1981–2010) at CHELSA's native ~1 km resolution, sharper than the map's standard 0.05° (~5.5 km) mean grid — sharp enough to keep a waterfront pin and the hills a few kilometres inland distinct instead of blended into one value.

GeoRank map with the Daytime high · 1km toggle active on the Temperature layer, showing a pin at L'Isuledda near Olbia, Sardinia reading a June daytime high of 26.8°C

Because the 1981–2010 baseline is three decades old, Daytime high adds a warming uplift on top: the per-month change in ERA5 near-surface temperature between 2011–2025 and 1981–2010, resampled onto the 1 km grid. Globally that adds +0.4 to +0.9°C by month; fast-warming Mediterranean and continental cells run +1.2 to +1.6°C higher, while ocean-damped coasts warm less — the correction is spatially variable, not a flat offset across the map.

City July mean July daytime high Uplift
Cefalù26.6°C27.8°C+1.2°C
Rome30.0°C31.4°C+1.4°C
Athens33.1°C34.2°C+1.1°C
Dubai40.7°C41.9°C+1.2°C
London21.5°C22.2°C+0.7°C
🌡
Temperature — Daytime high · 1km
Calibrated
●●●●
SourceCHELSA v2.1 tasmax + ERA5 uplift
Baseline1981–2010 · uplifted to 2011–2025
Native grid~1 km (0.0083°)
Map zoomRenders to z7 · mean layer stops at z6
CoverageMonthly only — no annual daytime high
Compare tableAdds "Daytime high (warmest/coldest)" rows
The warming uplift is a smooth 0.5° ERA5 field resampled onto the 1 km CHELSA grid, not a new observational climatology — treat it as a present-day-adjusted estimate, documented the same way as the ERA5 + SARAH-3 sunshine blend.
Full methodology →

Top 10 hottest cities worldwide (annual mean, calibrated)

These are the warmest major cities in the GeoRank dataset, ranked by calibrated annual mean 2-metre air temperature. Tropical capitals near the equator and oil-economy Gulf cities dominate the list. For affordable warm relocation options, the same map layers cost on top — try the cheapest countries to live ranking for the price-adjusted version of this question.

🌡 Top 10 hottest cities (annual mean)
  • 1Bangkok, Thailand28.5°C
  • 2Manila, Philippines27.8°C
  • 3Dubai, UAE27.6°C
  • 4Doha, Qatar27.1°C
  • 5Singapore26.7°C
  • 6Mumbai, India26.6°C
  • 7Miami, Florida (US)24.5°C
  • 8Phoenix, Arizona (US)23.3°C
  • 9Cairo, Egypt22.0°C
  • 10Brisbane, Australia20.1°C

Top 10 coldest capitals (annual mean, calibrated)

❄ Top 10 coldest capitals (annual mean)
  • 1Reykjavik, Iceland5°C
  • 2Helsinki, Finland6°C
  • 3Oslo, Norway7°C
  • 4Stockholm, Sweden8°C
  • 5Tallinn, Estonia6°C
  • 6Riga, Latvia7°C
  • 7Vilnius, Lithuania7°C
  • 8Moscow, Russia6°C
  • 9Ottawa, Canada6°C
  • 10Berlin, Germany9°C

Mediterranean & mild-climate benchmark cities

City Country Annual mean Notes
LisbonPortugal16.3°CAtlantic-tempered, 38°N
AthensGreece18.1°CMediterranean basin, 38°N
MadridSpain14.8°CContinental plateau, 40°N
Mexico CityMexico16.1°CHigh altitude 2,240 m, 19°N
ParisFrance11.2°COceanic, 49°N
LondonUK10.6°COceanic, 51°N

How to use the global temperature map

The interactive GeoRank map renders the calibrated temperature layer as a global orange gradient — the darker the swatch, the warmer the annual mean. Pan, zoom, and click any pin for the annual figure plus the monthly profile. The same view can stack cost, tax, sunshine, and safety, which is the real relocation question: not just "where's warm" but "where's warm and affordable and visa-accessible." The companion World Sunshine Map uses the exact same calibration discipline for sun hours.

01
Open the temperature layer
Deep-link straight into the temp layer at /?layer=temp. The orange gradient is the calibrated annual-mean field. Click any city pin to inspect the monthly profile.
02
Pin your home as baseline
Set "home" once and every other pin shows the delta from where you live now — temperature, sun, cost, all sourced and calibrated. Mexico City vs Bangkok, Lisbon vs Madrid, all instant.
03
Stack cost, tax, safety overlays
Temperature is one of nine data layers. Stack cost, safety, and tax on top — the colour-coded chips on every pin tell you the trade-off at a glance.
04
Save, share, decide
Save shortlists, share a pinned URL with a partner, or open the result in Compare for the full side-by-side. Free; no signup; the calibration formulas are public.

Temperature by climate zone

Annual mean temperature alone doesn't tell you everything — a coastal Mediterranean city at 16°C and a high-altitude tropical city at 16°C feel completely different in July. Use this rough zone mapping as the mental model before drilling into specific cities. For more granular monthly data, the climate & rain layer breaks each city into its full twelve-month profile.

Zone Example city Annual mean Character
TropicalBangkok, Thailand28.5°CHot & humid year-round, monsoon May–Oct
Hot aridCairo, Egypt22.0°CDry heat, <30 mm annual rain
MediterraneanLisbon, Portugal16.3°CMild wet winters, dry warm summers
High-altitude tropicalMexico City, Mexico16.1°CEternal spring; 6.5°C/km lapse offsets latitude
Oceanic temperateLondon, UK10.6°CMild damp, narrow seasonal swing
Continental coldBerlin, Germany9.4°CCold winters, warm summers, wider swing
SubarcticReykjavik, Iceland4.7°CCool maritime, short cool summer

Why GeoRank's temperature map is different

01
One reference period, named sources
Every value uses the 1981–2010 CHELSA v2.1 climatological normals — one consistent 30-year reference window across the whole map. Source named on every cell. No "various" rows, no undated Wikipedia means. Methodology link on every chart.
02
Published accuracy bounds
±1–2°C typical, ±3°C in mountain or tight coastal terrain. Most temperature maps don't publish a single error bar; ours sits on the layer passport above. Cross-checked against 219 WMO stations and national met archives (NOAA, Met Office UK, DWD).
03
Set home, read deltas
Climate numbers in isolation are noise. Pin London at 11°C as home and Lisbon shows +6°C, Bangkok +18°C, Reykjavik −6°C. The whole point of the tools suite is making destination numbers comparable against your current life.
04
Climate is one layer, not the answer
Temperature alone doesn't pick a city. The same map shows cost, tax, air quality, and safety — try the comparisons hub for the bigger trade-off picture. Free, no signup, all data sources cited.

Related GeoRank tools & rankings

Frequently asked questions

How accurate are these temperatures?
±1–2°C for most locations after calibration against 219 WMO reference stations. Mountainous terrain and tight coastal microclimates widen to ±3°C because even the CHELSA 0.05° (~5.5 km) map grid can't fully resolve local lapse rates or land-sea contrasts. Each value represents the mean inside the grid cell, not a single weather station. Full bounds in the methodology.
How is annual mean temperature different from monthly average?
Annual mean is the simple arithmetic average of twelve monthly means across the 1981–2010 CHELSA climatology. It collapses summer and winter into one number — Lisbon 16°C, Madrid 15°C, Berlin 9°C — which is the most stable single comparator for relocation. Monthly means matter once you've shortlisted; the live map shows both, and the climate layer breaks out the full annual profile.
How is Daytime high different from the annual mean?
The annual mean is a single yearly average of day and night temperatures. Daytime high swaps that for the monthly average daily maximum (CHELSA v2.1 tasmax, 1981–2010 climatology), rendered at CHELSA's native ~1 km resolution instead of the map's standard 0.05° mean grid, plus a small warming uplift (ERA5 2011–2025 minus 1981–2010) so the number reads closer to a present-day month. It's monthly only — there's no annual Daytime high figure — and the map renders it out to zoom 7, one level past the mean layer's zoom-6 ceiling. Full breakdown in the methodology.
Why is Mexico City 16°C when it sits in the tropics?
Altitude. Mexico City sits at 2,240 m elevation, and air temperature drops roughly 6.5°C per kilometre of altitude (the standard environmental lapse rate). At the same latitude, sea-level Bangkok averages 28°C — the ~12°C gap is almost entirely the altitude correction. Quito (2,850 m) and La Paz (3,640 m) follow the same pattern: tropical latitudes, temperate temperatures.
What's the difference between air temperature and "feels like"?
We publish WMO-standard 2-metre air temperature measured in the shade. "Feels like" indices (heat index, wind chill, humidex) fold in humidity and wind to estimate physiological perception, but those values are not standardised across countries and add noise to relocation comparisons. Doha at 27°C annual mean feels much hotter in July due to humidity; the air-temperature number stays comparable across cities.
Are urban heat islands modelled?
No. Even at CHELSA's 0.05° (~5.5 km) map resolution, dense urban cores typically run 1–3°C warmer than the surrounding grid cell in summer — that signal is too local for a downscaled climatology to capture. Treat the published number as the regional value; if a city is large and dense (Cairo, Phoenix, Tokyo), expect downtown to run warmer than the mapped figure.
Is the data current for 2026?
Values are the 1981–2010 CHELSA v2.1 climatological normals — a fixed 30-year reference window, not a live feed. Ongoing warming means a recent year typically runs warmer than that baseline. That's exactly the gap the Daytime high mode's ERA5 warming uplift (2011–2025 normals vs. 1981–2010, +0.4 to +0.9°C globally) is built to correct for.

Find your climate match.

The map combines temperature, sunshine, cost, tax, and safety on every pin. Set your home as the baseline and see every destination as a delta from where you live now.

About the data: GeoRank uses CHELSA v2.1 monthly mean temperature (tas) for the 1981–2010 climatology, downscaled at ~1 km native resolution and processed to a 0.05° (~5.5 km) map grid, then cross-checked against 219 WMO reference stations and national met service archives (NOAA in the US, Met Office in the UK, DWD in Germany, IPMA in Portugal, AEMET in Spain). Urban heat islands are not modelled — values represent the regional grid cell. See the methodology for source-by-source detail.

Sources: CHELSA v2.1 climatology (WSL / University of Zurich) · ERA5 reanalysis (Copernicus Climate Data Store, EU) for the Daytime high warming uplift · 219 WMO reference weather stations · NOAA climate archives (United States) · Met Office UK (HadCRUT cross-reference) · DWD Germany · IPMA Portugal · AEMET Spain. Methodology and accuracy bounds at methodology.