Rarest WorldNot NormalSynthetic demonstration series · not observed data
Not Normal
One full-bleed world deviation surface, built from 53 instruments.
Every reading is judged against that place’s own 75-year record rather than against a global
average, because a temperature on its own tells you nothing and a temperature measured against local
history tells you everything. The instruments are drawn as hard-edged survey marks on a continuous
field; the field between them is interpolated, and the application says so wherever it is doing
it.
What an anomaly percentile actually means
An anomaly is the gap between today’s value and what is normal for that specific
place, that specific variable and that specific time of year. It is expressed here two ways.
Sigma (σ) counts how many standard deviations today sits from the baseline
average. Roughly two thirds of years land within ±1σ, about 95% within ±2σ, and
about 99.7% within ±3σ. A reading at +4σ is therefore not “a hot day” —
it is a value the historical distribution barely admits is possible.
Percentile restates the same thing as a rank: the 99th percentile means only one
year in a hundred, in that baseline climate, would be expected to reach it. The reciprocal of the
tail gives a return interval — a 1-in-500-year value — which is the form most people find
intuitive, provided you remember that it describes the odds under the old climate, not a promise
about the calendar.
The critical move is choosing which stretch of history counts as normal. A value can be
a once-a-century event against a 1951–1980 baseline and an ordinary year against a
1991–2020 one. Nothing about the measurement changed; only the yardstick did. This
application lets you switch that yardstick — including to the thirty years that began the year you
were born — and watch every verdict on the wall move.
What the surface between the instruments is
53 points cannot support a continuous global field, and this application
does not pretend otherwise. The coloured surface is built in two parts: a broad-bandwidth
background carrying synoptic-scale structure, plus a narrow-bandwidth interpolation of each
instrument’s departure from that background. Everything is computed on the sphere using
great-circle distance, which is why an instrument at a pole constrains a band of latitude rather
than a point, and why the antimeridian is not a seam.
Near an isolated instrument the surface follows that instrument closely. Away from every
instrument it relaxes toward a background prior and the application screens it — the
colour visibly breaks into a halftone — because at that distance the picture is a model rather
than a reading. Each instrument also carries a dashed ring at one kernel bandwidth
(9.5°, about 1,060 km); the ground outside every ring is inference. Where two instruments
sit close together and disagree, one scalar field cannot carry both, and pointing at either of them
says so in words.
Do not read a number off the colour. Every number in this application is an
instrument’s own, one point at a time.
Data provenance
These series are modelled, not measured. Each station’s 75 annual values
are synthesised by a seeded deterministic generator from documented parameters — a baseline mean, an
interannual standard deviation, a per-decade trend and an autocorrelation term — set to published
climatological magnitudes so that the shapes, units and directions are plausible. They are a
demonstration of how anomaly reasoning works. They are not observations, they must
not be cited as observations, and no reading in this application is live. Where a real-world record
is well documented it is cited separately and labelled as such.
The instruments (53)
Each entry lists today’s modelled reading and its departure from that station’s own
1951–1980 average. These are the points the surface is built from; everything between them is
interpolated.
temperature · 16 stations
Phoenix, Arizona, United States — July mean daily maximum: 44.8 °C, +4.60σ from its own 1951–1980 average.
Lytton, British Columbia, Canada — June peak daily maximum: 46.3 °C, +5.80σ from its own 1951–1980 average.
Delhi, India — May mean daily maximum: 42.5 °C, +2.60σ from its own 1951–1980 average.
Coningsby, Lincolnshire, United Kingdom — July peak daily maximum: 38.0 °C, +4.20σ from its own 1951–1980 average.
Floridia, Sicily, Italy — August peak daily maximum: 42.3 °C, +1.90σ from its own 1951–1980 average.
Mitribah, Kuwait — July mean daily maximum: 48.6 °C, +1.50σ from its own 1951–1980 average.
Verkhoyansk, Sakha, Russia — June mean daily maximum: 31.6 °C, +4.40σ from its own 1951–1980 average.
Utqiaġvik (Barrow), Alaska, United States — Annual mean temperature: -7.5 °C, +4.50σ from its own 1951–1980 average.
Esperanza Base, Antarctic Peninsula, Antarctica — February peak daily maximum: 14.0 °C, +2.20σ from its own 1951–1980 average.
Onslow, Western Australia, Australia — January peak daily maximum: 46.7 °C, +1.10σ from its own 1951–1980 average.
Zürich, Switzerland — Winter (Dec–Feb) mean temperature: 3.5 °C, +2.90σ from its own 1951–1980 average.
Manaus, Amazonas, Brazil — September mean daily maximum: 35.7 °C, +3.90σ from its own 1951–1980 average.
Hanoi, Vietnam — May mean daily maximum: 33.4 °C, +0.90σ from its own 1951–1980 average.
Sevilla, Andalucía, Spain — August mean daily maximum: 38.8 °C, +2.40σ from its own 1951–1980 average.
Chicago, Illinois, United States — January mean temperature: -12.9 °C, −2.80σ from its own 1951–1980 average.
Vostok Station, Antarctica — Annual mean temperature: -54.9 °C, +0.50σ from its own 1951–1980 average.
precipitation · 8 stations
Mawsynram, Meghalaya, India — June–September monsoon total: 6950 mm, −1.80σ from its own 1951–1980 average.
Cape Town, Western Cape, South Africa — May–August winter rainfall total: 145 mm, −2.90σ from its own 1951–1980 average.
Dubai, United Arab Emirates — Annual rainfall total: 359 mm, +4.80σ from its own 1951–1980 average.
Valencia, Spain — October maximum daily rainfall: 272 mm, +5.20σ from its own 1951–1980 average.
Nairobi, Kenya — March–May long-rains total: 130 mm, −1.60σ from its own 1951–1980 average.
Manila, Philippines — July–September rainfall total: 2058 mm, +1.90σ from its own 1951–1980 average.
Perth, Western Australia, Australia — May–September rainfall total: 318 mm, −3.10σ from its own 1951–1980 average.
Chennai, Tamil Nadu, India — October–December north-east monsoon total: 1346 mm, +2.10σ from its own 1951–1980 average.
wind · 5 stations
Cape Hatteras, North Carolina, United States — August–October peak sustained wind: 140 km/h, +0.80σ from its own 1951–1980 average.
Lerwick, Shetland, United Kingdom — Winter peak gust: 173 km/h, +1.40σ from its own 1951–1980 average.
Naha, Okinawa, Japan — Typhoon-season peak sustained wind: 164 km/h, +1.00σ from its own 1951–1980 average.
Boulder, Colorado, United States — December–January peak downslope gust: 206 km/h, +3.60σ from its own 1951–1980 average.
Punta Arenas, Magallanes, Chile — Annual mean wind speed: 31.7 km/h, +2.00σ from its own 1951–1980 average.
river flow · 6 stations
Rhine at Kaub, Germany — August mean discharge: 487 m³/s, −2.10σ from its own 1951–1980 average.
Colorado River at Lees Ferry, Arizona, United States — Annual natural flow: 215 m³/s, −2.20σ from its own 1951–1980 average.
Po at Pontelagoscuro, Italy — July mean discharge: 70 m³/s, −3.00σ from its own 1951–1980 average.
Mekong at Chiang Saen, Thailand — Wet-season mean discharge: 3654 m³/s, −0.70σ from its own 1951–1980 average.
Amazon at Manacapuru, Brazil — Dry-season minimum discharge: 43,600 m³/s, −3.20σ from its own 1951–1980 average.
Danube at Budapest, Hungary — September mean discharge: 1408 m³/s, −1.10σ from its own 1951–1980 average.
wave height · 5 stations
Rockall Trough, North Atlantic, International waters — Winter mean significant wave height: 4.85 m, +1.30σ from its own 1951–1980 average.
Nazaré Canyon, Portugal — Winter peak significant wave height: 12.09 m, +2.30σ from its own 1951–1980 average.
Paradip, Odisha, India — Cyclone-season peak significant wave height: 6.26 m, +0.60σ from its own 1951–1980 average.
Gulf of Alaska (buoy 46001), United States — Winter mean significant wave height: 3.23 m, −1.90σ from its own 1951–1980 average.
Southern Ocean, south of Tasmania, International waters — Annual peak significant wave height: 16.92 m, +1.70σ from its own 1951–1980 average.
sea ice · 4 stations
Arctic Ocean (basin-wide), Arctic — September minimum sea-ice extent: 3.89 million km², −8.60σ from its own 1951–1980 average.
Bering Sea, Arctic — March maximum sea-ice extent: 0.42 million km², −3.60σ from its own 1951–1980 average.
Antarctic (circumpolar), Antarctica — February minimum sea-ice extent: 1.69 million km², −4.40σ from its own 1951–1980 average.
Gulf of Bothnia, Baltic Sea, Finland / Sweden — Annual maximum ice cover: 88 thousand km², −1.50σ from its own 1951–1980 average.
snowpack · 5 stations
Sierra Nevada (Phillips Station), California, United States — 1 April snow water equivalent: 81 mm, −1.90σ from its own 1951–1980 average.
Wasatch Range (Alta), Utah, United States — 1 April snow water equivalent: 1692 mm, +2.20σ from its own 1951–1980 average.
Weissfluhjoch, Davos, Switzerland — 1 April snow depth: 84 cm, −2.90σ from its own 1951–1980 average.
Sapporo, Hokkaido, Japan — Seasonal total snowfall: 496 cm, −0.90σ from its own 1951–1980 average.
Hunza Valley, Gilgit-Baltistan, Pakistan — 1 April snow water equivalent: 170 mm, −2.00σ from its own 1951–1980 average.
humidity · 4 stations
Houston, Texas, United States — July mean dewpoint: 24.7 °C, +2.70σ from its own 1951–1980 average.
Jacobabad, Sindh, Pakistan — June peak wet-bulb temperature: 32.7 °C, +2.90σ from its own 1951–1980 average.
Ras al Khaimah, United Arab Emirates — August peak wet-bulb temperature: 32.6 °C, +1.80σ from its own 1951–1980 average.
Athens, Greece — August mean afternoon relative humidity: 25.0 %, −2.60σ from its own 1951–1980 average.
Read further
Learn — climate normals, sigma anomalies, return periods and shifting baselines, in plain language.
Documented records — 9 real, well-documented extremes, each read against its place.
All 53 stations — one page per instrument, modelled series and its two normals.