Methodology
Every number on this site comes from a public source we can name, and every derived value uses a published formula. Below: where measured readings, forecasts and normal ranges come from, how often they refresh, and the equations behind values like station pressure and pressure altitude.
How readings are chosen and refreshed
- Measured pressure comes from METAR reports filed by airport weather stations and published by NOAA's Aviation Weather Center. Each city uses the nearest station within 31 miles (50 km) that's currently reporting pressure; if the closest one is silent, we try the next two.
- Stations file routine reports on a fixed schedule, every hour at US airports, plus extra reports when the weather changes quickly. City pages pick up new reports about every 30 minutes, and every reading shows the time it was measured.
- Past readings are each station's own reports for the past 7 days, from the same NOAA feed. Yesterday's low and high come from the reports between the city's local midnights, and the 24-hour change compares the latest report with the one nearest to 24 hours earlier, within 45 minutes. At stations above 3,300 feet the history uses the altimeter setting, because sea-level pressure there also moves with the temperature. Nothing is smoothed or filled in.
- When no station qualifies, the page shows the MET Norway forecast value for the current hour and labels it as a forecast.
- The forecast comes from MET Norway's Locationforecast. Outside the Nordic countries it's based on the European Centre for Medium-Range Weather Forecasts' global model and updated four times a day. It's hourly for about the first 60 hours, then six-hourly, and runs about 10 days ahead.
- Normal ranges (US cities only) are NOAA's 1991–2020 hourly climate normals for the nearest normals station within 47 miles (75 km). If that station lacks pressure normals for more than one in ten of the 288 month-and-hour slots, the nearest station within that distance that has all of them is used instead, when there is one. NOAA gives the average sea-level pressure and its 10th and 90th percentiles for every hour of every day; we average each hour over the days of the month.
- US highs and lows rank the sea-level pressure that stations below 3,300 feet (1,000 m) reported in the last two hours, and leave out any station more than 10 hPa (0.30 inHg) away from the median of its neighbors within 186 miles (300 km). The top-10 lists rank our cities by the reading on each city's page, under the same height, time and neighbor rules; where a report has no sea-level pressure, as at most stations outside North America, they use its altimeter setting.
- A sea-level pressure that contradicts its own report is dropped. At stations below 3,300 feet we take the report's altimeter setting and temperature back down to sea level; if the reported sea-level pressure is off from that by more than 3 hPa plus 0.5 hPa per 100 m of station height, the city page shows the altimeter setting instead, and the map and the US highs and lows leave the station out.
- The pressure map uses the same station rules as the national highs and lows, but only reports that include sea-level pressure. A two-pass weighted average (the Barnes method) spreads the readings onto a grid of points a quarter of a degree apart; points with no station close enough stay empty. Isobars are traced every 4 hPa (about 0.12 inHg). The fronts, highs and lows come from NOAA's Weather Prediction Center: its coded surface bulletin, read within the hour of each 3-hourly analysis. When that analysis is more than 6 hours old, an H or L marks a center only where a closed isobar surrounds it inside the area with readings.
- High and low thresholds on our guide pages come from the typical US station: the median, across the NOAA normals stations below 3,300 feet (1,000 m) that our US cities use, of each station's 10th and 90th percentiles at noon local standard time for the month.
- Station weather shown next to the national and world extremes is decoded from the same METAR reports into plain words, using the National Weather Service's list of METAR abbreviations.
- Pressure altitude and density altitude on the pressure altitude calculator use the National Weather Service formulas below. An airport reports its altimeter setting rather than its station pressure, so the calculator first converts it back with the altimeter-setting formula.
Normals and records outside NOAA's official set
NOAA NCEI publishes official 1991–2020 hourly normals for US stations only. For every other city, and for US cities with no NCEI hourly-normals station within 75 km, we calculate the same three numbers ourselves from NOAA's hourly airport reports (the Integrated Surface Database, ISD-Lite files on NOAA's open-data service): for each month and each hour of local standard time, the 1991–2020 average sea-level pressure and its 10th and 90th percentiles, the middle 80% the pages call normal.
A month-hour counts only when at least 20 of the 30 years have at least 10 readings at that hour. Where a station reported every 3 or 6 hours, the hours in between are interpolated in a straight line, across at most 6 hours, and every month needs at least 4 measured hours a day. A station that misses either rule gets no normals. 1,034 stations pass.
To check the method we ran it on 169 US airports that also have NCEI's official normals. The two agree to a median of 0.05 hPa for the average, 0.14 hPa for the 10th percentile and 0.12 hPa for the 90th (the worst airport differs by 0.34 hPa on the average).
Airport weather reports (METARs) outside North America carry the altimeter setting but not sea-level pressure, so where an airport sends nothing else, its ISD-Lite file has gaps. For 101 such airports we take the altimeter setting from NOAA's full hourly files instead and build the same normals from it; those pages compare the current altimeter setting with them and say so. Near sea level the two measures are almost the same: at 7 US airports below 50 m, our altimeter-setting averages differ from NCEI's sea-level normals by 0.07–0.22 hPa.
706 city pages use these computed normals. On 194 of them the normals come from a station other than the one the reading comes from, and the page names it with its distance: the city's next listed airport or, where the reading airport has no complete 1991–2020 record (some close at night, some have fewer than 20 years of reports) or there is no airport within 50 km, the nearest station within 50 km and 200 m of the city's height that has one. To see what a nearby station costs we ran 15 US cities the same way: normals from a station a median 18.6 km from the city's NCEI station differ from NCEI's official ones by a median 0.6 hPa on the average and 0.57–0.87 hPa on the 10th and 90th percentiles.
A city's reading is compared with these normals as sea-level pressure: the report's own sea-level value when it has one, otherwise sea level worked out from the altimeter setting and the air temperature (the same formula as our check on reported values, at airports up to 3,300 ft), otherwise the forecast model when there is no current report.
The records (943 stations) come from every report in NOAA's full hourly files, 1991 to the end of the archive in August 2025: the report's sea-level pressure or, at stations up to 100 m above sea level, its altimeter setting (worked out from the station pressure when the report has only that), which that close to sea level is within a few tenths of a hPa of it. US airports stop reporting sea-level pressure when their temperature sensor fails, which is what happens in the storms that set records. Every report counts except those NOAA marks as erroneous. Readings NOAA marks as suspect, and those worked out from station pressure, count only when they link to readings that passed NOAA's checks in steps of at most 15 hPa within 3 hours, which is how a storm's eye links through its eyewall and how a lone bad value fails to. Then our own checks decide. Readings outside 870–1,085 hPa are dropped, as are lone one-hour spikes and stretches that start and end with opposite jumps of over 20 hPa while the hours around them are calm. A reading more than 30 hPa from the station's usual level for the month stays only if the readings before or after it lead there step by step: a storm deepens over hours and a high builds over days, while bad data (station pressure filed as sea-level pressure, or a fill value such as 900.1 hPa) jumps there. A record low must be backed by a second reading within 3 hours and 15 hPa, a high within 4 hPa, and 24-hour changes over 60 hPa are treated as errors. A record that no station within 150 km and 300 m of height comes near (a high more than 8 hPa above every neighbour's own record, a low more than 25 hPa below) is not shown (20 stations). Against published station lows on the same day, ours come out at Philadelphia 953.1 hPa (953 published, Sandy, 2012), Atlantic City 948.6 hPa (948.3 published, Sandy, 2012), Trenton 958.3 hPa (958 published, Sandy, 2012), Baltimore 965.7 hPa (964 published, Sandy, 2012) and Lake Charles 958.7 hPa (956 published, Laura, 2020); an hourly report can miss a low that falls between two reports. These are our calculations from NOAA data, not NOAA's own published normals or records.
Equations
Altimeter setting from station pressure (NWS)
Altimeter setting = (P − 0.3) × [1 + k × h ÷ (P − 0.3)^0.190284]^(1 ÷ 0.190284)where k = 1013.25^0.190284 × 0.0065 ÷ 288
with P the station pressure (hPa) and h the elevation (m).
Station pressure from altimeter setting (NWS, inverted)
P_station = (A^0.190284 − k·h)^(1/0.190284) + 0.3, where k = 1013.25^0.190284 × 0.0065 / 288, A = altimeter setting (hPa), h = elevation (m).
Pressure altitude (NWS)
h(ft) = (1 − (P/1013.25)^0.190284) × 145366.45.
Pressure at altitude in a standard atmosphere (NWS)
The pressure-altitude formula solved for pressure, with h the height in feet and P in hPa. The altitude tables on this site use it.
Altitude from a pressure reading (from the NWS formula)
With the current altimeter setting p0 (sea-level pressure in a standard atmosphere) in place of 1013.25, the same formula turns a pressure reading p into a height in feet. Android's developer documentation describes the same approach for phones:
Density altitude (NWS)
Vapor pressure e in hPa from the dew point Td in °C, then virtual temperature Tv from the air temperature T in kelvins and the station pressure P in hPa:
Then density altitude in feet, with the station pressure P in inHg and Tv in degrees Rankine:
Standard temperature at altitude
15 °C (59 °F) at sea level, falling by 0.0065 °C per meter (about 3.6 °F per 1,000 feet), as in the standard atmosphere behind the altimeter setting.
3-hour tendency (Met Office bands, hPa / 3 h)
< 0.1 steady; 0.1–1.5 slowly; 1.6–3.5 (no qualifier); 3.6–6.0 quickly; > 6.0 very rapidly.
We compute the 3-hour change from the station's own reports three hours apart; when that isn't available we use the tendency the station reports, if any.
Unit factors (NIST SP 811)
1 inHg = 3386.389 Pa; 1 mmHg = 133.3224 Pa; hPa = mb.
City elevation
City elevation is the terrain height at the city's reference point from the SRTM elevation model (GTOPO30 north of 60° N, where SRTM ends), via GeoNames.
Data sources
| Domain | Source | Publisher | License |
|---|---|---|---|
| observed-pressure | METAR/SPECI surface observations (Data API + cache files) | NOAA / NWS Aviation Weather Center | U.S. Government work (17 U.S.C. 105) - public domain |
| forecast-pressure | Locationforecast 2.0 (compact) | Norwegian Meteorological Institute (MET Norway) | CC BY 4.0 and NLOD 2.0 (dual-licensed; attribution required); free incl. commercial use |
| unit-conversions | NIST SP 811 Appendix B — conversion factors | NIST | U.S. Government work - public domain |
| pressure-formulas | NWS meteorological calculator formulas (pressure altitude, altimeter setting, station pressure, density altitude) | NOAA National Weather Service (WFO El Paso wxcalc; WFO San Diego calculator) | U.S. Government work - public domain |
| pressure-science | Pressure definitions (station pressure, altimeter setting, mean sea level pressure) | NOAA National Weather Service, WFO Boulder | U.S. Government work - public domain |
| pressure-tendency-terms | Marine forecasts glossary — pressure tendency in station reports | Met Office (UK) | Met Office website content (UK Open Government Licence per Met Office terms) - definitions cited with attribution |
| world-pressure-records | WMO Weather and Climate Extremes — records table (as of 31 July 2025) | World Meteorological Organization | WMO publication - facts cited with attribution |
| us-pressure-normals | U.S. Hourly Climate Normals 1991–2020 (sea level pressure mean, 10th and 90th percentile by hour) | NOAA NCEI | U.S. Government work - public domain |
| surface-analysis | WPC surface analysis (fronts, highs/lows, isobars) + coded surface bulletin | NOAA NWS Weather Prediction Center | U.S. Government work - public domain |
| zip-geocoding | Gazetteer Files 2025 - ZIP Code Tabulation Areas (internal point lat/lon) | U.S. Census Bureau | U.S. Government work - public domain |
| city-locations | GeoNames cities (coordinates + DEM elevation) | GeoNames | CC BY 4.0 |
| pressure-daily-cycle | JetStream: Air Pressure (daily pressure cycle; where storms change pressure most) | NOAA National Weather Service (JetStream online weather school) | U.S. Government work - public domain (JetStream: 'You are free to use the materials in any manner you wish') |
| aviation-references | Pilot's Handbook of Aeronautical Knowledge (FAA-H-8083-25), chapter 11; Density Altitude (FAA-P-8740-2) | Federal Aviation Administration | U.S. Government work - public domain |
| phone-altitude-reference | SensorManager.getAltitude (Android API reference) | Google, Android Developers | Android Developers documentation (content license on developer.android.com); fact paraphrased with attribution |
| station-history-archive | Integrated Surface Database (ISD), global hourly station observations | NOAA NCEI | U.S. Government work - public domain |
| bombogenesis-definition | What is bombogenesis? | NOAA National Ocean Service | U.S. Government work - public domain |
| state-boundaries | Cartographic Boundary Files 2017 — states (us-atlas 3.0.1 states-albers-10m) | U.S. Census Bureau, Geography Division (TopoJSON redistribution: topojson/us-atlas) | U.S. Government work (17 U.S.C. 105) - public domain |
| metar-abbreviations | METAR/TAF list of abbreviations and acronyms, with the key to decode an ASOS (METAR) observation | NOAA National Weather Service | U.S. Government work - public domain |