People have measured air pressure since the 1640s, and it is still the weather reading most often quoted without enough context to be useful. A figure like 1013 hPa means very little on its own: it needs a place, an hour, and above all a direction. This article covers the missing context: what the number is, why it is corrected before you see it, who uses which unit, and how the forecasts are made.

What the Number Is

Atmospheric pressure is the weight of the column of air above a given point. The SI unit is the pascal; meteorology works in hectopascals (hPa), where one hectopascal is 100 pascals. The standard atmosphere, the internationally agreed reference value at sea level, is 1013.25 hPa.

Because pressure is the weight of the air overhead, it falls as you climb. Near sea level the drop is roughly one hectopascal for every 8.4 metres of altitude, easing to about 10.4 metres per hectopascal at 2,200 m, where the air is already thinner. That single fact explains most of the confusion people have about pressure readings, and it is the reason for the correction described below.

The Units, and Why a Millibar Is a Hectopascal

Pressure is written five different ways depending on where you are, but one hectopascal is exactly one millibar: the same quantity under two names, not a conversion.

Unit	Standard sea-level pressure	Used by
hPa	1013.25	Weather services in most of the world
mbar	1013.25	Home weather stations, shipping forecasts
mmHg	760	Russia and neighbouring countries
inHg	29.92	The United States; aviation in the US, Canada and Japan
kPa	101.325	Canada

The conversions are straightforward: hPa to mmHg, multiply by 0.75; hPa to inHg, divide by 33.864; hPa to kPa, divide by 10.

One detail trips people up regularly. The familiar 29.92 inHg belongs to the exact standard figure of 1013.25 hPa. A plain 1013 hPa works out at 29.91 inHg, and 760 mmHg likewise belongs to 1013.25 rather than to 1013.

Why Every Published Reading Is Corrected to Sea Level

A raw barometer reading tells you more about where the instrument is than about the weather around it. Mexico City sits 2,238 m up, so a barometer there reads roughly 786 hPa on an ordinary day, a figure that would signal a catastrophic storm at sea level and signals nothing at all up there. Compare that raw number against a coastal city's and you learn which city sits higher, not which one has the deeper low.

For this reason national weather services publish mean sea-level pressure: the reading adjusted to what it would be if the station stood at sea level, using the station's elevation and the temperature of the air column. It is a calculated figure, not a measured one, and over high terrain it describes an air column that does not physically exist — but it is what makes two places comparable, and it is what every synoptic chart, shipping forecast and weather app quotes.

So if your home barometer disagrees with the forecast, altitude is almost always the reason. In Mexico City the two figures were 786 hPa and 1019 hPa at the same moment. Height beats weather comfortably: a dramatic week of weather might move the pressure 30 hPa, while walking up 250 metres does the same thing.

What Counts as High and Low

At sea level, most weather happens between roughly 980 and 1040 hPa. Deep mid-latitude storms reach into the 950s; strong winter anticyclones over continental interiors can exceed 1040.

The verified extremes are far outside that band:

Lowest: 870 hPa, measured in Typhoon Tip over the western Pacific on 12 October 1979, the lowest sea-level pressure recorded on Earth outside a tornado.
Highest: 1083.8 hPa at Agata, Siberia, on 31 December 1968, under an exceptionally cold and dense air mass.

That is a total observed range of a little over 200 hPa, on a scale where a 10 hPa move in a day counts as a brisk change.

The Daily Tide That Has Nothing to Do With the Weather

Surface pressure also carries a thermal tide: a twice-daily oscillation driven by the sun heating the atmosphere, chiefly through ozone and water vapour absorption, rather than by any weather system. In the tropics it swings about 1 hPa either side of the day's average, cresting two to three hours before local noon and midnight and bottoming out near 04:00 and 16:00 local time.

In the tropics this signal dominates. Near the equator, the pressure trace over a quiet day is essentially a clock: it rises and falls on schedule whether or not anything is happening meteorologically. The tide weakens toward the poles, and at mid-latitudes it is largely buried under the weather itself, which is why most European and North American readers have never noticed it.

This matters when you read a barometer: a tropical station that has "fallen 2 hPa since this morning" may simply be on the descending half of the tide.

Why the Change Matters More Than the Level

Knowing that the pressure is 1004 hPa answers almost nothing. Knowing that it has fallen 9 hPa in the last day tells you a weather system is arriving, and roughly how vigorously.

The synoptic convention is the three-hour tendency, reported in the standard SYNOP observation code and drawn on charts as a small rising or falling trace. It is the right window for a forecaster tracking a front across a mid-latitude chart. It is a poor window near the equator, where the thermal tide can account for most of a three-hour change and the resulting figure ranks calm tropical stations as though they were in a storm.

A 24-hour comparison sidesteps the problem: the tide returns to roughly where it started after a full day, so what remains is the weather. That is the window barometer.today leads with on every city page, expressed as a sentence rather than an arrow.

How Pressure Forecasts Are Produced

Forecast pressure fields come from numerical weather prediction. A model such as the ECMWF Integrated Forecasting System (IFS) is initialised from observations (surface barometers, radiosondes, aircraft, satellites) and integrated forward, producing a mean sea-level pressure field on a global grid. The version ECMWF publishes openly comes at 0.25° resolution and is updated four times a day.

Two things about that output are easy to get wrong:

The recent past is not a measurement. Model values for hours that have already passed are the model's analysis of those hours, each initialised from real observations. They are the part of the series most tightly tied to what was actually observed, but still a gridded field, not a reading from a barometer in your town.
Timing is the weak point, not magnitude. In an August 2026 comparison for London, ECMWF, GFS, ICON and UKMO agreed on the forecast pressure to within 0.7 hPa. Where models genuinely disagree is when a system arrives: a low can turn up half a day earlier or later than drawn.

Where to Find Pressure Data

ECMWF — the Integrated Forecasting System, and documentation of the model behind most open pressure forecasts
Open-Meteo — open-source weather API serving pressure_msl and surface_pressure for any location
Barometer Today (barometer.today) — current sea-level pressure and its 24-hour change for cities worldwide, written as a plain sentence, with the reading at the city's own altitude alongside it
barometer.today/en/units — the full conversion table between hPa, mbar, mmHg, inHg and kPa
barometer.today/en/altitude — how far pressure falls with height, with reference values from sea level to 3,000 m

Barometric Pressure Explained: Units, Altitude and Forecasts