Cloud base calculator: estimating cumulus height from the spread

Tafrob editorial team · Updated · 8 min read

Short answer

The cloud base of convective cumulus is estimated by dividing the spread between temperature and dew point by 2.5 and multiplying by 1,000 feet. That is 2.5 degrees Celsius of spread per 1,000 feet, or about 4.4 degrees Fahrenheit in the usual rule of thumb. A 5 degree spread gives about 2,000 feet above ground. The rule does not apply to fog, stratus or frontal cloud.

Runs in your browser. Training and planning aid only, not for operational use.

Key points

  • Base above ground in feet = (temperature minus dew point in C) / 2.5 x 1,000, which is 400 ft per degree of spread.
  • A rising unsaturated parcel cools about 3 C per 1,000 ft while its dew point falls about 0.5 C, so the spread closes at 2.5 C per 1,000 ft.
  • The result is the lifting condensation level computed from surface values, valid for afternoon cumulus in a well-mixed layer.
  • METAR temperatures are rounded to whole degrees, so the estimate carries an uncertainty of roughly 400 ft either way.
  • The page returns height above ground and, with a field elevation, height above mean sea level.
Cumulus mediocris clouds with flat bases against a blue sky
Photo: Janne Naukkarinen, Public domain, via Wikimedia Commons

What the calculator does

The calculator on this page takes a temperature, a dew point and, optionally, a field elevation. It returns the estimated base of cumulus cloud above ground level in feet and metres, and the same height above mean sea level when an elevation is typed. Nothing is sent to a server: the arithmetic runs in the browser.

The code behind it is one line:

cloud base (ft above ground) = (T - Td) / 2.5 x 1,000

T and Td are the air temperature and the dew point in degrees Celsius, and the spread T - Td is the difference between them. Dividing by 2.5 and multiplying by 1,000 is the same as 400 feet per degree of spread. If the spread is zero or negative, the page shows 0 ft rather than a negative height. The metric figure is the height in feet multiplied by 0.3048. The default values, 24 and 14, give a spread of 10 and a base of 4,000 feet, or about 1,219 metres.

The temperature and dew point are the two numbers found in the TT/TdTd group of a METAR, described on the temperature and dew point page. In 18/13 the temperature is 18 C, the dew point is 13 C, and the spread is 5 C.

Where 2.5 C per 1,000 ft comes from

The rule is a statement about a rising parcel of air. The FAA Aviation Weather Handbook describes the process in its chapter on vertical motion and clouds. An unsaturated parcel that is lifted expands, and the expansion takes energy from the parcel, so it cools. The cooling rate of an unsaturated parcel is about 3 C per 1,000 ft (9.8 C per km), called the dry adiabatic lapse rate. At the same time the dew point of the parcel decreases by about 0.5 C per 1,000 ft (1.8 C per km), because the pressure falls and the water vapour mixes into a larger volume.

The temperature drops by 3 C per 1,000 ft and the dew point by 0.5 C per 1,000 ft, so each 1,000 ft of lift narrows the spread by 2.5 C. The Handbook defines the lifting condensation level, or LCL, as the level at which a parcel lifted dry adiabatically becomes saturated: the spread is zero there and the relative humidity is 100 percent. Above it, further lifting makes cloud. A spread of 5 C therefore closes after 2,000 ft of lift, and the Handbook works the same case in its Figure 12-2: a parcel at 18 C with a dew point of 13 C reaches saturation at an LCL of 2,000 ft.

In metric terms, 9.8 C per km minus 1.8 C per km is 8 C per km, which is 125 m per degree of spread. The page uses feet and 400 ft per degree, about 122 m, so the two versions differ by a few percent only. The 125 m figure is a calculation from the Handbook values, not a number printed in it.

The Fahrenheit version

A common form of the rule uses degrees Fahrenheit. A NASA Glenn educational page for students writes it as the surface temperature minus the dew point, divided by 4.4, multiplied by 1,000, with the result in feet. Converted exactly, 2.5 C of spread is 4.5 F, so a divisor of 4.5 would match the Celsius rule. The 4.4 version gives a base about 2 percent higher, which is far smaller than the other uncertainties listed below. The NASA page gives no derivation, so the Celsius derivation above is the one that can be checked against the Handbook.

Worked examples

Temperature / dew point Spread Estimated base above ground
18 / 13 5 C 2,000 ft
24 / 14 (page default) 10 C 4,000 ft
30 / 12 18 C 7,200 ft
28 / 27 1 C 400 ft

At an airport at 1,500 ft elevation, the 24 / 14 case gives 4,000 ft above ground, which is 5,500 ft above mean sea level. The distinction matters because a METAR reports cloud heights above the airport, and an altitude read from the altimeter is above sea level. The coding of heights in reports is covered on the cloud height page.

The last row shows where the rule stops being useful. A spread of 1 C near the ground means the air is nearly saturated at the surface already, and the lowest cloud is probably fog or stratus, not cumulus formed by rising bubbles. The number is arithmetic, not a forecast of a layer.

What the estimate means physically

The result is the LCL computed from surface values. For cumulus on a sunny afternoon, the surface layer is heated and stirred by convection, so thermals carry surface air upward without exchanging much heat with their surroundings. The base of the cloud forms where such a thermal reaches saturation, which is the situation the formula describes. That is why the page labels the result as the estimated base of cumulus and why the output carries a warning that it does not predict stratus, fog or frontal cloud.

Several limits follow from the physics:

  • Stratus and fog form by cooling, mixing or advection, not by the dry lift of surface air. The spread at the surface says little about their height.
  • Frontal cloud and cloud forced up a mountain slope depend on the motion of the whole air mass, not on the surface spread.
  • In the morning, before heating has mixed the lowest layer, the surface values may not represent the air that will rise later. The estimate is best in the afternoon.
  • A parcel does not always start from the surface. Over a dry slope the thermals may come from a layer different from the one the thermometer sees.
  • Moisture is not uniform. A station near a lake, a wet field or the sea can report a dew point that is not representative of the air a few kilometres inland.

Rounding and uncertainty

A METAR reports temperature and dew point as whole degrees Celsius. The WMO guide to the codes states that each value is rounded to the nearest whole degree, and the Federal Meteorological Handbook sets the reporting resolution in the body of the report to whole degrees. Each reported value can therefore be wrong by up to 0.5 C in either direction, and the spread by up to 1 C. At 400 ft per degree, a reported spread of 5 C stands for a real spread between 4 and 6 C, which means a base between 1,600 and 2,400 ft. The remarks section of a US METAR can carry a T group with tenths of a degree, which reduces the error, as described on the T group page.

Beyond rounding there is natural variation. The surface spread changes through the day, and the thermals that make the cloud are not all alike, so a field of cumulus often has a base that varies by a few hundred feet from one cloud to the next.

A schematic report

METAR LXXX 191450Z 27008KT 9999 SCT055 28/14 Q1018

It reads: wind from 270 degrees at 8 knots, visibility 10 km or more, scattered cloud at 5,500 feet, temperature 28 C, dew point 14 C, QNH 1018 hPa. The spread is 14 C, so the estimate is 14 / 2.5 x 1,000 = 5,600 feet above ground. The reported layer at 5,500 feet sits close to it, which is the pattern expected on a warm afternoon with cumulus. The base estimate is not an extra datum: it is a check on whether the reported layer looks consistent with the air mass.

Compare a second line:

METAR LXXX 190650Z 00000KT 0300 FG VV001 11/10 Q1021

The spread is 1 C, so the formula gives 400 feet, but the cloud is fog with a vertical visibility of 100 feet. The formula does not apply to fog and the example is only a reminder of that limit.

In real reports

In a real METAR the lowest layer is measured by a ceilometer or judged by an observer, not computed. The reported layer is the observation, and the estimate from the spread is only a cross check. The way a ceilometer measures cloud height is explained on the ceilometer page, and the rules for what counts as a ceiling are on the ceiling page. When a report shows a small spread and a broken or overcast layer, the cloud is more likely a low stratiform layer than convective cumulus. When it shows a large spread and a few or scattered layer, convective cloud is more likely.

Method and limits

The formula and the 3 C and 0.5 C per 1,000 ft rates come from the FAA Aviation Weather Handbook (FAA-H-8083-28A), chapter 12, which also gives the 18 C and 13 C example with an LCL of 2,000 ft. The 400 ft per degree figure and the 125 m per degree conversion are derived from those rates, not quoted from a source. The Fahrenheit form with a divisor of 4.4 comes from the NASA Glenn page in the sources; exactly, 2.5 C equals 4.5 F, so the two published forms differ slightly and are left as found. The whole-degree rounding of temperature and dew point is from the WMO guide to the codes (WMO-No. 782) and from the Federal Meteorological Handbook No. 1. The estimate is the lifting condensation level of surface air, an approximation of the base of convective cloud and not a prediction for any other cloud type. The METAR lines are schematic, not real reports, and the photograph shows cumulus without being tied to a report. This page is not an operational weather source and must not be used for flight decisions.

Source code of this tool (MIT)
// Tafrob calculators - pure functions, MIT licence (c) Tafrob.
// Every formula is documented on the tool page with its source.

const rad = (d: number) => (d * Math.PI) / 180;

/** Wind components on a runway. dir/heading in degrees (same reference: both true or both magnetic). */
export function windComponents(windDir: number, speed: number, runwayHeading: number) {
  const angle = ((windDir - runwayHeading + 540) % 360) - 180; // -180..180, + = from the right
  const cross = speed * Math.sin(rad(angle));
  const head = speed * Math.cos(rad(angle));
  return { angle, cross, head }; // cross > 0: from the right; head < 0: tailwind
}

/** Saturation vapour pressure (hPa), Magnus formula with Alduchov & Eskridge (1996) coefficients. */
export const satVap = (tC: number) => 6.1094 * Math.exp((17.625 * tC) / (tC + 243.04));

/** Relative humidity (%) from temperature and dew point (deg C). */
export const relHumidity = (tC: number, tdC: number) => (100 * satVap(tdC)) / satVap(tC);

/** Dew point (deg C) from temperature (deg C) and relative humidity (%). */
export function dewPoint(tC: number, rh: number) {
  const g = Math.log(rh / 100) + (17.625 * tC) / (243.04 + tC);
  return (243.04 * g) / (17.625 - g);
}

/** Convective cloud base estimate above ground (ft) from the temperature/dew point spread (deg C). */
export const cloudBaseFt = (tC: number, tdC: number) => ((tC - tdC) / 2.5) * 1000;

/** Pressure altitude (ft) from field elevation (ft) and altimeter setting (inHg). FAA rule: 1,000 ft per inHg. */
export const pressureAltitude = (elevFt: number, altInHg: number) => elevFt + (29.92 - altInHg) * 1000;

/** ISA temperature (deg C) at a pressure altitude (ft). */
export const isaTemp = (paFt: number) => 15 - (1.98 * paFt) / 1000;

/** Station pressure (inHg) from altimeter setting (inHg) and elevation (m), NWS formula. */
export const stationPressure = (altInHg: number, elevM: number) => altInHg * Math.pow((288 - 0.0065 * elevM) / 288, 5.2561);

/**
 * Density altitude (ft), NWS method (WFO El Paso): station pressure and virtual temperature (humidity included).
 * e = 6.11 * 10^(7.5 * Td / (237.3 + Td)) hPa; DA = 145366 * (1 - (17.326 * P / Tv)^0.235), P in inHg, Tv in degrees Rankine.
 * Same constants as the NWS page, so results can be checked against it (a dry ISA sea level case gives about 18 ft, not 0).
 */
export function densityAltitude(elevFt: number, altInHg: number, tC: number, tdC?: number) {
  const p = stationPressure(altInHg, elevFt * 0.3048); // inHg
  const pHpa = p * 33.8639;
  const e = tdC === undefined ? 0 : 6.11 * Math.pow(10, (7.5 * tdC) / (237.3 + tdC)); // NWS vapour pressure
  const tK = tC + 273.15;
  const tvK = tK / (1 - (e / pHpa) * (1 - 0.622));
  const tvR = tvK * 1.8;
  const da = 145366 * (1 - Math.pow((17.326 * p) / tvR, 0.235));
  const pa = pressureAltitude(elevFt, altInHg);
  const thumb = pa + 120 * (tC - isaTemp(pa)); // rule of thumb: 120 ft per deg C above ISA
  return { da, pa, thumb, isa: isaTemp(pa), stationInHg: p };
}

/** Pressure conversions. */
export const HPA_PER_INHG = 33.8639;
export const HPA_PER_MMHG = 1.333224;
export const hpaToInhg = (h: number) => h / HPA_PER_INHG;
export const inhgToHpa = (i: number) => i * HPA_PER_INHG;
/** METAR A group: hundredths of inHg rounded down (FMH-1 11.5.1); the epsilon absorbs float error (29.92 -> 2991.9999...). */
export const aGroup = (hpa: number) => 'A' + String(Math.floor(hpaToInhg(hpa) * 100 + 1e-6)).padStart(4, '0');
/** Pressure altitude offset (ft) from 1013.25 hPa with the AIM rule of 1,000 ft per inHg (about 29.5 ft per hPa). */
export const pressureOffsetFt = (hpa: number) => ((1013.25 - hpa) / HPA_PER_INHG) * 1000;

export type Category = 'VFR' | 'MVFR' | 'IFR' | 'LIFR';
/** US flight category (NWS / Aviation Weather Center definitions). ceiling in ft AGL (Infinity if none), visibility in statute miles. */
export function flightCategory(ceilingFt: number, visSm: number): { cat: Category; byCeiling: Category; byVis: Category } {
  const c: Category = ceilingFt < 500 ? 'LIFR' : ceilingFt < 1000 ? 'IFR' : ceilingFt <= 3000 ? 'MVFR' : 'VFR';
  const v: Category = visSm < 1 ? 'LIFR' : visSm < 3 ? 'IFR' : visSm <= 5 ? 'MVFR' : 'VFR';
  const rank: Category[] = ['LIFR', 'IFR', 'MVFR', 'VFR'];
  const cat = rank[Math.min(rank.indexOf(c), rank.indexOf(v))];
  return { cat, byCeiling: c, byVis: v };
}

/** Speed conversions (from knots). */
export const KT = { mps: 0.514444, kmh: 1.852, mph: 1.150779 };

/** Interpret a METAR/TAF day-time group DDHHMMZ relative to a reference date (UTC). */
export function dayTimeGroup(group: string, ref = new Date()) {
  const m = group.trim().toUpperCase().match(/^(\d{2})(\d{2})(\d{2})Z?$/);
  if (!m) return null;
  const [d, h, mi] = [Number(m[1]), Number(m[2]), Number(m[3])];
  if (d < 1 || d > 31 || h > 23 || mi > 59) return null;
  // Choose the month that puts the date closest to the reference date.
  let best: Date | null = null;
  for (const off of [-1, 0, 1]) {
    const c = new Date(Date.UTC(ref.getUTCFullYear(), ref.getUTCMonth() + off, d, h, mi));
    if (c.getUTCDate() !== d) continue;
    if (!best || Math.abs(+c - +ref) < Math.abs(+best - +ref)) best = c;
  }
  return best;
}

Training and reference only. For flight planning, always use official briefing sources (your national weather service, AIS or approved provider).

Frequently asked questions

How do you calculate cloud base from temperature and dew point?

Subtract the dew point from the temperature, divide by 2.5 if both are in degrees Celsius, and multiply by 1,000. The result is the height of the cloud base above ground in feet. For a METAR group of 18/13 the spread is 5, so the estimate is 2,000 feet above ground level.

Why 2.5 degrees per 1,000 feet?

The FAA Aviation Weather Handbook gives about 3 C per 1,000 ft for the cooling of a rising unsaturated parcel and about 0.5 C per 1,000 ft for the fall of its dew point. The two curves approach each other by 2.5 C per 1,000 ft. They meet where the parcel becomes saturated and cloud forms.

Does the formula work in Fahrenheit?

Yes, with a different divisor. A NASA Glenn educational page gives the spread in degrees Fahrenheit divided by 4.4, multiplied by 1,000. Strictly, 2.5 C equals 4.5 F, so dividing by 4.5 is the exact conversion and dividing by 4.4 gives a slightly higher base, by about 2 percent.

Can the result be compared with the cloud layers in a METAR?

Only for convective cloud on a day of afternoon heating. A scattered or broken cumulus layer in the report often lies near the estimate. Stratus, fog, cloud under a warm front or cloud forced by hills follow other processes, so the same spread can sit under a layer at a very different height.

Why can the estimate be wrong by several hundred feet?

Temperature and dew point in a METAR are each rounded to the nearest whole degree, so the true spread can differ from the reported one by up to one degree. At 400 feet per degree, that alone moves the estimate by about 400 feet before any real variation of the air mass is counted.

Sources

  1. Federal Aviation Administration (2024). Aviation Weather Handbook (FAA-H-8083-28A)
  2. NASA Glenn Research Center. Module 1, Part 2, Lesson 2 (Aero2000 student web)
  3. World Meteorological Organization (2025). Aerodrome Reports and Forecasts, A Users' Handbook to the Codes (WMO-No. 782)
  4. Office of the Federal Coordinator for Meteorology (2019). Federal Meteorological Handbook No. 1 (FCM-H1-2019)