Why does rain matter only above about 10 GHz?

Because raindrops become large compared with the wavelength. A drop is up to about 5 mm across, while the wavelength is 3 cm at 10 GHz and 1 cm at 30 GHz. Below a few GHz the drops are tiny by comparison and barely touch the wave. At 50 mm/h on horizontal polarisation, the specific attenuation is 0.17 dB/km at 5 GHz, 1.66 dB/km at 10 GHz, 9.8 dB/km at 30 GHz and 19.7 dB/km at 100 GHz.

Why does horizontal polarisation lose more than vertical?

Falling raindrops flatten into a shape wider than it is tall, so a horizontally polarised wave meets more water. At 50 mm/h and 20 GHz, horizontal loses 5.72 dB/km and vertical 4.53 dB/km. The gap shrinks higher up, to 19.7 against 19.3 dB/km at 100 GHz. That is why microwave links in rainy climates often run vertical. Circular polarisation sits in between, at the average of the two coefficients.

Is this the total rain fade on my link?

No. It is the loss for one uniform rain rate along the whole path, which is exactly what P.838 defines. Real storms are patchy, and on a long hop the heaviest rain rarely covers every kilometre. ITU-R P.530 (terrestrial) and P.618 (Earth-space) turn a rain rate into a fade for a given percentage of time by applying an effective path length. This calculator stops before that step. Gases, clouds and a wet radome add their own losses on top.

Which rain rate should I enter?

For link design, the rate exceeded for 0.01 % of an average year, R0.01, which is about 53 minutes a year. The presets give R0.01 for the 15 rain zones of ITU-R P.837-1 from 1994, from 8 mm/h in zone A to 145 mm/h in zone P. Much of temperate Europe falls in zones E to K, the wet tropics in N and P. P.837-1 is superseded. The current P.837-8 gives R0.01 from digital world maps, so use it, or local rain-gauge statistics, when the exact figure matters.

Why does the elevation angle change nothing with circular polarisation?

Because the elevation enters only through cos²θ · cos 2τ, and cos 2τ = 0 at τ = 45°. The H and V coefficients are then averaged with equal weight at any angle. For linear polarisation the elevation does matter. Raising the path towards vertical blends H and V more and more, until at θ = 90° every polarisation gives the same value.

← All tools

Rain attenuation calculator (ITU-R P.838)

Rain attenuation is the loss a radio wave suffers when raindrops absorb and scatter it, and it grows fast above 10 GHz. This calculator applies Recommendation ITU-R P.838-3 to get the specific attenuation in dB/km from frequency, rain rate, polarisation and path elevation, then multiplies it by your path length for uniform rain.

Frequency

P.838-3 is fitted from 1 to 1000 GHz.

Rain-rate preset

R0.01 per rain zone from ITU-R P.837-1 (1994). The table is historical, the current P.837-8 uses digital maps instead. The zone maps are in P.837-1. Values re-read 7 November 2026.

Rain rate

Link planners use R0.01, the rate exceeded for 0.01 % of an average year (about 53 minutes).

Polarisation

Path elevation angle

θ in the formulas: 0° for a terrestrial link, the dish elevation for a satellite path. Only linear polarisation feels it. At 90° every polarisation gives the same value.

Path length

The same rain rate is assumed along the whole path.

Specific attenuation

4.76 dB/km

γR = k · Rᵅ, one uniform rain rate

Attenuation over the path

23.8 dB

γR × d, uniform rain · no P.530/P.618 path reduction

Power-law coefficient

0.09164

k · H 0.09164 · V 0.09611

Power-law exponent

1.0568

α · H 1.0568 · V 0.9847

Specific attenuation against frequency, both axes logarithmic, for the rain rate you entered. The bold green line is your polarisation and elevation. The dashed (H) and dotted (V) lines are pure horizontal and vertical polarisation on a level path. The pale band runs from 8 to 145 mm/h, the driest and the wettest rain zone of ITU-R P.837-1. The dot marks your frequency and its specific attenuation.

Implements Recommendation ITU-R P.838-3 (03/2005), equations (1) to (5). The path attenuation A = γR × d assumes uniform rain along the whole path. For outage prediction use ITU-R P.530 (terrestrial) or P.618 (Earth-space), which apply an effective path length. Gases, clouds, fog and wet antennas are not included. The rain-zone presets are historical values from P.837-1 (1994).

How it works

01
Fit the coefficients to the frequency

P.838-3 does not run a scattering calculation. It gives four curve-fitted coefficients, k_H, k_V, α_H and α_V, as sums of Gaussian terms in log₁₀ f, valid from 1 to 1000 GHz. At 20 GHz they are 0.09164, 0.09611, 1.0568 and 0.9847.
02
Combine them for your polarisation

Equations (4) and (5) blend the horizontal and vertical values through cos²θ · cos 2τ, where θ is the path elevation and τ the polarisation tilt. Horizontal takes the H values, vertical the V values, and circular (τ = 45°) the plain average. On a vertical path (θ = 90°) every polarisation gives the same result.
03
Apply the rain rate

The specific attenuation is γ_R = k · R^α in dB/km. Near 20 GHz α is close to 1, so doubling the rain rate roughly doubles the loss. At 20 GHz on horizontal polarisation, 42 mm/h gives 4.76 dB/km and 84 mm/h gives 9.90 dB/km. Higher up α falls (0.77 at 60 GHz, 0.68 at 100 GHz), so heavier rain adds less than in proportion.
04
Multiply by the path length

Over 5 km of uniform 42 mm/h rain at 20 GHz, A = 4.76 · 5 = 23.8 dB. Heavy rain rarely covers a whole long hop at full strength, so on long paths this overstates the fade. ITU-R P.530 (terrestrial) and P.618 (Earth-space) shorten the path to an effective length for that reason. This calculator does not.

Formulas

Coefficient fit, equations (2) and (3)

G = Σ a_j · e^−u_j² + m·x + c₀

x = log₁₀ f, u_j = (x − b_j) / c_j
a_j, b_j, c_j, m, c₀ — constants of P.838-3 Tables 1 to 4
k_H = 10^G with Table 1, k_V = 10^G with Table 2
α_H = G with Table 3, α_V = G with Table 4
f — frequency, GHz (1 to 1000)

Polarisation and elevation, equation (4)

k = k_H + k_V + (k_H − k_V) · C2

C = cos²θ · cos 2τ
θ — path elevation angle, degrees (0° for a terrestrial link)
τ — polarisation tilt from horizontal: 0° H, 90° V, 45° circular

Exponent, equation (5)

α = P_H + P_V + (P_H − P_V) · C2k

P_H = k_H · α_H, P_V = k_V · α_V
k — the combined k of equation (4)

Specific attenuation (equation 1) and path attenuation

γ_R = k · R^α, A = γ_R · d

γ_R — specific attenuation, dB/km
R — rain rate, mm/h (R_0.01 for the 0.01 % design case)
d — path length, km
A — attenuation over the path, dB, for uniform rain (no effective path length applied)

Worked example

Zone K rain (42 mm/h) on a 5 km link at 20 GHz

at f = 20 GHz: k_H = 0.09164, α_H = 1.0568, k_V = 0.09611, α_V = 0.9847
horizontal on a level path: τ = 0°, θ = 0°, so cos²θ · cos 2τ = 1, k = k_H and α = α_H
γ_R = 0.09164 · 42^1.0568 = 4.76 dB/km
A = 4.76 dB/km · 5 km = 23.8 dB
vertical instead: γ_R = 0.09611 · 42^0.9847 = 3.81 dB/km, 19.1 dB over the path
circular: k = 0.09388, α = 1.0199, γ_R = 4.25 dB/km

FAQ

Why does rain matter only above about 10 GHz?: Because raindrops become large compared with the wavelength. A drop is up to about 5 mm across, while the wavelength is 3 cm at 10 GHz and 1 cm at 30 GHz. Below a few GHz the drops are tiny by comparison and barely touch the wave. At 50 mm/h on horizontal polarisation, the specific attenuation is 0.17 dB/km at 5 GHz, 1.66 dB/km at 10 GHz, 9.8 dB/km at 30 GHz and 19.7 dB/km at 100 GHz.
Why does horizontal polarisation lose more than vertical?: Falling raindrops flatten into a shape wider than it is tall, so a horizontally polarised wave meets more water. At 50 mm/h and 20 GHz, horizontal loses 5.72 dB/km and vertical 4.53 dB/km. The gap shrinks higher up, to 19.7 against 19.3 dB/km at 100 GHz. That is why microwave links in rainy climates often run vertical. Circular polarisation sits in between, at the average of the two coefficients.
Is this the total rain fade on my link?: No. It is the loss for one uniform rain rate along the whole path, which is exactly what P.838 defines. Real storms are patchy, and on a long hop the heaviest rain rarely covers every kilometre. ITU-R P.530 (terrestrial) and P.618 (Earth-space) turn a rain rate into a fade for a given percentage of time by applying an effective path length. This calculator stops before that step. Gases, clouds and a wet radome add their own losses on top.
Which rain rate should I enter?: For link design, the rate exceeded for 0.01 % of an average year, R0.01, which is about 53 minutes a year. The presets give R0.01 for the 15 rain zones of ITU-R P.837-1 from 1994, from 8 mm/h in zone A to 145 mm/h in zone P. Much of temperate Europe falls in zones E to K, the wet tropics in N and P. P.837-1 is superseded. The current P.837-8 gives R0.01 from digital world maps, so use it, or local rain-gauge statistics, when the exact figure matters.
Why does the elevation angle change nothing with circular polarisation?: Because the elevation enters only through cos²θ · cos 2τ, and cos 2τ = 0 at τ = 45°. The H and V coefficients are then averaged with equal weight at any angle. For linear polarisation the elevation does matter. Raising the path towards vertical blends H and V more and more, until at θ = 90° every polarisation gives the same value.

Rain sets the fade margin, but terrain decides whether there is a clear path to fade at all. Waveshed checks line of sight and Fresnel clearance over real elevation data, so you know the hop is open before you budget for the storm.

Check the path over terrain →

Related guides

LOS vs RF (FSPL & ITM) — when a few dB of level decide the link, and when geometry alone does
Plan radio coverage (RF / ITM) — where a fade margin sits in a full coverage prediction

Sources & further reading

ITU-R Recommendation P.838-3 — Specific attenuation model for rain for use in prediction methods ↗ — edition 03/2005, in force: equations (1) to (5), coefficient Tables 1 to 4 and the check values of Table 5
ITU-R Recommendation P.837 — Characteristics of precipitation for propagation modelling ↗ — the presets come from Table 1 of the superseded P.837-1 (08/1994). The current P.837-8 uses digital maps
ITU-R Study Group 3 — Validation examples for Earth-space propagation prediction methods (Rev 8.3.0) ↗ — spreadsheet, sheet “P.838-3 Sp.Att”: this calculator reproduces all 16 test cases to 12 significant digits
ITU-R Recommendation P.530 — Propagation data and prediction methods required for the design of terrestrial line-of-sight systems ↗ — turns specific attenuation into a rain fade for terrestrial links, with an effective path length
ITU-R Recommendation P.618 — Propagation data and prediction methods required for the design of Earth-space telecommunication systems ↗ — the same step for satellite paths

Rain attenuation calculator (ITU-R P.838)

How it works

Fit the coefficients to the frequency

Combine them for your polarisation

Apply the rain rate

Multiply by the path length

Formulas

Worked example

FAQ

Related guides

Sources & further reading

Related tools

Atmospheric gas attenuation calculator (ITU-R P.676)

Cloud and fog attenuation calculator (ITU-R P.840)

Free-space path loss calculator

Link budget calculator

Weather radar dBZ to rain rate converter