Propagation via an aurora

Last Updated on November 4, 2026 by John Berry

Propagation via an aurora is similar to that via the Moon and that via a meteor trail. Both exploit reflection and both therefore involve estimating the effectiveness of that reflection.

In the case of the Moon, we have a significant body of knowledge stemming from empirical measurements. That body of knowledge comes perhaps from the fact that the Moon is always there. So we now know the path loss to the Moon and back. Meteors and auroras are fleeting and hence our knowledge is much lesser.

There’s similarity too with reflection from the F2 Region of the ionosphere since that also occurs from elevated density of electrons providing pseudo-metallic layers. It’s just that in the case of auroras, the reflecting layers are broadside on and to my knowledge there’s been little work on reflection mechanisms and reflection loss.

This page builds a reflections model of auroras such that the path budget can be built and the signals environment to the aurora and back can be understood.

The geometry

The diagram below explains the geometry. A radio amateur at a transmitting station in Scotland points their antenna toward an aurora (typically north). Their antenna has a HRP 3dB beamwidth of around 34 degrees. If the southernmost edge of the aurora is over Iceland, they are around 1,300km away from the aurora as reflector. This suggests that the useful reflector is around 800km wide. And we do know that the vertical height of the E Region in which reflection occurs is around 10km. That gives a reflector area of about 8,000 square kilometres. For a much closer aurora of say 500km, the area drops to 3,000 square kilometres. These dimensions are  shown as w and h in the diagram.

The TX station is at distance d_TX from the aurora. The RX station is at d_RX. We don’t actually know how deep inside the aurora the reflection occurs. The Breit & Tuve theorem which I applied to ionospheric HF propagation suggests that the process is equivalent to mirror type reflection from a single virtual height. Since the aurora is broadside to the propagation, we can apply this to postulate virtual distances d’_TX and d’_RX to the problem. However, since we don’t know how deeply the wave goes before it is reflected, we can only guess at these virtual distances.

Disco ball analogy

The scenario can also be likened to that of reflection from the Moon. There, the analogy of the disco ball applies with a plurality of reflectors bouncing off in all angles. Despite the Rayleigh roughness criteria not being obeyed, many amateurs refer to this plurality as scattering or backscattering. Strictly it is not wave scatter, though the term has stuck. Like the model of Moon reflections, we can estimate the area of the aurora that provides useful reflections. This varies with wavelength and is around 50% for VHF wavelengths.

Unlike the Moon, however, there are multiple disco balls, or rather disco cylinders. It’s difficult to estimate how this multiple cylinder model will affect the useful reflection area.

The final element of reflection is the reflection coefficient. For the Moon, this is estimated as between 7% and 11% and varies with wavelength. The Moon is a lump of rock and that perhaps explains the poor reflectivity. Radio auroras are characterised by high electron density, and our knowledge suggests that when this is high enough, pseudo-metallic layers exist. Like the insertion loss of the F2 Region, auroral layers are likely to have low insertion losses, and hence can be quite efficient reflectors.

Billboard reflection

The billboard reflection model describing the aperture of a large reflecting object has been used for many years by microwave engineers to direct energy over hills and round terrain obstructions. It’s simple to apply and I use it here in absence of any other available approach. It’s given below.

Billboard Gain, G_B=10 log₁₀(4π⋅A_eff/ λ²) dB

A_eff is the effective area of the billboard – the physical area times any modifiers for reduced efficiency and effectiveness (see below). λ is the signal wavelength.

For an auroral area of between 3,000 and 8,000 square kilometres, the billboard gain works out at between 93dB and 97dB. Without such a high number, the bistatic path would not work. Adding the two Free Space Path losses gives around 270dBi which gives a path loss far beyond the System Value. Adding in the billboard gain makes it all work.

Practical QSOs

Radio amateurs exploit auroras to communicate with stations off-perpendicular. Pointing due north into the aurora returns a collinear reflection. But pointing off-perpendicular by several degrees or tens of degrees, exploits the notions of scattering and diffraction to cause a reflection into Norway, for example. The state is shown in the diagram below. Station A can communicate via the aurora with station B.

We see in the above diagram Station A to the south of the UK pointing north.

It’s likely that the efficiency of the reflection off-beam will be somewhat lower than those more collinear. This is borne out in practice. It takes a strong aurora to yield QSOs off-beam. For every station, there is a locus of points where reflection from the auroral columns is possible. The strongest returns will be collinear, due south. For a QSO, station B would need to recognise that there’s an aurora in progress and point its antennas somewhat west of north to take advantage of the geometry.

I’ve built a calculator to model all these effects and return a received signal and fade margin. If the fade margin above receiver threshold is positive the paths will work. If negative, no communications will be possible.

Importantly, this calculator is for educational purposes. It’s not intended to define exactly how auroral contacts happen. It provides instead a thought experiment into the various characteristics of auroras.

Here’s the calculator.

The variables in this calculator fall in two camps. First there are the characteristics of the aurora itself and its ability to reflect signals. Second, there are all the characteristics of the stations and the geometry of the scenario.

Aurora characteristics

The billboard area of the aurora is estimated from knowledge that reflections come from an E Region layer that is likely about 10km thick. The width of the aurora is then dictated by the aurora itself, but also by the antenna beamwidths. The width therefore also depends on the distance of the stations from the aurora – greater distance away means a wider reflector. An area of 5,000 square kilometres is a good starting point.

The useful percentage of the reflector will be around 50% for a collinear reflection at 144MHz. This will reduce off-beam. It’s difficult to estimate the percentage off-beam but perhaps 25% might be a good starting point. The reflection coefficient is likely to be around the same as the insertion loss quoted for the F2 layer when considering ionospheric reflection. This is around 3dB.

The foreground losses for both TX and RX reflect the fact that few radio amateurs have full first Fresnel Zone clearance and hence there will be diffraction losses in the path, particularly at low VHF frequencies.

Setting these coefficients allows useful QSOs of distance TX to aurora of 1,300km with RX to aurora of 1,000km. This is sensible. What’s needed now is to exercise these variables in light of experience of QSOs.

Other characteristics

I’ve defined Data Processing Gain and Threshold Degradation on other pages. My page on Data Modes and the Shannon Limit gives processing gain values in the High Sensitivity column of the data rates and Eb/N0 table.

Select an appropriate receiver threshold from your manufacturer’s information. Set a threshold in dBW for CW or SSB. When using Q65 or other data system, input an SSB threshold and then modify this in the data processing gain field. But remember that the data processing gain is usually referenced to the SSB noise floor and not to the SSB sensitivity. You will need to adjust by adding 10dB to the data processing gain for a sensitivity given for 10dB S/N for example.

Giving comments

I’d welcome positive, considered comments. Have I got the overall theory about right? Can an aurora be modelled in this way? And are the variables I discuss the only variables, or are there other variables? And how do we move to a point of sound physics in auroral propagation?

Of course, what’s needed now is to link these variables to the geomagnetic storm coefficients Kp and dst.