Last Updated on December 4, 2024 by John Berry
A prediction asks the question, ‘Will I be able to communicate from here to there at a given time on a given date?’ In ham circles, perhaps this can be flipped to read, ‘What DX can I work today?’ In this page I give the background to ionospheric predictions, showing the engine of the prediction.
There are two tools popular in making predictions: Recommendation ITU-R P.533 and VOACAP. They are basically the same thing, though VOACAP inner workings are not so well defined. My experience is twofold. First I’ve used VOACAP. And second with Rec. ITU-R P.533, I’ve implemented it as a FORTRAN kernel in a C++ wrapper. Most people who offer HF prediction software simply wrap the old FORTRAN code.
The following is taken from the Recommendation.
Critical incidence frequency
Over many years, radio ionosondes have made measurements of three critical parameters: foE, foF2 and M(3,000)F2.
foE is the critical vertical incidence frequency for the E region. An antenna is pointed directly upwards toward the ionosphere. Then the highest frequency of transmission that is reflected from the E region is sensed.
foF2 is the critical vertical incidence frequency for the F2 region. An antenna is pointed directly upwards toward the ionosphere. Then the highest frequency of transmission that is reflected from the F2 region is sensed.
foE and foF2 are different and easily discernible.
M(3,000)F2 is the ratio of the maximum usable frequency at a distance of 3,000km to foF2. foF2 here is at the mid-point of a 3,000km path. It’s sensed by setting up a path. The highest frequency that is returned towards the distant station is determined that enables the communication. M(3,000)F2 therefore represents the optimum frequency for a 3,000km path.
I’ve given the figure below to show the parameters.
Sunspot number
Measurements have been taken for many values of smooth sunspot number (SSN), also known as R12. The SSN/R12 is a 12-month rolling monthly average number. For the purposes of prediction modelling, critical incidence frequency data is stored against R12=0 and R12=100. Typically, R12 lies in the range 0 to about 285. The median value is about 160 and this is taken as the limiting value in predictions, so if R12>160, R12=160.
It’s possible to use better indices to optimise predictions, like R5, the five-day rolling average sunspot number, in place of R12. Single daily values are not useful on their own.
Critical frequencies have been measured for many locations and paths across the world. Hence, relationships have been developed between the variables. Therefore, data can be made available for all points on the globe and all values of R12 between 0 and 160.
Dataset
The dataset is large and freely available from the ITU-R. I’ve used it here as a description of the background to ionospheric predictions. It contains coefficients for foF2, M(3000)F2, and foE. These are at the two levels of solar activity, R12 = 0 and R12 = 100, all against date and time (over many years). Coefficients for median foEs, upper decile foEs, lower decile foEs, h’F and h’F,F2 have been developed from these.
Median foEs, upper decile foEs, and lower decile foEs are the descriptive statistics of the critical incidence frequency. This is for the E region for sporadic-E. h’F and h’F,F2 are the minimum observed virtual height of reflection for the F region during day and night.
For reference, you’ll find that typical values of foE are around 4MHz. Typical values for foF2 are from about 5MHz (at R12=0) to about 12MHz (at sunspot maximum).
Ionosonde data has also been collected to enable the prediction of spread-F. Spread-F is a phenomenon where the ionosphere is re-shaped from two layers into one region at sunset. Spread-F enhances propagation for between seconds and hours for trans-equatorial paths. Since it occurs at sunset, it gives rise to the term ‘greyline’ propagation.
Making a prediction
Propagation prediction and modelling software used for HF uses the dataset described above.
Prediction software uses a simple premise. If the critical incidence frequencies and the R12 are known for the path (at each station, and at the mid-path), the path performance can be calculated.
Path performance at the required day and time is described by its operational maximum usable frequency (MUF). The operational MUF is sometimes referred to as the FOT, the fréquence optimale de travail. The descriptive statistics of the operational MUF are available from data, and hence the circuit reliability, can be reported. An example might be for 10% of time.
The value of R or SSN is an input to the calculation from space weather data for the day of interest. If relevant, the E-region screening (of F region modes) is also calculated and applied.
Calculating the receive field strength
As the saying goes, “it’s all in the geometry”. Given the MUF calculation, what we now needed is a path calculation. This will expand further in giving background to ionospheric predictions. And for that we need a path. The path is dependent on the likely point from which the signal is returned to Earth – the mirror reflection point. That determines the height – and the elevation angle. The mirror reflection height is given from the above work. Then the angle is calculated for different modes, E and F2.
Field strength & path loss
The field strength of the received signal is given by:
The result is in decibels relative to 1μV/m. This can be easily converted to signal level in μV for a given antenna system. f is the TX frequency, Pt is the transmit power, Gt is the TX antenna isotropic gain and Lb is given by:
Where p’ is the virtual slant angle range in kilometres.
Lb is the path basic transmission loss in decibels for the mode considered, E, or F2.
You’ll notice here that the first three terms comprise the free space loss equation, and I’ve discussed this elsewhere on this site.
Modifications
Li is the absorption loss for an n-hop path assuming two penetration points per hop. Li is given by a complex set of maths including terms for angle of incidence, penetration points, the electron gyrofrequency, the solar zenith angle, absorption factor at local noon, and diurnal absorption factor as a function of magnetic dip.
Lm is a term to cover losses above the MUF, considered separately.
Lg is the ground reflection loss at the intermediate reflection points. This is assumed to be Lg=2(n-1) dB or 2dB for a two-hop path.
Lh is a term accounting for auroral and other losses.
Lz is a coverall for other losses not descibed above – effectively a constant, recommended as about 9dB.
So, in summary, the received signal is determined by the free space path loss. But, that’s modified by a number of terms covering losses due to the coefficients of reflection at ground and ionosphere reflection points.
The figure below shows the concept. Click the image for a brief video explanation. Or skip if you already watched by clicking the equation above.
In the discussion elsewhere I give the free space path loss equation. I note that in its simplest, the loss at HF is simply this applied to the slant paths of the multiple hops. Above I’ve shown that there are some local corrections to this needed.
Other solar activity indices
Some making ionospheric predictions using solar activity indices prefer to use the Ottawa 10.7cm flux instead of R12. The 10.7cm flux (the F10.7) is reported in units of 10-22Watts per m2 per Hz and measured at 10.7cm wavelength or 2,800MHz. Its units are known as solar flux units varying from about 50 to about 300 over the course of a solar cycle. The 10.7cm flux and R12, the smooth sunspot number, are directly related, and some ionospheric prediction applications can take either as input.
Any preference tends to centre on ease of measurement rather than accuracy – the F10.7 can be measured reliably in any weather. Since radio hams aren’t doing measurement of R12 and F10.7, the index used by them matters little.
Summary
The assumption is that the normal ionosphere changes consistently over an 11 year cycle. Hence, it behaves consistently and will continue to do so. The assumption is that this background to ionospheric predictions will continue to apply.
An empirical data set has been assembled (over many years). That relates the behaviour of the ionosphere in supporting radio wave propagation to geographic location, time, and sunspot activity. However, this data is generalised.
That generalised data is then applied to the specifics of the path from one station to another. The application is at an intended date and time and the path loss and angles calculated. The R12 or SSN value is input for the day in question from space weather forecasts.
Once the path loss is available, assertions can be made about the viability of the path. The questions, ‘Will I be able to communicate from here to there?’ or, ‘What DX can I work today?’ can be answered.
I’ve used the concept here of a normal Sun. And normally a viable prediction can be made. But the Sun is quirky, and there’s a chance that when we test the prediction and operate, a flare will blow on the Sun’s surface. The result can be one of several effects, all reducing the our ability to communicate on the day.
Note that atmospheric and man-made radio noise is always well above the bench value of receiver threshold. It therefore dictates the real receiver threshold at the receiving location. Therefore, data must be included for this in determining what will and won’t work. It’s not enough on its own to predict received signal if that signal is then below the local noise.