Contacter l'auteur / Contact the author

Recherche dans ce site / Search in this site







HF Propagation tutorial

F2-region critical frequency map showing some "islands of ionization" along the equatorial region called the "equatorial anomaly".

by Bob Brown, NM7M, Ph.D. from U.C.Berkeley

Down-Sizing of the Ionosphere (V)

In the previous page, I showed one sample contour of a global map of the F-region, for 10 MHz when the SSN was 137. You can go back to the map to see how it spilled over into the hours of darkness. But that was only one contour. So the question comes down to the rest of the map, what other contours were like and their limits in critical frequency.

Looking at the sample contour, it is easy to think that parts of the globe closer to the sub-solar point would have higher values of critical frequency, up to 16-17 MHz. After all, the sun was more overhead for there and the solar UV had less atmosphere to penetrate. But at larger zenith angles, particularly toward the polar regions, the critical frequencies would be lower, going down to 6-7 MHz. All that for a SSN of 137.

What about lower SSN, say toward solar minimum? Then, for the region where the critical frequency was 10 MHz earlier, you can just put in 5-6 MHz and at higher latitudes, you can put in 3-4 MHz while at low latitudes, the value is 11-12 MHz.

But whatever the SSN, the highest critical frequencies are always found at the lower latitudes. As a practical matter, that is an explanation why contest DXpeditions go toward equatorial regions; the bands are always open there and it is just a matter of how far their signals go poleward before running out of sufficient ionization.

So I like to say that the low-latitude regions are the most robust of the ionosphere. But there is a difference between "robust" and "ROBUST", say for solar minimum and solar maximum.

Before getting to that, I should point out there are "islands of ionization" at low latitudes, as shown by the additional contours given at right. These islands of strong ionization develop in the afternoon/evening local hours. This is called the "equatorial anomaly" and has profound effects for propagation, giving rise to long, chordal hops on HF and DX on VHF

Those regions are a regular part of the ionosphere, day in and day out, and the high level of ionization there adds to the robustness that I spoke of earlier.

A few paragraphs earlier, I made mention of the fact that global maps of the F-region change with solar activity. One way of making these ideas more vivid in one's mind is to think of them like relief maps, with a "frequency surface" that rises or falls in height as critical frequencies change with increasing or decreasing SSN.

N-S cross-section of the F-region for two different sunspots numbers.

The quantitative side of that approach can be shown by means of a N-S slice through the global maps that one obtains for two different sunspot numbers as displayed at left, adapted from an original plot from the PropLab Pro program.

Those N-S cuts across the F-region maps show the two "islands" of the equatorial anomaly as well as the deep notch in between them. Also, it shows again the geomagnetic control of the ionosphere by the asymmetry of the ionosphere at 120°E, due to the fact that the magnetic dip equator is about 5 degrees north of the geographic equator at that longitude.

Admittedly, the above graphics are pretty crude but they cover the main aspects of the ionosphere - E-, F- and F2-region maps - showing how ionization is distributed and how it varies with changes in solar activity. It is within those regions that we are trying to propagate signals.

So we should lay down some great-circles to see where the paths are going relative to the ionization. The test, of course, is if the effective vertical frequency along a path is less than the critical frequency encountered. As long as that's true, propagation will continue; otherwise, the RF will penetrate the F-region and be lost. 

Looking at the last graphic, you can see that "the test" gets tougher at high latitudes where the critical frequency is on the low side, a few MHz. Thus, there will be angles at which the RF penetrates the ionosphere and is not returned to ground level. That is "skip", discovered by John Reinartz back in the mid'20s, and obviously gets worse at higher frequencies.

In that regard, there is one "side light" to that on the higher bands. Thus, it is quite easy to "pass the test" and work to the south on 21 MHz, for example, as the ionosphere is quite "robust" in the N-S direction. But looking at the last figure, one can see that the ionosphere is "puny" in the E-W direction, with very low critical frequencies. As a result, when chasing DX on 21 MHz, skip makes it impossible to hear the station east or west of you that got the South American contact that you were trying for.

At this point, our discussion comes down to exploring the aspects of the distribution of ionization, vertically and horizontally. The vertical distribution determines how signals are refracted or bent along a path while the horizontal distribution determines whether a hop is completed or how long it might be. There are two approaches we can follow, the rigorous one would be to trace ray paths through a model ionosphere while the practical one would be to use the model in a propagation program, looking at the critical frequencies at the two control points on a path to see what the MUF would be and whether one's RF passes the test.

Ray-tracing takes us back to the analogy between the flight of a baseball and RF across the ionosphere. Mathematically, the flight of the ball is worked out using Newton' Laws, with equations of motion in two or three dimensions. You should not be surprised if I tell you that equations of motion for RF can be worked out, with the ionosphere playing the role of gravity. So, like any baseball or even spacecraft, the methods of mechanics work with RF and the equations of motion solved, step by step, to find the path of RF. In that regard, the PropLab Pro program is outstanding; all you have to do is put in the locations of the terminii, the date and time as well as the sunspot number, and it solves those equations of motion and traces out the path of the RF. Just fantastic!

But there is one more thing to add; PropLab Pro also includes the role of the geomagnetic field in the equations of motion. At the upper end of the HF spectrum, that is not important as the QRG is large compared to the electron gyro-frequency about the field lines. But down around 160 meters, the 1 MHz gyro-frequency is comparable to 1.8 MHz and the effects of the magnetic field no longer appear to be negligible in the equations of motion. There are some interesting consequences for wave polarization as well as signal absorption. In addition, signals can get trapped in that valley above the night-time E-region and ducted to great distances with low loss. But we'll get to that later; first, MUF programs.

At left, the intensity of the vertical component of the world main geomagnetic field correlated at right with the electron gyro-frequency map created with PropLab Pro for AM bands from 630 to 1630 kHz and of course with the 160 m ham band.

Note by LX4SKY. The correlation between the geomagnetic field and the electron gyro-frequency (EGF) explains the propagation of the lowest band. This correlation requests some explanations. EGF is a measure of the interaction between electrons present in the Earth atmosphere and the vertical component of the geomagnetic field (Z-field). The closer a transmitted AM or SSB frequency is to the electron gyro-frequency, the more energy is absorbed by the gyro electrons from that carrier wave frequency. This phenomenon mainly occurs with AM signals traveling perpendicular to the geomagnetic field (especially along high latitude NW and NE propagation paths). This kind of absorption is always present and cannot be avoided.

Reference Notes

Originals of all the figures mentioned above can be found in my article, "On the Down-Sizing of the Ionosphere", that appeared in the July/August '94 issue of The DX Magazine. Also, the two main F-region maps are on p. 29 of my book on long-path propagation and also found in Davies' book, Ionospheric Radio (IEE Electromagnetic Waves Series, Vol. 31).

In addition, there are a number of ray traces shown in my Little Pistol book, illustrating skip and showing how RF hops vary with frequency as well as radiation angle.

Performance of ionospheric models

Now we are in a position to talk about propagation predictions. I say that as you understand that predictions require some sort of representation of ionospheric maps, both E- and F-regions, and a method that looks at how effective vertical frequencies compare with critical frequencies along a great-circle path.

I must admit that I have injected "effective vertical frequency" (EVF) into the discussion; you normally don't see that term when you read about propagation. In McNamara's book, Radio Amateurs Guide to the Ionosphere, he uses another form, "equivalent vertical incidence frequency", in his discussion but I find that just too wordy and besides, my choice of EVF fits the bill and tells the story. I hope you agree.

Anyway, we know the test which our RF undergoes as it ascends after launch: if its effective vertical frequency is less than the local critical frequency, it will be contained by the ionosphere and if not, it will go past the F-layer peak and be lost. The propagation prediction business has to do with how that test is carried out - to what approximation or detail the test is made and with what sort of model of the ionosphere.

Ham CAP propagation map for the 17-m band. Created by VE3NEA it is interfaced with VOACAP, and with DXAtlas in option.

I've already mentioned the control point method in which the test is made at the first and last hops on a path. That method was developed back in WW-II, by Smith in the USA and Tremellen in the UK, and was based on the notion that if a path failed, it was usually at one end or the other. 

I pointed out that works well as long as any hops in the middle of the path do not have LOWER critical frequencies. Beyond that, you should remember that the method represented a great step forward at the time, even though it was when ionospheric mapping was in its infancy.

So the control point method was based on an approximation and its use involved a database which was both limited and uncertain, at least at the outset. Nowadays, the database has improved quite a bit but still will undergo some revisions in the future as the Internation Reference Ionosphere is updated from time to time. Today the standard IRI model is the release IRI-2001.

I really don't know the details of the first uses of the control point method but I am familiar with some at the present time. For example, the pioneer program in amateur radio circles was MINIMUF from NOAA, with source code first published in QST in December '82. That method used M-factors, numbers between 3 and 4, for division of the QRG to obtain EVF for comparison with critical frequencies at about 2,000 km from the ends of the path; for that, MINIMUF used a database founded on oblique ionospheric sounding.

One can fault the source code of MINIMUF for not taking into account the earth's field, leaving out the equatorial anomaly and organizing the ionosphere only with geographic coordinates. Beyond that, the database was rather limited in scope. But MINIMUF caught the imagination of the amateur radio community and all sorts of accessories were attached to MINIMUF, ionospheric absorption and man-made noise, to mention just a few.

MINIMUF's shortcomings, the lack of geomagnetic control in the method and no consideration of radiation angle, placed it in a poor position to compete with other programs that came along and corrected those deficiencies. Here, I have in mind the work of Raymond Fricker of the BBC External Services. In the mid-80s, he published programs like MICROMUF and MAXIMUF which included the role of the geomagnetic field and put in radiation angles so one could compare MUF predictions for more than just the lowest mode. MICROMUF exists also in Pascal and an interactive web form has been developed by Pete Costello for Unix.

Propagation Prediction, aka PP, by DL6RAI taking advantage of FTZMUF2.

Somewhat later, the german club FTZ Darmstadt introduced a program name FTZMUF2.DAT, that used a grid point method to obtain critical frequencies from the CCIR database and used interpolation to obtain the spatial and temporal data for making predictions. They went on to show that FTZMUF2 gave a better representation of the CCIR-Atlas data for 3000 km MUFs than did MINIMUF. Beyond that, they incorporated FTZMUF2 in their own MUF prediction program, MINIFTZ4.

Note by LX4SKY. Four years later, in 1991, Bernhard Büttner, DL6RAI, also used FTZMUF2 in his own applicated named Propagation Prediction, PP. This is one of the first DOS application to display MUF and other signal strength in a colored line graph as displayed at left. Then in 1994 Cedric Baechleris, HB9HFN, released HAMFTZ based on the same grid point method.

But Fricker used an entirely different approach when it came to the database for his calculations; he used mathematical functions to simulate the CCIR database, now in the International Reference Ionosphere. Then he used the functions to calculate foF2 at the midpoints of the first and last hops in his programs, MICROMUF 2+ and MAXIMUF, as in the control point method. 

Those were the propagation prediction programs available until the IONCAP program developed in the late '70s by George Lane from VOA then by Teters and al. for NTIA/ITS was brought down to a smaller size where it could be incorporated in home computers. Unlike IONPRED, which Fricker's method was based only on F-region considerations - but that gave accurate results in its limitations - IONCAP deals with fluctuations of signal strength, it uses a D-region factor, and takes into account man-made noise. Today the only application always maintained and using a reduced set of IONCAP functions is PropView from DXLab suites.

Note by LX4SKY. In 1985, pressed by the broadcasters' interest, George Lane improved the IONCAP model, corrected some algorithms, added new functions, and after years of research and development created the famous VOACAP that was released free of right in 1993. Today VOACAP is considered as the best ionospheric model, the standard for comparison.

At left the Windows 32-bit VOACAP interface, at right ICEPAC. Both use a improved version of the IONCAP model rather than the International Reference Ionosphere (IRI). From IRI model has been derivated various smaller models to name electron density models, electron temperature models, auroral precipitation and conductivity models, F2-peak models, etc. IONCAP has also been improved and was incorporated in VOACAP (1985) and ICEPAC (1993) and "borrowed" by commercial products like ACE-HF Pro (2001) or WinCAP (2003) prediction models to name a few. The IRI model is used by very few software accessible to amateurs, to name PropLab Pro from the Solar Terrestrial Dispatch and DXAtlas (2004).

Then came series of programs, some as accurate as the VOACAP model for Windows 16 and 32-bit plateforms. Most of them used the new functions devised by Raymond Fricker and other scientists or directly the VOACAP engine without additional algorithms.

In any event, the upshot of the comparisons, is today that Raymond Fricker's programs and the improvements made by George lane are close in agreement with the International Reference Ionosphere (IRI), then came all non-VOACAP-based applications that give a rough estimation of propagation conditions, and far behind all DOS executable like MINIFTZ4 and other MINIMUF considered as the poorest and displaying often very few information.

But how well the underlying VOACAP database matches the real ionosphere (e.g. CODE GIM model) compared with IRI, the best representation available at the present time  ?

Real-time status of the ionosphere from GNSS Group, Belgium

Software  : CODE GIM model

In that connection, I undertook a study of how the mathematical F-layer algorithm in Fricker's MAXIMUF compared with IRI, not just for a path or two but over the entire world. Thus, foF2 values were calculated at intervals of 5° in latitude and 5° in longitude from Fricker's mathematical functions and compared with corresponding values from IRI. That method showed where Fricker's values were low, where high and an overall measure of his methods.

The result was that Fricker's method, when used to make a map of the F-region, gave good agreement over the entire globe with the values from IRI, point by point, but the agreement could even be improved considerably by the simple offset of 1 MHz added to the foF2 values calculated by his methods.

SNR calculated with VOACAP for a circuit between UA3 (Moscow) and ON (Wépion) on August 5, 2004 at 2200 UTC for 100 W output (SSN = 27, SFI = 85). VOACAP predicts a signal strength in Belgium of -140 dBW or S3 and a S/N ratio of 27 dB, thus quite weak. Figures matched signal reports given on the air.

Put another way, Fricker's foF2 map was very much like the map from IRI, with details such as the islands of ionization showing up as well as various aspects of geomagnetic control, but the critical frequencies were a bit low. All in all, I found it amazing! 

And that approach proves to be just another way of testing F-layer algorithms, seeing if they can make a good ionospheric map or not. MINIFTZ4's algorithm gets good marks in that regard but with problems from its interpolation methods while MINIMUF's F-region map has little resemblance to a real ionosphere on a global scale. That accounts for some of its erratic predictions for DXing.

Unfortunately, when I made my tests the F-layer algorithm of IONCAP was not available, so comparisons with the IRI remain to be done with VOACAP which sources are available from NTIA/ITS. Perhaps some of the VOACAP developers will do that in the future. But whatever the outcome, VOACAP is always the best HF propagation program and provides some of the other aspects of propagation prediction that are important. Thus, in addition to having methods for calculating MUF, LUF and other HPF, it deals with the range of values of critical frequencies resulting from the statistical variations in the sounding data.

Here, I refer to statistical terms like the median as well as the upper and lower decile values of critical frequencies from the sounding data. In a propagation setting, the median value of the data at a particular hour during a month would be one such that half the observed values lie above it and half fall below it. If a median value is used in propagation calculations, one obtains what is termed the Maximum Useable Frequency (MUF) for the path. The upper and lower decile values of critical frequency have to do with the 90% and 10% limits. Thus, the upper decile value during a month of observation is a frequency which is exceeded only 10% of the time, 3 days, while the lower decile value during a month is a frequency which is exceeded 90% of the time, 27 days.

When those values are used in propagation calculations, one then obtains the Highest Possible Frequency (HPF) and the Frequency of Optimum Transmission (FOT) for the path. A sample of that kind of calculation is given below (in MHz), for a path from Boulder, CO to St. Louis, MO in the month of January and when the SSN is 100 :

 GMT    FOT    MUF    HPF      GMT   FOT    MUF    HPF

  1     10.7    13.6    17.4       13     6.4     7.5     8.4

  3      7.4     9.6    12.0       15    13.0    15.3    17.1

  5      5.7     6.9     8.7       17    16.6    19.3    22.0

  7      6.1     7.4     9.7       19    18.1    21.1    24.0

  9      6.5     8.0     9.4       21    17.7    20.6    23.5

  11     5.0     6.1     7.2       23    15.9    18.5    21.1

Looking at those numbers, you can see that the HPF and FOT values lie about 15% above and below the MUF values. That should put you on notice; if the propagation program you use gives only MUF values, the real-time values for the ionosphere could differ by as much as +/-15%. And that is only from the statistical variations in the basic data; there are still the approximations in the method to worry about as well as geophysical disturbances.

But those remarks apply mainly to the higher HF bands; down on 80 and 160 meters, ionization is not a concern on oblique paths. Instead, noise and ionospheric absorption limit what can be done. And propagation programs are useless for those bands as the main criterion is darkness along paths, not MUFs. But the role of the geomagnetic field is important and affects the modes that are possible. All that in due time.

As for geophysical disturbances, those will be our main effort in next chapter and need not concern us at this point. We are really concentrating on the undisturbed ionosphere and its properties or modes, variable though they may be. And while still talking about the VOACAP program, it is worthwhile to note that its methods deal not only with the statistics of F-layer ionization, through MUFs and the like, but also down lower where absorption and noise become have their origin. So VOACAP has F-region methods which give not only the availability of a path, the fraction of days in a month it is open on a given frequency, but also D-region methods which give the reliability of a mode, the fraction of time the signal/noise ratio exceeds the minimum required for the mode.

This was not meant to be something just in praise of VOACAP but for me it is the best HF propagation analysis and prediction program that I have at my disposal in a point-to-point prediction perspective. True, there are other programs based on it and you will have to judge for yourself whether those programs meet your requirements or not. You should read the reviews out there, on this website, in QST and The DX Magazine, to get a feeling for what they can offer you in your pursuit of DX. If possible, check with a user to see if the program matches your goals or needs for DXing.

At this point, we've come to where ionospheric disturbances from the impact of the solar wind on the magnetosphere are of real importance. Needless to say, they add to the uncertainties that have been cited above. But in contrast to the statistical side of propagation, there are clues that help deal with the geophysical side of propagation. That will be our task in the next pages.

Next chapter

Propagation modes and DXing

Page 1 - 2 - 3 - 4 - 5 - 6 - 7 - 8 -

Back to:


Copyright & FAQ