Contacter l'auteur / Contact the author

Recherche dans ce site / Search in this site







Software review

The VOACAP interface, the "golden standard" in HF propagation prediction matter.

VOACAP propagation analysis and prediction program (I)

Foreword. This review was written in 2004 (with updates until today) at a time where NTIA/ITS provided (and continue to do) an executable version of the VOACAP engine that you can downloaded from Greg Hand's website. We will see at the end of last page that since 2010 an online version is available with new functionalities, preventing users the tedious task to select the appropriate model and entering all data manually.

However, the web interface is based on the same original engine that everybody can download if needed, and which review of menu items and paramaters is always interesting to know to well understand what are its performances but also its limitations. Therefore I keep this full review online.

Readers who prefer to test the online version without background can connect to VOACAP online.

VOACAP stands for the Voice of America Coverage Analysis Program. It is an ionospheric model predicting the expected performance of HF transmissions. It takes into account tens of parameters to support you in the planning and operation of long distance amateur traffic or broadcast transmissions.

Contrary to the online version which is much more complete, the original VOACAP (classic) that we are going to review is a "Point-to-Point" analysis tool. Point-to-Point means that the propagation is calculated not for a general coverage over a location or to provide a global status of the ionosphere, but rather for a specified path or circuit between a transmitter and a receiver.

VOACAP is based on the Ionospheric Communications Analysis and Prediction Program (IONCAP model), a famous scientific product released in the '80s by the U.S. National Telecommunication and Information Administration and free of rights.

Its history being associated to many researches and developments, if you discover the product for the first time it is not always easy to understand who's who, what are those institutions and acrynoms listed, and all these additional models that complete the product. Therefore a review could not be complete without introducing VOACAP in telling how all that began, because it's a long story, all the more interesting that the program is accessible to radio amateurs for many years, and is still considered by many experts as being the "rolls" and the "golden standard" of propagation programs.

Back in time

In the years 1920s and 1930s amateurs and professionals discovered HF and DX communications, and SSB. In the late '30s, many organizations were involved in the study of HF communications. A considerable effort was made in the U.S.A. and other countries to investigate ionospheric parameters and determine their effect on radio waves and the associated reliability of HF circuits. Soon a worldwide network of vertical incidence sounders was established to measure values of parameters such as critical frequencies foE, foF1, foEs, foF2, and h'F. After the first two-way commercial transatlantic transmissions, operators noted the influence of the sun on daytime propagation conditions. In addition, worldwide noise measurement records were started to understand what phenomenon disturbed these communications, and steps were taken to record observed variations in signal amplitudes over various HF paths. These analysis and researches extended over half a century up to recently include soundings and satellite measurements. Of course these studies are always pending. Thanks to these data recorded at all levels of the ionosphere (50 to 800 km or so) and under all working conditions, scientists have got a better view of propagation and how the different parameters interacted each others.

Among the tens of discoveries made during these experiences like Van Allen belts, the South Atlantic Anomaly or sporadic plasma clouds, sientists have discovered that HF system performance is related in a very complex manner to solar activity, time of the day, season, and last but not least to the details of the shortwave path, like the ground properties and noise. Thanks to the first mainframes and their famous punch cards, scientists translated these discoveries in equations and developed the first ionospheric models.


In 1978, the Institute for Telecommunication Sciences of the National Telecommunications and Information Administration (NTIA/ITS) released for the U.S. Army a program developed by George Haydon, John Lloyd and Donald Lucas called the "Ionospheric Communications Analysis and Prediction" program, IONCAP for short. Written in Fortran77 for mainframes, it was designed in modular format to allow more flexibility for development of models for the key parts of the program, as many extensions that we should call today "subroutines" or "add-ons". Earlier much of the analysis had to be handled manually, a very time consuming process. Thanks to IONCAP scientists and engineers could get result in a few minutes with a greater accuracy.

Most large scale ionospheric data available in IONCAP were collected during the International Geophysical Year (IGY) between 1958 and 1963. Data were then reduced to look up tables which were valid for 2 or 3 month periods at the even hour. These maps were dedicated to the estimation of critical frequencies foE, foF1, foF2, as well as to the F-days distribution, M-3000 (for the MUF) and excess system loss tables. In addition, radio noise was mapped in 3 month groups and 4 hours time interval blocks, the original data of the an hour basis being lost, remain averages.

This way to create a model using averaged values was subject to many comments until it was decided by George Lane from the Voice of America broadcasting company, and the Signal Corps Radio Propagation Agency engineers to force the program to give a smooth transition across the days of the month. This is in this way, using statistical values, which accuracy is no more to demonstrate, that the concept of reliability is born. IONCAP could quite accurately calculate the distribution of the day-to-day variation and forecast for example that such conditions 'd be met three days at a specific hour during the month. But it was unable to define when will they occur during the month.

In parallel John Wang at the FCC developed a medium wave prediction model suited for long distance paths into darkness that was adopted by the ITU-R. However, IONCAP and Wang's model using different semi-empirical data set and functions, from a pure statistical point of view, both models were not compatible and both theories were never merged together.

IONCAP was probably the best studied propagation program and, until the late 90's was arguably considered by some engineers as the best basis for judging other programs. However the source code has evolved but the concept is stayed the same.

In 1983, George Lane cleaned up and corrected IONCAP to retain all of the theory as put forth by Lloyd, Haydon and Lucas >and developed a new model for the account of VOA. Two years later VOA adopted the new version of IONCAP as the approved engineering model to be used for broadcast relay station design and antenna specification.

Then VOA funded the U.S. Naval Research Laboratory (NRL) to make specific changes to the IONCAP methodology, and renamed it, as expected, to VOACAP so as to avoid confusion. That version of VOACAP was completed in April 1993 and distributed to participants at IES 93 (Ionospheric Effects Symposium May 1993, Alexandria, VA., USA). Simultaneous to funding NRL to enhance the model, VOA also funded the NTIA/ITS to enhance the user interface. A few months later NTIA/ITS had converted the program in a DOS-based application.

After Microsoft released Windows 95, on May 15,1996, VOA released the first official Windows 16-bit version, also called VOACAP. All previous released were Beta test versions. The next year the Windows NT 32-bit version was released. Today OACAP is at release 04.0324W and runs on all Windows 32-bit plateforms.

For years the development of this program was an endless project and new versions were released from time to time. Unfortunately, today all future development of this software will be limited to the Windows version and currently there is however no plan to modify VOACAP further.

VOACAP comes thus in several versions under different names : VOACAP, VOAWIN and HFWIN32, this latter being dedicated to Windows 32-bit plateforms although previous versions work fine in these environments as well.

Actually VOACAP and alike (e.g. WinCAP Wizard, ACE-HF, Ham CAP, etc) are simple graphic user interfaces, GUI or shells as say programmers, added to create a user-friendly layout between you and the VOACAP engine that processes data.

In the meantime, as expected, additional models have been added to the VOACAP engine like ICEPAC, S_I_VOACAP, VOAAREA, REC533, HFANT, etc, as many very accurate models that deserve your attention and that we will review later.

In parallel geophysicists have worked on a more complete ionospheric model called the International Reference Ionosphere (IRI) from which have been extracted several specialized models like the F2-peak model that we often find in many amateur propagation programs.

Today the VOACAP engine is used by tens of programs that have tried to extend the user interface. You will find it to HF broadcasters like VOA, BBC, ITU, HFCC, military, defense contractors, airliners, users of HF Email – Sailmail / Airmail, BPL and at least in the radio amateurs' ham shack.

This close our short history of VOACAP. Now it's time to speak more seriously, HI!

Download, support and installation

As of October 1998, NTIA/ITS does no longer accept funds to register VOACAP. As they quote on their website, that means that the program is only available to users thanks to an executable to download via the Internet. NTIA/ITS provides also information and the program (engine) at respectively these both addresses : input screens and outputs and the Windows 32-bit version.

If you require a software change, in the past there was a way to fund ITS to develop your project but nobody took advantage of this opportunity.

As stated in a popup menu that opens just after installation and in the readme.txt, Greg Hand who handled the officious support retired : "In February 2005, I have retired from the US Government. Questions, comments, suggestions may be addressed to Greg Hand Responses will depend on how much fun I am having!"

For short, if since 1998 there was no more warranty of results and accuracy, today the sole support to expect for the Windows version is the user assistance given by Christopher Behm from ITS through email, and it can only be done on a "time-available" basis.

Today, if you want a change, knowing that Jari Perkiömäki, OH6BG/OG6G, and his team have developed the web interface, before any action contact him first.

The latest Windows version of VOACAP "classic" is named "HFWIN32" and the final DOS version should only be used if you are really desperate and can't find another more recent version, HI! The latest fix for the Windows 32-bit version was incorporated in February 2008.

The main files packaged with the VOACAP engine. VOACAP, ICEPAC and REC533 are the main input/ouput for the Point-to-Point models. In addition HFANT permits to explore radiation patterns of any predefined or custom antenna.

The installation process is intuitive and very fast. After you downloaded the 4.4 MB EXE file, run it to install it in a specific directory. The program setup will automatically start. During the process it calls the Tarma Installer to prevent problems with creating the desktop icons on Windows systems. The program requires also to be installed in a NON long file name directory and without blank space, thus not under "\Program Files" or any other folder using more than 8 characters otherwise some modules will not start (thus use e.g. \itshfbc or \voacap).

All the application is saved under a folder name \itshfbc. If you open this small folder you will find 17 subdirectories gathering 12 MB of data among them VOACAP and additional models ICEPAC, REC533, VOAAREA, HFANT, etc. If you need free space and only use the VOACAP module, you can delete some data set files (e.g. additional antennas, extra locations, etc), including many samples and output files that you will never use. But let them there for the moment.

IONCAP heritage

Apart some approximations that have been corrected, VOACAP does not differ much from the IONCAP interface developed by John Lloyd in 1978, and inherited thus of most of its drawbacks but also of its flexibility and power. Like in IONCAP, there are various models to estimate the absorption, noise level and reliability of a service (mode) at some frequencies.

VOACAP deals not only with statistics of F-layer ionization (e.g. through the MUF and the like), but also down lower where absorption and noise have their origin. So, like IONCAP, 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.

Another interesting heritage is for example the antenna analysis program, HFANT. This external module is used by VOACAP each time that the path, the gain of the antenna and the takeoff angle of the specified signal need to be considered. With time, from a simple module able to process simple isotropic vertical or various dipole designs, it became a complete program able to take into account many antenna geometries from the longwire to the beam and it can compete against other programs offering more complex functions. In addition HFANT allows you to create your own antenna design and to check its properties in injecting its parameters into VOACAP.

Where IONCAP program is still more powerful is when you call the ionospheric parameters subroutine (the first calculation method). Explicit electron density profiles are not calculated by mathematical approximations but rather in the form of look up tables, thus using predefined data. These tables gives excellent results and, better, increase the computation speed for very complex systems, allowing the use of alternate ionospheric models in VOACAP.

Like IONCAP which models involve both paths longer than 10000 km and a "single hop" method for distance less than 3000 km, VOACAP uses the same functions to consider all possible ray paths for the specified circuit. However, in the specific case of DX where extension of paths require three or more hops (over 5 to 6000 km away depending on solar conditions), the system concludes that a path is available, which is far to be the case if we take into account the various reflection modes and degree of ionization of each layer. Therefore VOACAP uses a correction for these multi-hop paths longer than 10000 km, often important for DXing. It achieves this using Method 21.

On another side, if a propagation path does exist in a circuit, it has been shown empirically that usually these circuits are dominated by what we call "control areas" which are the regions within 2000 km around each end of the circuit (transmitter, receiver). At the transmitter location the control area allows the sky wave to do its first hop up to about 2000 km away, while the control area at the receive location allows the remaining signal to be returned after the last hop. In between, the path is simply characterized by a simple loss per distance statistical function, the noise and signal statistics being the same for all paths. If this approximation gives good results, we must recognize that the VOACAP model is more complete and thus more accurate for shorter paths, say below 5000 km or 3 hops.

Unlike IONCAP that divided the year into twelve months without considering days, VOACAP uses a decimal integer (e.g. 25 December is 12.25 instead of simply 12 in the previous model). The improvement avoids rough average and possible inaccuracies using data of the current and the next month together. But IONCAP used also two additonal coefficient tables for the ionosphere, CCIR/Oslo and URSI/88 to get more accurate predictions.

IONCAP developers corrected the predicted values using the CCIR database by using Transmission Loss tables and determined the variability of the MUF over the month using the F-tables. These corrections were important because they are based on predictions calculated with CCIR data and substracting them from the performance recorded between 1958-63. Differences were collected on these two sets of databases. The changes apply to output parameters calculating the signal power, SNR, reliability and the required power gain. But currently, if you use a model considering only the F-region, and do not recalibrate the predictions against measurement, nobody call tell if the correlation is maintained or not. You can made an improvement or ruin your prediction. VOACAP inheritated of both coefficients tables but they don't work in a similar way. CCIR was calibrated to work with the month only while URSI/88 requests the month and the day to establish a prediction. URSI provides also a linear fit between the SSN calculated during the minimum and the maximum or the solar cycle, and consider data over the oceans as well. But VOACAP doesn't support the day-of-the-month entry because URSI coefficients have not been validated during the development phase. This is one of the major error that might do users when they request a forecast. We will come back of coefficients later.

The smoothed sunspot number (SSN) is also used but, under specific conditions, it can be replaced by updated daily value instead of using monthly smoothed average. However, one must say and underline twice that when the sun activity is quiet with a low SSN, its value doesn't matter much, while near the paroxysm of its activity, when the sun is very active, differences can be large and conduct to false predictions of signal strength and circuit reliability.

In fact using median values like SSN must be considered as an advantage rather than a drawback. If we take the SNR Distribution tables or Excess Gain tables for example that are internal to the model, they were conceived and calibrated by VOA engineers for years from many listeners reports against a wide range of solar and geomagnetic conditions and at the higher statistical levels (higher required reliabilities, SNRxx). This is for these reasons that any attempt to enter daily sunspot numbers for example instead of the SSN can cause inaccuracies in VOACAP predictions. Therefore ITS recommend to use SSN - and specially the one from NGDC - and no other "equivalent" solar indice. We will come back on this important subject later too.

Heritage of the past, VOACAP does not either include input for planetary indices (A- and K-indices). This lack gives rise to inaccuracies in disturbed conditions, mainly at high latitudes as well as for top band predictions where geomagnetic and gyro-frequencies effects are not taken into account. However some newer interface do include corrections for the geomagnetic effect when high latitude paths are involved (e.g. ICED model used in ICEPAC) but indices are not explicitely defined and predictions are sometimes less conform to the reality than using VOACAP. Only some VOACAP-based competitors ask explicitely for geomagnetic indices.

At last, heritage of the first Fortran program using punch cards, like in IONCAP, when you request a report, VOACAP as well as ICEPAC and some VOACAP-based competitors output on screen an austere text file as long as an old listing. But as long as charts are available and can be printed, these reports become almost obsolete. Hopefully, conversely to IONCAP, the input text file that mimicked the punch card format has been replaced by a user-friendly interface that we are going to detail in the next pages.

Now that we know VOACAP's origins and heritage, come back to the present, and launch the program, Enter.

Next chapter

Main menu and settings

Page 1 - 2 - 3 - 4 -

Back to:


Copyright & FAQ