Horizontal Heights and Sound Bites
A Short Note Relevant to NVIS Antennas

L. B. Cebik, W4RNL (SK)

All too often, we absorb information about antennas in the form of sound bites. The information is not necessarily false, but it is context-limited. However, the sound bite does not carry a trace of the context. Context, of course, tells us when we may accurately apply the information. Perhaps even more important, context tells us when not to apply the information. Misapplied information can lead us well astray of good antenna design and application.

Let's focus on just one of those pesky sound bites. "For a horizontal antenna, ground losses for any soil quality are (insignificant) (immaterial) (irrelevant) [pick any one or more of the 3 options]." Curiously, I find such statements not only in amateur literature, but as well in commercial brochures of antennas designed for commercial or government services.

My question is not whether the quasi-quoted statement is true. Rather, I want to ask when it is true, that is, under what conditions it is true--and to what degree.

A Context for Making the Sample Statement True

Perhaps the main arena in which the statement concerning ground losses and horizontal antennas may be true involves a dipole or similar antenna placed well above (say, 1 wavelength) ground. We can easily test the situation with almost any modeling software. Let's make a dipole from AWG #12 copper wire and place it exactly 1 wavelength above ground. The physical length will be about 0.48 wavelength to be resonant at the test frequency: 3.5 MHz.

Next, let's systematically alter the ground quality, starting with a perfectly reflective ground. Then we can go to very good ground (conductivity 0.0303 S/m, permittivity 20) and proceed downward through average ground (conductivity 0.005 S/m, permittivity 13) and wind up at very poor ground (conductivity 0.001 S/m, permittivity 5).

Table 1 shows us the results of our little survey in tabular form. From perfect ground to very poor ground, the total gain differential is only 0.65 dB. That difference does not make much of a difference. The table also lists the depth of the null above the lowest and strongest elevation lobe of the antenna pattern. As the gain decreases, so too does the depth of the null. Perhaps in numeric form that fact seems to make no great difference, so let's examine the elevation patterns for our surveyed ground types. In all cases, the antenna remains unchanged.

Besides outlining the test case, the graphic shows why it is dangerous to present antenna patterns over perfect ground if the ground of actual use is not perfect. First, note the relative strength equality of the first and second lobes. Second, note the depth of the null between the first and second lobes. Third, note the absence of any radiation straight upward. Then, compare the pattern for perfect ground with the other three patterns for increasingly lossier ground.

As the ground becomes lossier, the second lobe becomes weaker relative to the lower and stronger lobe. As well, the null between these two lobes becomes shallower. Finally, note that over real ground, there is almost always a component of radiation straight upward, a lobe or dome that grows stronger as the ground becomes lossier.

Nevertheless, in most amateur installations, we are only interested in the lowest lobe, the one most likely to coincide in angle with the prevailing skip conditions. This first lobe carries the small gain differential that we noted from the tabulated data. As well, the TO angle or angle of maximum radiation does not change with changes in the ground quality. Now let's add in one final fact. Since the ground reflections that form the lobes of the antenna pattern occur numerous wavelengths away from the antenna, there is virtually nothing that most amateurs can do about the ground quality in the critical zone.

We have noted a few minor differences created by ground quality for the performance of our horizontal antenna, but they are too small to make large arguments out of them. 0.65-dB is too small a gain differential for us to detect during operation of the horizontal dipole. Moreover, short of moving to a different part of the country, we can do virtually nothing about the ground conditions. So an a real sense, the quality of the ground beneath our horizontal dipole when it is 1 wavelength above ground is indeed a non-critical facet of antenna design, installation, and operation.

A Case in Which Ground Quality Can Make a Difference to a Horizontal Antenna

Suppose that we wish to operate in the Near Vertical Incidence Skywave (NVIS) mode for short-range communications. We can use our near-resonant dipole simply by lowering it to the height above ground that yields maximum gain straight up. Of course, the pattern will have a certain beamwidth both broadside to the wire and off the ends of the wire. So we shall not be restricted to talking to our next-door neighbor. Although most NVIS activity on our 80/75-meter band occurs in the SSB portion, we shall leave the test frequency at 3.5 MHz for consistency with the data on the antenna when it was 1 wavelength above ground.

As we lower the antenna, 2 important questions come to mind. 1. At what height shall we place the antenna? 2. What performance can we expect from the antenna? The answer to both questions is the same: it all depends on the quality of the ground. As a test, I placed the antenna at a height of 0.185-wavelength (just about 52') above ground. I checked the maximum gain upward for perfect ground and for several lesser quality grounds. Next, I searched for the height (in 0.005-wavelength increments) at which the antenna yielded maximum gain. The results of the small investigation appear in Table 2.

Each part of the table tells us something useful. With the antenna at a constant height, the quality of the ground alone created a nearly 3-dB differential in gain from the best to the worst tested. When signals may be very close to the threshold of being detected at all, 3 dB can make a significant performance difference. Second, the better the soil quality, the lower the antenna for maximum gain. Fig. 2 graphs the 2 gain curves--one for the constant height and one for optimized height. Only over salt water and perfect ground does the difference become at least visually significant. So for most situations, a height between 0.175 and 0.185 wavelength makes a good compromise. In practical measures at 3.5 MHz, that is 49' to 52' up.

Fig. 3 overlays the elevation patterns for very good, average, and very poor soil. The gain differential is evident. Also clear is the fact the a dipole has a significantly greater beamwidth broadside to the wire than along the axis of the wire. This feature of a dipole in NVIS service is useful if I need bi-directional coverage, but it may hinder communications if I need uniform or omni-directional coverage.

Although optimizing the antenna height for ground quality over most soils is not especially significant once we have a good range with which to work, we can still go some distance in eliminating the gain differential between very poor and very good ground. We need a reflector. Simple one wire and multi-wire ground screens are simply not good enough to make a major difference. We need a screen to create a planar reflector. For most applications, the screen needs to extend at least 1/2-wavelength beyond the limits of the antenna array above it in order to perform effectively as a flat reflector based on principles derived from optics. In our dipole case, that requirement means a reflector 1.5 wavelengths long by 1 wavelength wide. The reflector can go directly on the ground, but not below it. You do not need a tight screen or solid surface. Chicken wire will do, although its lifetime may be limited. Fig. 4 outlines the screen used with our test dipole.

Table 3 compares with modeling results with and without the screen. The table omits perfect ground, since the screen is nearly touching ground. Note the very small difference in maximum gain from one end of the soil spectrum to the other. As well, note that the optimal height above ground virtually constant.

Table 3 also contains a result that may seem contrary to expectations. The gain tends to improve as the soil gets worse. Is something wrong with the model? Yes and no. Actually, NEC-4 is doing exactly what it says it will do. It is calculating gain based on the antenna array's geometry and placement above a homogenous earth below it. The reflection of waves occurs not only within the boundaries of the screen, but well beyond it. Hence, to some degree--limited in this instance--the pattern is the result of reflection off the copper screen and off the ground. Some of the reflections result in far-field cancellations. The better the soil, the stronger those reflections are. As the soil grows "worse," it actually is less conductive and begins to approach the status of an insulator, which is transparent to RF. However, the wire-grid beneath the antenna is not 100% adequate, due to the conflicting needs of having a workable file size, a screen very close to the ground, and a wire diameter that coincides with the wire-grid cell size to simulate a nearly solid surface. Hence, part of the deviation may be due to this shortcoming of the model.

Do not expect a real-world antenna to provide even the small bit of extra gain shown in the model. Real ground is highly stratified. Below the ground that someone has measured and entered into a chart lie other layers of ground that may have properties quite different from the near surface values. The most likely effect is that most soils will yield gain values close to the lowest values in the table. Since the total range is small, you will not be able to detect the differences.

Ground effects are not constant from one frequency to another, although general trends persist throughout the HF spectrum. As a sample, I scaled the dipole (without a screen) to both 1.8 MHz and 7.0 MHz. The results appear in Table 4.

Except for highly conductive salt water and for a perfect ground, the performance trends are clear. For very good down to very poor soil, increasing the frequency increases the optimum antenna height for maximum gain, when we measure height as a function of a wavelength. The differential is very small and makes little difference in practice, but it is numerically noticeable. More significantly, the higher the frequency, the lower the maximum gain for all soils up through very good. Hence, a screen of sufficient size is advisable wherever it is feasible.


The old sound bite about ground quality not being significant to horizontal antenna operation turns out to be true only under certain circumstances. If a horizontal antenna is high enough above ground, then the gain and pattern variations created by different soil qualities are relatively small and are largely beyond control.

However, for horizontal antennas very close to the ground--such as is necessarily the case with NVIS antennas--the quality of the ground can make a significant difference in antenna performance. The difference is great enough to suggest system improvements that will largely equalize performance, regardless of the soil quality.

In the end, sound bites are a poor way to store antenna information. Most antenna data requires a context to give the information sensible, reliable, and usable truth.

Go to Amateur Radio Page