The purpose of the ground plane, radial system or counterpoise is two-fold: (1) It provides a conductive path that diverts returning ground currents away from the lossy earth. In a sense, the ground plane can be thought to "shield" the radiating element of the vertical antenna from the nearby lossy earth. (2) In the case of a Marconi type antenna, it efficiently provides the "missing portion" that would be present in a self-resonant conductor, a half wavelength (or multiple thereof), isolated in space.
A vertical, inverted-L or tee antenna of an even number of quarter-wave lengths (half-wave, full wave, etc.) functions as a Hertz type antenna similar to an ordinary horizontal dipole rather than as a Marconi, therefore no counterpoise is needed to achieve resonance, but a radial system is still necessary to shield the radiating antenna from the lossy earth. It is a common misconception by hams that no ground system is needed for a half wave vertical. Yes it could be made to resonate, but 50% to 80% of the rf power would be wasted heating the soil in the vicinity of the antenna. Such an antenna used to be called a “worm warmer” by old timers. A radial system is still needed, even with a self-resonant vertical if it is mounted close to the ground. A half-wave vertical may be made to work efficiently if the bottom end is mounted a substantial fraction of a wavelength or more above ground on an insulated pole. The old WWV transmitters on the east coast employed half wave centre-fed vertical dipoles mounted on wooden utility poles.
The smaller the distance between the ground and the base of the vertical, the more radials are needed to effectively shield the lossy earth from the antenna. Ideally, a ground mounted vertical would have at least 60 quarter wave radials, and commercial installations usually have at least 120. OTOH, a VHF ground plane, mounted several wavelengths above ground, may work efficiently with only 3 quarter-wave radials because the vertical radiator is far enough away from the lossy earth that less shielding is needed, and the radials serve mainly to complete the resonant length of the antenna. At intermediate distances between those two extremes, it follows that more radials will be needed as the radial system approaches the ground, up to the maximum number when the radials are actually in contact with the ground.
It is not necessary to bury the radials; they may just as well be left lying on the surface for the ground system to function. The primary purpose for burying them is for protection from damage from surface traffic such as wayward feet and lawn mowers. But you don’t want to bury the wires very deeply; an inch or two is enough. Burying the radials more deeply than just slightly below the surface would tend to defeat the whole purpose of the ground plane in the first place, by inserting a layer of lossy earth between the vertical radiator and the ground plane. Avoiding the arduous task of burying thousands of feet of wire, newly-installed radials may be securely fastened to the surface of the ground using clips, stakes or other suitable mechanical means, and they will soon work their way down through the grass and become covered by thatch, and within a year or two will have buried themselves just beneath the sod and may actually be difficult to pull up by hand. Some broadcast stations, with FCC approval, have successfully deployed an elevated radial system, high enough off the ground to allow farm equipment and foot traffic to pass underneath, and experimental data suggests little or no loss of efficiency with far fewer elevated radials than the standard count of 120.
Beware of those commercially made, ground mounted, electrically short "no-radials-needed" amateur radio verticals; they are nothing more than a bogus rip-off designed to thin the wallets of the blissfully ignorant. The sales pitch would suggest the ability to violate the laws of physics. Better to save your money and construct a real vertical.
I would recommend studying the seven-part series of articles by Rudy Severns, N6LF, “Experimental Determination of Ground System Performance for HF Verticals” which appeared in January/February 2009 through January/February 2010 issues of QEX. The second instalment discusses a particularly interesting prediction using the modelling program NEC, verified in the experiments, that in ground systems with fewer than about 10 radials buried or lying close to the ground, there may be more ground loss when the radials are longer than about 1/ 8 wavelength, in other words, more copper = more loss! This is counter-intuitive to the logical assumption that making radials too long may be a waste of wire but otherwise causes no harm. The writer suggests that the increase in ground loss with longer radials may be due to the effect of resonances in the radial screen.
(to be continued in Part II)