Unless looking for NVIS at lower frequencies, a horizontal at approximately 1/2' wavelength above moderate to decent ground can return up to ~8.5dBi of gain. The problem for lots of us is getting a lower HF horizontal dipole for 160/80/40 that high, even if perhaps having the space lengthwise.
That said....
Want a ground-mount HF vertical without the huge radial field? Make the vertical a halfwave fed at the base with with a 49:1 or 64:1 unun. Now all ya' need is a few very short on-ground radials (and/or common-mode on the feedline) to satisfy Kirchhoff's first law.
If antenna height is an issue, as a halfwave on even 40m is ~66', look into deploying the halfwave as described above instead as an inverted-L. It might need a little tweaking to the length and/or unun ratio, but either seems easier than deploying dozens of ground radials IMO.
Alternatively, raise the vertical and use a couple or more elevated tuned radials. I keep meaning to add those to my elevated 31' vertical for improved ~7MHz tuning, but it is a receiving antenna, so it is just grounded to the existing mast and guys for now. If I raised it higher, I could actually even use tuned guys as elevated radials.