Topband: Low vertical and high inv-L

[Ulrich Heuberger]

Gentlemen,

I wonder if I could get some advice about the following situation:

My inv-L is 20m up and 33m horizontally, and has 55 radials. It works
fine into all directions. However, it is located on rented area, and
sooner or later I have to give it up.

My vertical is in my own garden, a tower of 30cm diameter, 42m high,
shunt-fed, 65 radials from 45 to 270 degrees. It is located 48m from the
inv-L and its base is 8m LOWER than the inv-L. To the north-west (USA,
Carribbean) the GP has 6 elevated radials, 6m high, some of them sloping
a little UP, two of them connected to the radial-system of inv-L. 

The inv-L is located towards NW seen from the GP. 

The horizontal part of the inv-L is suspended to the tower at 28m. The
ground is elevating from the GP in direction of the inv-L, first slowly,
the last meters are steep.

Although the GP is 14m higher than the inv-L, it ALWAYS is outperformed
at signals from US/Carr./even England. I observed W8JI some days ago
over some time, and the difference typically was 10-12 dB. 

Into the other directions, the GP is usually 6-8 dB BETTER than the inv-L.

Questions:
Is there anything that could be done to improve the GP towards
North-West with the existing feedpoint?
Any chance that the inv-L is better into NW because the GP acts as
reflector? (Detuning the GP doesn't show any difference)
If the feedpoint is elevated to the same level as the inv-L (like
grounded tower feed system by N4KG), would there be an improvement
towards NW?
Can elevated radials compensate for irregularities in the ground below them?
Has anyone a similar problem, and how was it solved?

Thank you very much in advance.
73s Uli, LA0CX