SV1BDS George, it is not 'that simple' (because you/we? see the noise floor +5 dB 'in Greece'). Suppose -in your situation- the tpx gain will be decreased with another 5 dB, you will see (with your setup) 0 dB tpx noise.
In other words, you can't see an increment due to the tpx.
So the noise of the tpx will be the same as your system noise. However, these 'noises' are not correlated, but ... add up stochastically, meaning you've 3 dB more noise (system_noise + tpx_noise = 2 = +3 dB). That means that the detoriation of weak(er) signals is more than 3 dB (refer to the 'noise figure formula' of a receiving system).
The tpx gain ('noise') is set up in such a manner that the given user requirements of the link budget (dish size, or better ... G/T, see ITU ES'hail-2 document) are met.
(I commented a few times about this on this forum, but can't find it now .. ; -) Iirc the ground station requirements are around 75 - 80cm dish diameter in the 'average coverage area' (<-- my words) when the uplink power (using the same diam dish) is around 5 - 10 Watts for 20 dB C/N @ 2.5 kHz.
So, if you use a 1.6m dish you receive these 5 - 10 W uplink with 26 dB C/N, and an increment of the tpx noise. As a consequence, with this 1.6m dish you can reduce your uplink power with 6 dB : -)
Summarized: dish size around 75-80 cm diam, and 5 - 10W uplink power has to produce 20 dB C/N in 2.5 kHz BW 'averagely'.
Having monitored the tpx since the beginning I think the reduction of 10 dB (in two steps, first 4 dB and then 6 dB) is a good choice and heeds the formulated requirements and ... improves dynamic range : -)