Hi,
If I remember correctly an antenna will only seen an equivalent to black body radiation if it's inside a perfect anechoic chamber since it will achive thermal equilibrium with the environment, in reality KTB it's KT_aB (T_a it's the antenna equivalent noise temperature). T=290ºK it's obtained through statistical measurements in earth environment in the more general case(or for a more realistic estimation) it's necessary to know the noise sources that "see" the antenna and apply:
T_a=n_radT_b+(1-n_rad)T_p
where :n_rad is the antenna efficiency
T_b:equivalent brightness temperature seen by the main beam(secondary lobes pick up noise also but usually are neglected for this calculation)
T_p:it's the physical temperature of the antenna(it takes into account the ohmic component of the antenna).
So in reality if an antenna picks up through it's main lobe a noisy source you can see a higher level of thermal floor(or viceversa).
Hope it helps,
PD:in which band are you working?, maybe  it could be a problem of fast fading on the received signal so the SNR will degrade(or improve) but not caused by noise, i.e it's not a true AWGN channel, it's a frequency selective-time varying one and the attenuation it's not fixed.

Hi,
In classical antenna book you will find the definition of T_a and how to calculate T_b(it's a ratio of two 2-D integrals...) but the general concept it's more likely to be found in radiocommunications texts. In this band I think that it's more dangerous fading than variation of the noise(10-20 dB's of variation of the incoming signal it's not at all strange from frame to frame), of course it depends on system specific details.
Hope it helps,
PD:as a reference for antenna theory I think that Antenna Theory:Analysis and Design by Constantine A.Balanis it's a good book .
For general radiocommunications I think that Wireless Communications:Principles and Practice by Theodore S. Rappaport it's a good reference(although I think that it doesn't cover MIMO...) .