For diversity gain, you'd need to have two receivers (or you just switch to the currently "better" antenna; that's called selection combining in MIMO technology and kind of is the worst possible diversity mechanism).
So, no, with two antennas and only a single receive chain, there's no diversity gain.
You can, of course, build an antenna array. That would actually work by using a impedance-matching combiner to make the reception of both antennas combine constructively. You'd basically achieve a higher directivity by adding array gain to your antenna gain.
Also, I wonder how you can put two antennas at "1 wavelength" distance for the whole UHF band – that covers wavelengths from 10 cm to 1m.
I'm not sure it would help much with the fading problems – these might be happening at larger scales.
As a more general comment: Diversity happens when you have two different signal paths transporting the same signal; different in the sense that the random effects on the signal are independent. That will almost certainly not happen for large-range communication with transmit antennas that are a mere single wavelength apart – I'd not expect any significant gain due to independent signal paths ("multipath") in an outside scenario with two close antennas pointing in the same direction.
In indoor scenarios (e.g. your usual cheap Wifi router with more than one antenna is a diversity/MIMO transceiver!), there's way more diverse multipath, and hence, antennas placed a mere more than half a wavelength apart have a serious chance of picking up a combination of at least two independent paths. Mathematically, the system is modelled such that:
- We assume one channel response from each transmit to each receive antenna
- We then put these into a table (and for reasons of simplicity, let's assume these channel impulse responses are completely characterized by a single complex number)
- We then consider that each receive antenna sees the sum of all the channels that "go to it", applied to the transmit signal
- And then we consider that the table from above is simply a matrix, and we can multiply that matrix with the transmit signal vector (ie. a vector with the entries of what each transmit antenna sent) to get the output at each receive antenna
With that model in mind, we can let the transmitter choose a transmit vector, so that the process of
- taking a vector of the signal to be transmitted
- an arbitrary mapping to a transmit vector
- applying the channel matrix
- and then applying a matrix that we can also arbitrarily choose
gives us multiple, independent channels, over the air, mathematically.
If this is of interest to you: that usually happens by applying a singular value decomposition to the (estimated) channel matrix, so that this matrix gets split into unitary (hence, nicely invertable) matrices and a diagonal matrix (which means that whatever you put into that diagonal matrix in one element has no effect on the other elements).