Regular tide-related fading was par for the course on the Braewynner to Thrumster 4GHz link which was on my patch in 1970. The description is in the website -
http://www.dgsys.co.uk/btmicrowave/sites/6.php
The link was 55 miles long and 50 of that was over water! Quadruple diversity was essential to get an acceptable reliability. You could watch the received signal levels in each of the four diversity channels rise and fall gradually between the high and low tides. The sheer length of the path was definitely pushing it, and there were times when the whole lot would fade to noise anyway. The equipment was a custom designed one-off, and it was very complex and fault prone, given the multiple diversity switches and (obviously) valve based technology. I think it may have been one of the longest over-water Line Of Sight links ever - anything further, and you're in troposcatter territory. The transmitter output was only one watt per channel, so there wasn't much margin in hand in the first place...