Known effect, and I can put numbers on the seasonal part from a 5.8 GHz link I maintain, 12 km hop between two hilltop sites, 6 dB antennas, 14 dBi.
Winter baseline: RSL -62 dBm, first-zone radius at mid-path 12.4 m (F1 = 17.32*sqrt(d1*d2/(f*d)) with d in km and f in GHz), obstruction clearance about 9 m, roughly 0.72 F1. Summer, same radios, same power: RSL -71 dBm. Nothing changed except the trees on the south side grew about 1.8 m and the ground stayed wet for weeks.
Why the drop is bigger than the textbook number: a knife-edge model with that obstruction predicted about 4.2 dB of extra loss. We measured 9 dB. The missing 5 dB came from two things the single-edge model ignores. First, the obstruction was not a clean edge; the tree canopy is a volume scatterer, and at 5.8 GHz I use roughly 0.3 dB/m of vegetation depth, so 1.8 m of new growth is only 0.55 dB by itself. Second, and larger, the wet ground raised the specular reflection. At our 2.6 degree grazing angle the reflection coefficient went from about 0.35 dry to 0.6 wet, and the two-ray notch moved onto the path, which alone accounted for 4-5 dB.
Boundary conditions worth stating: the 60 percent clearance target assumes a single dominant obstruction and a smooth surface. Over water, or over flat wet farmland, the reflection dominates and 60 percent clearance buys you little. If the obstruction is multiple ridges, run Deygout rather than one knife edge, and expect several dB more than the single-edge answer.
What I would put in the log: F1 radius in meters at mid-path, actual clearance in meters and as a fraction of F1, RSL in dBm, and the ground condition (dry, wet, snow). Then the winter-to-summer delta is explainable instead of mysterious.
Question: on your link, does the summer fade track rainfall in the previous week, or does it appear even in a dry summer?