A great question. The elevation plot is presenting the gain (dBi) in relation to the ground. The x-axis on the plot represents the ground. Most of the radiation directed at the ground is reflected, contributing to the the gain shown above ground in the plot. This is why ground conditions are so important.
The elevation pattern that you presented as an example of a Yagi is a model of an antenna in free space.
If we model this same Yagi antenna half a wavelength above an arbitary set of
‘typical’ ground conditions the far field plot looks very different.
Repeating the model with the same Yagi antenna lifted to 4 1/2 wavelengths above the ground shows the emergence of lobes and an increase in low angle gain.
Modelling allows us to visualise the gain of an antenna by making some assumptions. Your plot used the assumption that the antenna is in ‘free space’. Modelling the antenna above ground uses a different set of assumptions, but does not bring the antenna into the real world, we now have assumptions of a level, flat ground with specific ground conditions. As has been pointed out summits are not flat, and the soil conditions, water table and countless other parameters are different from the idealised world that I am able to model. The visualisations are an aid our understanding but only take us so far.
I have some hunches on what might happen if the ground is not flat and level. If the ground is perfectly flat but inclined I think that the entire far field plot will be inclined. If the antenna is on an area of land that is inclined in an otherwise level landscape I suspect that the far field plot will be inclined to a lesser extend that the angle of the ground. If the antenna is sat on top of a cone of ground I suspect that the far filed plot will be pulled down towards the ground in all directions. This leads me to believe that it may be possible to place an antenna too high.
The real world is complex, and SOTA summits doubly complex. I will never be able to create a model for a real summit, let alone all summits.
I have assumed that I want to maximimse the RF at an angle near the horizon. What goes on once the RF leaves the antenna is even more unpredictable and may proove this assumption to be wrong in certain conditions.
I was on [SOTLAS](https://GW/SW-035 Myarth) yesterday, on 145.450mhz fm. This is a difficult summit for VHF. A helicopter flew over and suddenly I could hear @M0TRI on [SOTLAS](https://G/SC-006, Periton Hill) in a QSO with MW0SAW on my frequency. The helicopter left and they were gone, a great pity as I was struggling for a 4th contact. It would be easy to attribute this to a reflection from the helicopter, but who knows.
SOTA is about far more than antenna modelling, or even the perfect antenna. One of my ambitions is to have a summit to summit with Fraser one day.
I have watched your videos and you do far more than ‘just climb hills and try to make contacts’.
Yesterday I had a summit to summit with Gerald MM0EFI, who I had met on Mickle Fell earlier in the year. @M5RJC walked up a hill specifically for a summit to summit with me. There were lots of other people who made my day so pleasant that I have not mentioned.
The antenna helps, but untimately, for me, its about people.