Germ,
I think I understood your concept, but for reasons as stated by others, I can't see it working, so I went off at a tangent!
Your principle starts with a large body of underground water, to all intents uninsulated from ambient ground temperature. In summer you want this body of water to be frozen, so you can slowly thaw it to give chilled water to the house?
All very well, but this water needs to be taken to close to zero for it to be effective for chilling, and you can only do this over winter if you need it frozen for the summer. I don't do GSHP, but as I understand it, the one thing that remains pretty constant throughout the year is ambient ground temps below 1m. Your idea would work to a point, the water could be used for winter GS heating, but it would never get below the ambient of the ground ( 10 deg? ) without further chilling, or massive thermal insulation. As the water descended to ambient ground temps, it would just strip the surrounding ground for heat, I assume this is your principle for winter heating.
But get back to summer, all you have is a body of water at ambient ground temps. You can use this for what? heating!
It seems to my uneducated GSH eyes, your concept is just using water as a heat transfer medium for what is basically GSH, and driving the GSHP is excess PV power in summer. Winter PV excess is minimal if anything, and once you move temperatures beyond either side of ambient, you need energy to do it.
Have I still missed something?, like I say GSH is not my bag!