On Monday, August 27, 2012 6:40:42 PM UTC-4, Ramy wrote:
we need the actual data to learn something from those accidents, but it is almost never provided.

I suggest that you take your glider to 2500 AGL or so, fly it like you normally do, circle in a weak thermal, record the 'start altitude', put it into a spin and record the altitude of the low point in your spin recovery.

Do this ten times. Take the worst case altitude loss and add a 'safety factor'.

The results will help you set a "personal minimum circling altitude'.

Why would actual flight path accident data (for a different pilot in a different glider) give you a better number?