From the report:
The Met Office study of satellite imagery indicated that the mesocyclonic storm front was demonstrating the properties of a significant supercell with associated downdrafts and possible near-surface winds in excess of 100 miles per hour (87kts). The study concluded that a mesocyclonic storm was highly likely with an associated supercell being probable. It also concluded that tornadic waterspouts and downdrafts were possible where local winds could reach extreme hurricane force well in excess of 64kts. Such tornadic waterspout features could be of very limited horizontal extent (50m to 100m wide)
The investigation has also established that, in the assumed loss condition, wind speeds in excess of 63.4kts on the beam were sufficient to knock Bayesian over. It is possible that Bayesian was similarly vulnerable to winds of less than 63.4kts.
These vulnerabilities (when in the motoring condition with sails lowered, the centreboard raised and 10% consumables on board) were not identified in the stability information book carried on board. Consequently, these vulnerabilities were also unknown to either the owner or the crew of Bayesian.
The Met Office study and local observations indicated the probable transient presence of hurricane force winds well in excess of 64kts at the time of the accident. These winds were sufficient to knock Bayesian beyond its angle of vanishing stability