I think a big point is being missed here.
Pilots don't so much need an "accurate" forecast (where "accurate" may be defined as being correct most of the time) as they need an "accurate" worst case scenario (where "accurate" should be defined as never failing to indicate the possibility of the worst case scenario).
If there is a 0.1% chance of a weather condition such as fog or TS occurring then I want a note to that effect on the forecast.