parados wrote:Are you arguing that weather forecasts are reasonably accurate only 50% of the time? That would be required for it to be random in relation to reality. I would argue that you can't show me any random sampling of forecasts that aren't close 50% of the time.
I am saying that weather forecasts, based on numerical integrations of the global numerical weather model are reliable only for about five days into the future. Beyond that their accuracy falls off very rapidly, and about 15 days into the future they are useless as an indicator of what will happen. Seasonal predictions based only on averages for the time of year are more accurate - indeed they remain the best we have.
I am also saying that the various doomsday projections on matters ranging from the 'Atlantic Conveyor' ocean current to projections of global rainfall and weather patters - all derived from the forward integration of analogous time-dependent models - are similarly subject to sensitive dependence on initial conditions, mathematical chaos, and the resulting gross inaccuracies in projections more than a short time into the future.
This is not an esoteric thing known to a very few. It is a well-known and widely recognized mathematical and empirical fact.
The static loadings of (say) the structure of a bridge, based on different applied loads, as they may occur in the future, is, from a mathematical perspective, a very different thing - and not subject to chaotic uncertainty. It is a sequence of static conditions, each resulting from different applied loads - not the time-dependent result of a coupled, non-linear dynamic process.
