The use of forecasts is forced on us by the midst that we have to fish out in the ocean and not in the terminal areas.

Oregon did a analysis on their wild coho a while back and came up with a really good model that accurately predicted run size. It used four different ocean parameters. The problem was that they were independent enough that 3 good ones could be over-ruled by the last being poor. Or vice-versa. That 4th one was essentially just before they entered the bays and rivers so it was useless for managing the ocean.

So long a ocean fisheries drive things we will be using rather wildly inaccurate models but it is all we have.