Fraser river sockeye: A failure of science?

Since 2006 commercial fishing for sockeye salmon on the Fraser River in British Columbia has been closed. In 2009 the run collapsed with a return of 1.5 million despite a prediction of 11 million. This instigated an expensive government inquiry to find out what happened. The end of Fraser River wild sockeye salmon seemed quite likely. Federal government regulation managed to completely fail in eastern Canada, with the cod fishery, and in 2009 it seemed as if they had done it once again in western Canada.
As a sport fisherman I am always under the strong impression that fishing is always getting worse, year after year. The old photos and fishing stories of huge and abundant salmon seem to reinforce my impression.
Now the 2010 sockeye salmon run is a record run with an estimated 25 million fish returning. This is the largest return in almost a century! Good news for salmon, fishermen and the environment but it is an indication of really inaccurate science. An estimate in late July suggested there might be 11 million returning sockeye, far from reality. At this point I think that it is clear that the Department of Fisheries and Oceans is incapable of making accurate estimates. I am not suggesting that estimating salmon returns is easy but making policy decisions based upon their predictions seems like a bad idea. Is there a very important variable that is not being accounted for?
How is it possible that the estimates can be so wrong? Taking a brief look at the pre-season estimate report for 2010 shows that estimates are strongly based on retrospective data. Perhaps this data is insufficient for estimating returns and other sources need to be utilized. I hope that this results in some changes in the methods used for making these estimates. Even better, the government could fund some more ecological studies of this important commercial resource in order to better understand how an error of this magnitude could have happened.