How The Springer Forecast Came About

Reaction to yesterday’s prediction of 470,000 spring Chinook back to the Columbia River in 2010 ranges from “We are going to have a very fun year next year it looks like” to “Even half of the prediction would be very nice!” to “I’ve got ten bucks that says they have over-estimated. Their track record for making estimates is horrendous.”

Those comments came from posters on ifish known as Fish Hawk Adventures, allwaysfishing and Bait Bucket.

Similar statements were also registered on three other Northwest salmon and steelhead fishing boards, piscatorialpursuits, gamefishin and steelheader.net, and anglers quoted by Allen Thomas of The Columbian all arched their eyebrows at the prediction.

Added Mark Coleman of All Rivers Guide Service (425-736-8920) , “The forecast of 470,000 springers sounds great and I’m hopeful that the actual return will be that large. Even if we only get a percentage of that number, we will still be looking at a lot of springers to catch. I normally start my guide trips around the middle of March, but if we are really getting over 40o,ooo fish, we could have good fishing a week or two earlier.”

Indeed, we springer fiends can’t help but be giddy and leery at this potential blockbuster of a Christmas present.

A) The run, as predicted, may be the largest of all time, topping the current mark by over 50,000!

B) This year saw a run that was only 54 percent of what the same team of forecasters said it would be!

Indeed, in the inexact science of trying to forsooth how many salmon might return to the Columbia from a 93,000,000-square-mile body of water based on how many of a certain age came back last year, the margins of error in recent years have been huge.

Four of the past six years’ runs have been well below forecast.

But anglers sometimes forget that we’ve also had years where managers got it right (see 2007), and there have been seasons where more than expected have returned (see 2001, The Best Year Ever) as this graph from ODFW and WDFW shows:

The difference, though, is that the number of underforecast fish is much fewer than the number of overforecast fish. That is, since 2000, 211,000 unpredicted fish have returned to the Columbia River, but 534,000 forecasted springers have failed to come back.

It’s led anglers to roll their eyes at any forecast, but the missing fish have also given managers fits in recent years.

Part of the run consists of threatened wild springers which require protection, and as Hal Bernton writes in today’s Seattle Times, “Accurate forecasts are necessary to set harvest at levels that meet treaty obligations and conservation requirements under the federal Endangered Species Act.”

Columbia Chinook are required to be managed by what is actually coming in — thus the all-important midseason update — vs. what is merely forecast. Part of the problem is that in the last five years, the midrun mark has unexpectedly moved deeper into the year.

It’s all led to sometimes herky-jerky seasons and angler discontentment.

This year, managers also have had to contend with a strange new signal from the salmon: A record 81,000-plus jack, or 3-year-old, springers returned past Bonneville, four times the previous high.

Managers figure that the jack Chinook run is a certain percentage of the overall year-class, and more or less multiplying out that percentage will give you the following year’s return.

When folks like Stuart Ellis of the U.S. v. Oregon Technical Advisory Committee did that “back of the envelope” math last spring, they came up with a run forecast of 1 million to 1.5 million adults in 2010.

In November, however, he told me that it was “unreasonable” to expect even a doubling of the current record return, 2001’s 416,000.

But he also told me, “Nobody’s going out on a limb to say a record return.”

And yet that’s what the forecast released yesterday says.

Four hundred and seventy thousand! “Holy f#$%*#@ *%t!” is what I said.

“That’s what we thought — wow!” Ellis tells me this morning.

I called him to find out more about how he and the rest of the committee — which includes Oregon, Idaho and Washington fish and wildlife departments, the Columbia River Inter-Tribal Fish Commission (on behalf of the Nez Perce tribe, the Umatilla tribe, the Warm Springs tribe and the Yakama Nation,) the Shoshone-Bannock tribe, the National Marine Fisheries Commission and the U.S. Fish and Wildlife Service — settled on that number.

Turns out, it’s just a midpoint from seven different mathematical models that spat out run sizes of anywhere from 366,000 to 528,000 adults next year.

And it almost sounds as if Ellis et al, in their attempts to deal with all those jacks, looked at 470,000 different ways to forecast the run.

“We looked at a huge number of models and then settled on 18 that had any validity,” he says.

In previous years, they’ve used a straight linear graph of jacks on the X axis, adult 4-year-olds on the Y axis to come up with a number, but among the new approaches was a nonlinear curve.

They also used models that factored in ocean indicators such as water temperatures, upwellings and the Pacific Decadal Oscillation, climactic and environmental conditions that haven’t been used before.

They shortened up their data set.

They looked at jack returns to home hatcheries, not just past Bonneville Dam.

And they calculated the absolute lowest number of adults that have come from a jack return.

“With the 18, we did ‘hindcasting’ to narrow things down based on error rates,” Ellis says. “We chose seven that would have been acceptable in the past. They tended to have the lowest error rates.”

The group then talked over those seven models’ pros and cons, and in the end averaged them, he says.

“They’re all quite reasonable as far as predictors,” Ellis says, but he adds that right now, they’re “not trying to assess how accurate (the forecast) will be. There’s no history with this size of jack return.”

“It’s kind of tricky when you’re dealing with records,” he says.

“You can make extreme theories  on jacks — outrageous inriver and ocean survival, so lots will come back. There’s lots of evidence pointing that way.

“But another theory you can’t disprove is, maybe there’s a big change in maturation rates in the ocean. For some reason, a huge proportion came back as jacks and we don’t have a lot of adults out there. That line of thinking leads to lower estimates. That’s possible; we can’t rule it out.

“But the sum total is, we should get a pretty good return,” Ellis says.

Looking at 2009’s actual return, 169,000, he says, “I think we’re going to do a fair bit better, but don’t bet the farm.”

According to today’s Columbia Basin Bulletin:

The upriver forecast includes 272,000 Snake River spring/chinook, of which 73,400 are expected to be wild fish, and 57,300 Upper Columbia spring chinook (including 5,700 wild). The balance of the upriver forecast is comprised of mid-Columbia spring chinook.

Now that we at least have a number, Ellis says that the season setters can begin to craft recreational, tribal and commercial fisheries. We already know that they’re going to chop 30 percent off the forecast as a run buffer and move some of the allocation upstream, as Bill Monroe of The Oregonian detailed, and which has pissed off two commissioners of a lower Columbia River county on the Washington side.

Ellis has no doubt we’ll be fishing — “It would be well worth making sure your spring Chinook gear is in tip-top shape, where you’re going to get your bait, and your boat motor will start up” — but he’s warned me before and he warned me again today that we will need to be flexible in our late winter and spring fishing plans.

“The opportunities may not be in your favorite area, or your preferred area to fish. You may have to go to choice number two or three,” he says. “Hang loose.”

Planning fishing trips months ahead may be more difficult, he suspects.

Asked why his committee’s guesses have been well off in recent years, Ellis says that springers are pretty difficult to forecast due to their timing, the wide geographic landscape they return to — not to mention the fact that as smolts a couple months out to sea, they literally disappear off the face of the earth for one to two years. Who knows what sort of ocean conditions they’re swimming through.

The “intense interest” in the species — widely viewed (at least in the Northwest) as the best-tasting fish on the planet, and a substantial cash cow for river cities and the sportfishing industry as a whole — magnifies those errors.

Ellis points to the Spring Creek tule Chinook stock that returns past Bonneville Dam in the fall. That run has come in 50 percent below forecast, or at 150 percent, and outside of the managers, not too many folks are wringing their hands.

“They’re just not as impressive to anglers. They don’t tend to gripe as much,” he says.

And even as Ellis says the run could come in above and below what those seven models spit out, he’s confident in all the work that’s gone into 2010’s prediction.

“We do think the effort we put into the forecast — we’ve made a real good effort to reduce the risks of very large errors. We hope it’s accurate, but we’re not guaranteeing it,” he says.