NOAA Salmon Predictors
by Bill Rudolph
NOAA researcher Brian Burke says the quantitative fish predictions his agency has produced over the past few years were not necessarily done to create numerical forecasts, "but to understand how the ecosystem works."
At last week's meeting of the Northwest Power and Conservation Council in Boise, Burke updated members on NOAA Fisheries' ocean indicators project that has been tracking climatic and biological data since 1996. He said creating a forecast "is a lot like suggesting a hypothesis about how we think the system is working, so when the fish come back, that basically tells us how that hypothesis turned out."
Burke said it's been a learning tool for his agency, and that it has been exciting to see the results have been useful to others. He said NOAA plans to continue the work for both these purposes, even though the original motivation stemmed from trying to understand the main drivers of salmon ecology.
Burke showed plots of the important factors that lead to their results. For spring Chinook, he said one important element is the number of juvenile fish caught in their ocean trawl surveys every June. For fall Chinook, the most important elements were various measures of copepod richness, and for coho, the top two were related to sea surface temperature.
He said they know these different species go to different places in the ocean at different times, "so it makes sense that they are interacting with the environment differently."
One thing that surprised Burke and his fellow researchers was the relative importance of the Pacific Decadal Oscillation to spring Chinook abundance, but not to fall Chinook or coho. Burke said the PDO cycle is a big picture item that explains the overall scenario (decades of generally cold and wet conditions versus warm and dry), but does not capture regional dynamics.
NOAA blew its 2013 upriver spring Chinook forecast big time--estimating about twice the number that actually returned, which Burke pointed out, didn't correlate well with the PDO data or a measure of growth hormone when juvenile fish were checked in 2011, after the fish entered the ocean. Columbia Basin fish managers' technical advisory committee [TAC], using their jack-based model, had predicted 141,000 (to river mouth) for the 2013 spring run, which was close to the 123,000-fish return.
But when the NOAA scientists took a closer look at the PDO data, they found large differences in sea surface temperatures (SSTs) between spring and late summer. In March, waters off the whole Northwest coast were cool all the way to Alaska, where SSTs were much warmer in some areas. But by September, warmer water from the central Pacific had reached the coast of Vancouver Island, and essentially split the cool areas in two, disturbing the horseshoe pattern of cold water off the West Coast, By then, the warmer water off central Alaska had cooled.
"The PDO is a great summary of the ocean, but it's not sufficient to describe the dynamics," said Burke.
When they checked in with Alaska scientists, Burke said they had not even been able to detect a spring plankton bloom in 2011, a sign that feeding conditions for juvenile fish were poor. He said these complications have led his team towards a new goal--to refine the indicators enough to estimate their effect at a certain time and place of a given stock.
He presented a map that reflected the differences between jacks and adults, showing that after traveling up the coast with the rest of the juvenile migration during the spring, when mortality is usually high, the jacks turn back at some point, and begin heading towards the Columbia. So, in many years, the number of returning jacks does reflect the run size returning the following year. But it doesn't when the young fish encounter much different conditions than expected once they reach the Gulf of Alaska.
Burke said by measuring environmental conditions when juvenile fish are migrating, noting prey and predator abundance, growth rate, and abundance, they may be able to tell something about the number of fish that will likely return as adults.
Both the jack-based models and NOAA's ocean indicators are very valuable ways of looking at the system, Burke said, and sometimes they come up with similar forecasts. But in some years, that correlation between jacks and returning adults doesn't hold. He noted the large discrepancy between the TAC forecast for upriver fall Chinook returns (1.4 million) and NOAA's (461,000) for this year's fall run.
"With the PDO, when things don't go the way we think, that's when we really learn something about the system. In a few years, I think we will all benefit from the dichotomy in these forecasts." Burke said.
Last year, TAC predicted about 700,000 fall Chinook (upriver and lower), and about 1.3 million showed up. NOAA predicted 420,000 (upriver only) and 953,000 had appeared by the end of the season. Burke noted that NOAA's 2013 estimate was actually a range between 270,000 and 680,000 fish, an admittedly wide prediction interval due to the fact they had only 16 years of data.
For the 2013 upriver spring Chinook run, NOAA predicted a 200,000-fish return (102,000-370,000), but only 83,000 came back. It was the biggest miss since they began announcing numerical predictions. But then came the under-prediction of the fall run.
For 2014, NOAA is predicting an upriver fall Chinook run of 420,000 (270,000-680,000), while TAC managers have developed a huge forecast--expecting a return of 1.6 million to the river mouth, and nearly a million upriver brights back to the Hanford Reach.
In the future, Burke said they may try a hybrid approach, that uses jack counts and ocean indicators.
In response to a question from Washington council member Phil Rockefeller regarding the project's declining budget, Burke said their funding has decreased quite a bit and it has been a challenge to redesign the project to collect enough information to get their questions answered. He said working with annual budgets over the past few years has made it "a little difficult" to plan for the longer term, particularly if you are not sure that collecting information for certain time-series will be maintained from one year to the next.
In 2011, BPA spent $2.1 million on the various components of the ocean survival work. That's been whittled down to about $900,000 annually. Burke said the cuts have meant that the annual May ocean survey cruises have been cancelled, and NOAA now funds its September cruise, while BPA still pays for the June cruise work. Without the May cruises, important information about steelhead survival is being lost, the scientists say.
learn more on topics covered in the film
see the video
read the script
learn the songs