Major fishery management decisions depend on the counts of fish entering the Cook Inlet rivers -- decisions like commercial openings, sports bag limits, catch-and-release, emergency closures and long-range run forecasts.
The Alaska Department of Fish and Game maintains two types of sonar fish counters in the Kenai River and works constantly to improve their accuracy.
However, how accurate the salmon counts really are remains unknown.
The science of salmon counts on the Kenai is a work in progress, said Debbie Burwen, a Fish and Game fisheries biologist in Anchorage specializing in sonar.
"At times it has been such a controversial thing," she said. "There is nothing else we have that can give us the in-season information."
The state has a sonar counter for sockeyes at River Mile 19, where conditions for counting are good. The counter for kings is at Mile 8.6, down in the estuary. Tides create challenges for the counter, but its location takes in all the habitat for kings, which may spawn or get caught low in the river.
The two counters use totally different sonar systems, Burwen said.
One problem has been distinguishing between the types of salmon. The early and late runs of sockeyes and kings have similar timing and, at the end of their second runs in August, silver and pink salmon enter the river and add to the confusion.
In general, the concern has been to regulate fishing to preserve all types of fish. A small error in classifying sockeyes can lead to big errors in king estimates, Burwen said.
"We are looking for very few chinooks in with the sockeyes," she said. "(King specialists) are looking for the needles; (the sockeye biologists) are looking for the haystack."
The sonar system has been revamped since the mid-1990s.
The sockeyes tend to run near the riverbank, the kings in the middle. The sockeye counter is a Bendix system dating back to the 1960s and focusing on the river's edge. This year, a new, state-of-the-art system to replace it is being tested on site.
The king counter estimates size and looks for fish in the middle of the river. It began producing data in 1987.
In 1995, a new split-beam sonar was installed to improve the king counts. It allows biologists to track targets in three dimensions and see their direction of travel. But it is labor intensive, requiring technicians to follow traces of dots on a computer screen for each fish.
That year two studies -- one using live fish on lines and the other using a net in the river -- checked the results of the counter.
In 1996 and 1997, the department followed up with a telemetry study. Biologists marked king salmon with radio tags and traced them up the river.
"What we found was the sonar tended to be biased about 25 percent high," Burwen said.
In response to the problem, the biologists sought to improve their counts. They looked for an alternative site for the counter but were unable to find one.
More successfully, they began a standardized netting program to give an independent measure of what was coming up the river. It has proven the most cost-effective way to cross check the numbers. This year, Fish and Game has nets out seven or eight hours a day to detect when sockeyes are running up the middle of the river and skewing the king counts, she said.
"We use the information from our nettings to adjust the estimates down," Burwen said.
She anticipates more improvements in the coming years.
Advances in computer software will allow Fish and Game to automate the king counter to lower costs and improve accuracy. Advances in acoustics will improve the ability to discriminate among salmon types.
The sonar team will know in December if they will get grant funding to push ahead on the advances in 2002.
"I'm pretty proud to say that we remain in the forefront with respect to conducting this type of research," Burwen said.
Despite its flaws, the fish-counting sonar is the best tool for the job at this time. It is improving and it is providing crucial information, she said.
"Most people feel very good about the sonar readings through most of the run," she said.
