What Happened After the Klamath Dams Came Down¿

Planet in Balance 🌎
1 follower
3 Video Views·Jan 2, 2026
This project's goal is to monitor the re-establishment of the Salmonid species returning to a habitat
that they haven't had access to. In some cases over 60 years, in some cases over 100 years.
We just want to see what's happening. We just want to see where these fish are going and learn as
much as we can from them directly. Some of the coldest water I've ever stood in is above these
these former dams. You stand in Fall Creek and it's so cold it hurts your feet.
You know and you know that's a good thing. So these salmon are really going to have access
to spring-fed streams where they can find refuge. And you know with with climate change
and global warming and salmon are going to need more cold water. So this project is really the
first step in kind of understanding what salmon do. I'm pretty new on the Klamath. Last year I was
sort of like things were kind of slow in the Elwha. Like sort of pumping the brakes on people's
expectations. And the week I was here we saw them going by the sonar. They were seen in the video
we were upstream. They caught them in a net and then like two days later they were in Oregon.
And the same thing happened when I was here two weeks ago. It was just like one or two showed up
and then all of a sudden they were everywhere. I would say they blew my expectations out of the
water. I never would have guessed that there was 7,000. And now this year I'm kind of looking at
the hatchery releases and I think there's going to be 10,000 this year.
Our fish counting station is running 24 hours a day, seven days a week, year round. And it's
recording footage that entire time. So counting a year's worth of footage can take a very long time.
I tell people the great thing about it is you can stick it in the water and press go and start
filling up hard drives full of data. The really bad thing about it is you can stick it in the
water and press go and start filling up hard drives of data. Human beings are can be trained
to count the fish. Now if you imagine counting fish for 360 days a year, 24 hours a day,
not much fun. So one of the big partnerships we have is with Caltech and MIT to develop a
machine learning AI for the sonar fish counting station specifically. It's like a shoebox size
device which is in the water and that gives us a video feed that shows the fish going by like
little shadows in the river. And so the challenge for us in AI is to identify the fish and measure
them, count them, hopefully also tell which species they are and hopefully within a couple
of years this will be working well for all the communities that are interested in restoration
of the river. So AI is currently being sort of tested here. Our end goal is to have this running
on every sonar sensor in the world, starting with the west coast of the U.S. and Canada. Here we're
sort of using this as a test site of it's a really important site, they know they have a certain
number of fish, they're set up to manually annotate in a way that we know we can get good comparison
readings on. So the end goal is to have this in situations where there's a person involved but
not necessarily the person annotating all of the data. I think the sonar is so interesting because
it's not an easy data type to work with. If you're watching this sonar footage, any given still
frame, if you just take a screenshot you're like where are the fish? Even the experts couldn't
tell you, they need the video, they need the context, they need to learn about the river.
It's a difficult problem and so this is one of those cases where actually the process of training
the AI models takes more time, takes more investment and though the benefits are still really there,
the process of getting it to be reliable is more complex. Thank goodness for Caltech and MIT for
stepping up and wanting to assist with this portion of the project and genuinely wanting
to help. Yes it's really cool technology they're working on but they also see the benefit that it
can bring to us. This technology will allow us to sort of you know you can open and close a
in the middle of the season if you want to. You can kind of like make real-time in-season
management decisions based on the counts coming out. I mean it really has the potential to change
the way we count fish across the Pacific Rim. In the Klamath you know we have a history of working
together and working in partnerships and diverse partnerships too. We have tribes and non-profits
and state agencies and federal agencies so it's a diverse group. Everybody brings something to
the table and that's that's really good to have that broad set of knowledge to inform these kind
of projects. It's incredible having the Kadook tribe out here working day in day out directly
with me and working on other projects. I mean the experience that those teams have is invaluable.
Being able to work with guys like Alex Quorum and Taz Soto and then the technician teams of Bo,
Eric, Ben, Clayton. Having real water happy people around. Excited to get in the water,
excited to jump in when it's cold and rainy out here. I could not be happier to have that
crew out here day in day out with me. I'm hoping we continue this work for a long time. If we only
do two three years of this work we won't truly know what's going on. It's going to take a decade
plus to really understand what's happening on this river in these new sections.
