The Icy Nenana River

Guest Post by Willis Eschenbach

The last time I was in Alaska, I had the good fortune to stop by the town of Nenana, home of the Nenana River Ice Classic. Nenana sits at the junction of the Tanana and Nenana Rivers. The dates of the ice breakup at Nenana form one of the longest-term modern temperature proxies in the area, extending back unbroken to 1917. Figure 1 shows my photo of the tripod which is set out on the ice to determine the exact tim of the breakup.

nenana ice classicFigure 1. The tower and the tripod. The tripod is placed out on the ice before the breakup. When the ice breaks up, the tripod falls, tripping the clock. These days the tripod is actually a quadripod, or perhaps a quadruped.

Every year people pay money to bet on the exact time of the breakup, with the winner taking the pot. At present, the pot is $318,500 …

There’s a recent WUWT post by Psalmon about the Ice Classic here. Although I’d written about it previously, there were a couple of things I didn’t understand about Nenana until I’d visited the place.The first was the reason the ice breakup was so important. It was critical because both then and now, the river is navigable, and becomes a main highway for people and supplies during the summer. Until the breakup, little villages and cabins and camps along the river can’t get their supplies or travel by water. Although this is less important now with the advent of highways, there still are many places along the river that can only be reached by traveling along the river. That made the breakup a huge event in the old days.

The second thing I didn’t understand was the reason why the breakup was so sudden and complete. The map shows the Tanana River and surroundings:

ge nenana ice classicFigure 2. The Tanana River and its tributaries flow north to the Yukon River. The Yukon flows from the top center to the upper left of the figure, with a portion appearing dark blue.

The reason the breakup comes suddenly is that unlike most US rivers, the rivers around Nenana are flowing north. As a result, the more southerly upriver parts of the drainage would tend to melt earlier. At some point this increasing upstream meltwater will put pressure on the downriver ice, and as the ice at Nenana rots and melts, the whole thing will break and collapse at once.

Now, you’d think that the river breakup dates would be a perfect temperature proxy. After all, urban warming surely won’t be an issue. However, nature always sides with the hidden flaw, so of course there is a confounding factor—rain. Rain can hasten the breakup significantly by melting the ice from the top. In addition to starting out warmer than the ice, rainwater pools have less albedo than ice, so they warm more for a given amount of sunlight. Rain also increases the volume of water flowing in the river, so it puts additional pressure on the ice. As a result, the breakup dates form the usual imperfect proxy for temperature.

Given all of that, here are the inverse dates of ice breakup since the Classic was first run in 1917.

ice breakup dates at nenanaFigure 3. Nenana ice breakup dates since 1917. The result for 2013 will be equal to or greater than today, May 15th. Blue dots show the standard error of the Gaussian average at the endpoint in 2013.

For me, it is clear that what we are seeing are the effects of the Pacific Decadal Oscillation. This is a slow, decades long cycle in the heat distribution in the Pacific Ocean. In about 1945 the PDO shifted to the cool phase, then went back to the warm phase in about 1975, and has recently switched again to the cool phase.

Overall? I’d say there’s not a whole lot to see in the ice records. Temperatures went up a bit, down a bit, up a bit, and appear to be on their way back down again … be still, my beating heart, it’s all too exciting.

I leave you with the current photo from the Nenana Icecam, at 8:47 Alaska Daylight Time May 15, still no breakup, but goodness, it’s a lovely spring day in Alaska … makes my heart leap just to look at it.

ice cam 945 am pdt

Regards to all,


DATA: the historical breakup days are available here.


newest oldest most voted
Notify of

Wow, looking at the breakup dates, Two distribution humps. Suggests an oscillatory system is at work there. It is odd that the later hump has a dip in the middle. Possibly a 3 state system?

D.J. Hawkins

Who’s the poor SOB who has to go and retrieve the tripod?

John Tillman

To appreciate the infantile state of climate science, consider that the PDO wasn’t discovered until 1997, by a Pacific NW fisheries researcher.

Willis Eschenbach

John Tillman says:
May 15, 2013 at 10:26 am

To appreciate the infantile state of climate science, consider that the PDO wasn’t discovered until 1997, by a Pacific NW fisheries researcher.

True that. The PDO has a huge effect on the fisheries of the US and Canadian Pacific Northwest.

if the measurement of the day in the year of break-up is exact,
there cannot be a standard error.
What error do you want to bring in? For 2013 we are already on 135
and counting.
Looks to me, on the black, we are exactly on course
= 1925
what did I tell you

Don B

My comment earlier today on the previous Nenana post…
“The early ice breakup date in 1998 was the last hurrah for the warm phase of the PDO. Since then the trend has been towards later breakups, at the rate of 0.4 day per year. (The Excel linear calculation assumes this year’s breakup is today, May 15, which is not my prediction, but an assumption to use in the calculation.)”

Robert M

It was snowing at my house when I left for work this morning. Cooler then normal weather is expected until at least next week. There is a pretty good Chance that Anchorage, Alaska’s largest city, will have measurable snowfall Friday night into Saturday. Which if it happens will be the third (I think…) latest on record…
In about 4 hours, 2013 will move into the top three latest breakups ever. (As measured by the Nenana Ice Clasic) The latest is 20 May.
If CO2 was truly a strong climate driver, I would expect so see it’s fingerprints all over your graph Willis, but you are correct, all I see is a noisy PDO signal.
One other thing, you mentioned rain as one of the confounding variables for using this as a temperature proxy. Last year was one of the earliest breakups, and I think what made it early was the absolutely enormous amount of snow that fell last year. The area broke snowfall records early and often. I had around 17 feet at my house, and nearby Anchorage had it’s snowiest year ever. I think the additional insulation from the snow kept the ice from getting very thick. If you look at last years thickness measurements, the ice was not very thick at all, even though that winter was one of the coldest we have had.


I wonder if climate science was funded by fisheries and agriculture instead of environmentalists, we would have had any of this carbon-AGW nonsense to begin with?


What would be interesting to know is if there are any similar records in Central Canada, Siberia and West of the Urals also?
That would give some insight as to whether temperature proxies were even by longitude, even but out of phase or closer to random.
Given that AMO is out of phase with PDO, you might expect west of the Urals to be phased, but perhaps Eastern Siberia to be similar to Alaska??
Anyone got enough data sets to answer scientifically?


The chart means something to me, so many thanks, Willis.
1. It clearly shows the earth’s climate is variable.
2. It shows the observed warming from the 1930s to the peak in 1940, which appears in many temperature datasets.
3. It shows the observed warming from the early 1980s to about 1997.
4. It shows the ending of the latest “global warming period” and first flattening then declining after 1997.
5. What it does not show is any influence of carbon dioxide from burning fossil fuels at any juncture.

True. CO2 is and never was a factor. Is is just nonsense to make people feel guilty and to collect taxes. H

Frank K.

From the website, if the breakup date goes past Monday, it will be a new record…then we can all sing Neil Sedaka’s “Breaking Up is Hard To Do”… 🙂

I like to cross-check against other data. Looking at HADCRUT3, there was a huge drop in global average temperatures in the mid 1940s, that seemed out of place next to GISSTEMP. These figures are more in line with HADCRUT3.
Last year it was early at 7:39 p.m. April 23. In fact the fourth earliest on record. Does anybody know if there was heavy rainfall?
This year is already the third fourth or third latest on record.

Stopped in Nenana earlier this year for breakfast. I put my money on 5/12 4:30 PM. Looks like I won’t be collecting the cash prize.

Tom in Indy

A quick google and here are ice in/ice out dates for hungary and finland. Surely, someone has collected all the ice out dates available on the web. No?

Fred Harwood

The late John Daly referenced Fairbanks, upriver, as a possible factor in the dates. One of many power plants there consumes more than 210,000 tons of coal a year.

Hey! That ice-cam is a terrific thing. I’m thinking about inviting neighbors over to watch for the ice-free moment. This is what life comes to when you retire.


D.J. Hawkins says:
May 15, 2013 at 10:24 am
Who’s the poor SOB who has to go and retrieve the tripod?

Being Alaskans they cannily have attached cables to the tripod so it can be retrieved without having to go wading among the ice chunks.

Greg Goodman

Interesting the tendency to alternate. I’ll have to grab the data and compare to Arctic melting season.

Greg Goodman

Also the arctic alternation comes from a complex pattern of three frequencies close to two years (an amplitude modulation triplet) that means it is not regular every two years. sometimes the pattern skips.
The triplet resolves as amplitude modulation of a 2 year cycle by one of 12.83 years.
The 2 year is probable some internal flip-flop: more melting causes more snow , which causes less melting the next year. It tends to oscillation.
12.83 is less obvious, but just to note the alignment of Jupiter and Neptune happens every 12.78 years which is within the uncertainty of the detection of the AM triplet. Neptune is the third ranking planet in it’s inertial effect on the sun (Jupiter is first) and hence earth. But that’s tale for another day.

Greg Goodman

Just dug out another version that uses both melting and freezing periods to give a second estimation of melting behaviour.
By eye it’s a bit hard from Willis’ graph but it does not look like a close match on year to year scale. Though recent cooling is clear in both.

Greg Goodman

Willis, where did you get the numbers from?
Looking at the menu of their site all the links are like : file:///E:\other files\blah….
Some ham fisted windows user messed up and used local file names.
If you have a direct URL for the data it would be good to have.

Eric Ellison

You spoilsport!
We were having so much fun treating this as a temperature proxy! Until you came a along and rained on the parade of articles with math and even history!
Next year we WUWT folks should start our own lottery on this even and use your math skills to predict the date of the breakup!
I appreciate you and anthony


Isn’t the Nanunenenunu “tripod” really a tetrapod?


As a kid living up in Northern Ontario I remember there being a pool for ice break up. Back then instead of a fancy time tripped tripod we used an old broken snow machine sitting out in the bay. It disturbs me now to think of that piece of junk sinking into pristine Northern waters or how many years they might have done that. I think they pulled them out later and reused it for the next year. That is what I choose to believe. Of course that was the seventies and we were supposed to die at any moment from Global Cooling so what did it matter anyway. LOL

Mark B

A good read as always.
A phrase in one of your sentences jump-out that I very much like: “ . . . nature always sides with the hidden flaw . . . “
It reminded me of the final sentence in Richard Feynman’s appendix to the Rogers Commission report on the Space Shuttle Challenger failure:
“For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled.”

Bill Illis

The reason for the two humps in the break-up data is probably due to the fact that there is a strong correlation to the ENSO here (from Alaska down down to Minnesota). It has the strongest correlation of any region in fact other than the ENSO regions themselves, Australia and Indonesia. Northwest North America is one of the most impacted regions from the ENSO.
Early break-up in El Nino years, late break-up in La Ninas. 3 month lag in effect. There was weak La Nina conditions early in 2013.

Willis Eschenbach

HenryP says:
May 15, 2013 at 10:38 am

if the measurement of the day in the year of break-up is exact, there cannot be a standard error.
What error do you want to bring in? For 2013 we are already on 135 and counting.

Sorry for the lack of clarity. It’s the standard error of the 11-year gaussian smooth. When you get to the end of the data using a centered 11-year smooth of some kind, the final points are an estimate of the eventual smoothed curve that you will obtain when you have the next five years of data.
That estimate, of course, has a standard error which I’ve indicated by the error bars. For a discussion of the issue, see my post “Dr. Michael Mann, Smooth Operator“.

Bob says:
May 15, 2013 at 12:12 pm
Hey! That ice-cam is a terrific thing. I’m thinking about inviting neighbors over to watch for the ice-free moment. This is what life comes to when you retire.
If you need some more excitement, I can send you a can of paint to watch dry !!
(only 25 years until I “retire”)

Willis Eschenbach

Mark Cates says:
May 15, 2013 at 11:44 am

Stopped in Nenana earlier this year for breakfast.

I would prefer my Nenana breakfast in August, but you’re a lucky man nonetheless …


Weather – even over a fair-ish chunk of Alaska, which is twice the size of Texas – isn’t climate.
But – and Willis, many thanks again! – lots of ‘bits’ of weather make up climate – in enough time.


Talking about ENSO and La Nina, the cold SST plume of Peru in the East Pacific at the end of the anomaly animation looks like something from the intro sequence of a James Bond movie just now.

Greg Goodman

Just thought I’d throw this in since this is all about melting ice: just found a striking correlation in recent AO index and rate of change of CO2.
Now since out-gassing of sea water is a function of temperature the dip may presage a cold 2013. Most of NH seems to be feeling it already but also a rebound for 2015.
Good news since, despite wanting to see an end AGW hysteria, I’m not sure I want pay the price of a replay of the Maunder Minimum to prove the point.
Since Willis has already pointed out the tropical oceans are fairly stable in temperature and polar regions are reckoned to show polar amplification of warming, I think this may be where a lot of the increasing CO2 is coming from.
Cold water holds a lot more CO2 than warm tropical waters to start with.
Now rate of change of CO2 does not cause temperature but the other way around does fit the physics and those wiggles are specific enough for this not to be coincidence.
So that dip supports a late melt this year but don’t bet on the same thing next year.


What about the run-off of the tributaries upstream ?, if you get enough water flowing under the ice, it will either melt from below or rise enough to loosen its grips on the banks of the river, thus moving the tripod and making someone rather wealthy. No ??

Willis Eschenbach

Fred Harwood says:
May 15, 2013 at 11:55 am

The late John Daly referenced Fairbanks, upriver, as a possible factor in the dates. One of many power plants there consumes more than 210,000 tons of coal a year.

Interesting thought, Fred. Let’s see. The coal is from the Usibelli mines, and that coal has about 1.64e+10 joules per tonne. Then 2e+5 tonnes of coal would have about 3.3e+15 joules of energy. Say 40% is used to generate electricity, so you’d assume that 60%, or 2e+15 joules of heat would be lost, mostly going into the river cooling water.
But what John Daly might not have known is that because of the extreme cold, the power plants in Fairbanks are co-generation plants. In addition to electricity, they produce both hot water and steam, for residential, commercial, and city use.
So during the time the Nenana River is frozen, much of the waste heat is going to providing steam and hot water rather than to warming the Tanana River.
Now, co-generation is typically pretty efficient. So I’d estimate that rather than 60% of the total heat content of the coal going into the river, maybe 10% heats the river water. That’s still a big number, 3.3e14 joules of heat each year. That works out to 9e+11 joules entering the river daily.
It takes about 4.2e+6 joules of energy to heat a tonne of water one degree C. So how much water is in the river? Man, I love the web … here’s the Tanana river flow in Fairbanks earlier this month. It was about 25,000 cubic feet per second. That’s about 708 cubic metres per second, which is 6.12e+7 tonnes per day.
So now, we have all the information we need. We have 9e+9 joules heating 6.12e+7 tonnes of water each day, at 4.2e+6 joules per degree per ton, crunch the numbers, the Tanana River is warmed about 0.003°C by the power plant … so even if there were ten plants and twice the waste heat per plant, it’s still in the hundredths of a degree. But there aren’t ten power plants upriver of Tanana, there are three. The other two together use about 120,000 tons of coal per year. So the heating is still trivially small.
Check my numbers, I’ve been known to make mistakes …

Willis Eschenbach

Greg Goodman says:
May 15, 2013 at 1:12 pm

Willis, where did you get the numbers from?

As I said in the head post,

DATA: the historical breakup days are available here.


Bruce of Newcastle

Just now gone past 1935 into third place. Second looks set to go too in another 19 hrs, the ice still looks very solid.


I’ve read that to be near a river during ice break-up, is a thing to behold.
Just for the sounds.
Maybe one of these days.

it’s the only tripod with four legs.

Navy Bob

Looks like a nonopod to me (one less than the decapods doubtless waiting below to see the sun).

Thanks, Willis.
Interesting geography lesson.


With nine legs, its got to be a Nenanapod.
A Nenananonopod?

Felix Scholl

The ice break up on the Moose River at Moose Factory, Ontario probably has long record of break up dates. This is also a north flowing river, but at lower latitude. Break up was last week for 2013.

A few things for your consideration:
Fairbanks gets most of its electricity from burning diesel these days. The Healy clean coal plant has been the center of a decade long pi$$ing contest between Golden Valley Electric (GVEA) (Fairbanks area electric co-op) and the State of Alaska since 2000. It has just recently been transferred to GVEA and may get fired up depending on how hard the EPA fights the air quality permits.
One mode of travel in the Bush during the winter is to drive diesel pickups up and down the rivers while they are iced over. Still have to be careful as there are springs.
Interior Alaska has more than a little trouble with flooding after break up, as it is not uncommon to get ice dams on the rivers and flood the surrounding countryside. This is the other reason that breakup is monitored closely – first you can’t travel on the ice any more and then the ice and ice cold water comes to visit for a while.
Still a lot of snow on the mountains around Anchorage. Rivers and creeks are going to be high for most of the summer as it melts and runs off. Cheers –

Gary Pearse

PDO oscillating? Well it looks like it mirrors the global temp trace – 1936 high, the following lows, upturn in the 80s and 90s to 1997-98 and then curving downward into cooler temps. Since it is argued that that global temp curve is a signature of CO2 increase (except for after 1997 – a pretty damning blow to CO2), I guess faithful can argue its mainly CO2 and the darn ocean began swallowing heat (for some reason) after 1997. Or, all we need are the oceanic oscillations to explain the whole shitteree.

Greg Goodman

“As I said in the head post, DATA: the historical breakup days are available here.” Awe crap , you mean I have to go through and format it all by hand !

Greg Goodman

BTW last years data seems to be missing ? Did you find it ?


Under the “Ice” heading of their website, they have weekly ice thickness records from January through May (or until breakup) for every week for every year since 1989.
A quick look shows no significant pattern, nor any visible changes from this year to the two recent years of most significant Arctic Ocean minimum sea ice extent: 2007 and 2012.
So, is minimum Arctic sea ice extents really indicative of any “temperature proxy” anywhere else in the Arctic at all?

Donald L. Klipstein

The graph appears to me reasonably honest. It appears to me as usable
to support a contention for a warming trend, or for lack of warming in the
past 25 years.
Keep in mind that this is a specific location, as opposed to a region or
the world as a whole. I would give more weight to satellite-based
determinations of global lower troposphere temperature. I see those,
despite the century-class 1998 spike, indicating 12 years of lack of
warming as opposed to the bantied-about 15-17 years.
Not that I think manmade global warming exists to the degree that most
advocates of its existence insist upon, but I see that it exists, as surely as
a roughly ~65 year (if temporarily so) natural cycle – which shows up
well in HadCRUT3.

Willis Eschenbach

Greg Goodman says:
May 15, 2013 at 7:07 pm

“As I said in the head post, DATA: the historical breakup days are available here.” Awe crap , you mean I have to go through and format it all by hand !

Here you go, we’re a full service website …