Spring Contrasts

Clear skies and calm winds produced a widespread freeze across eastern and northern Alaska this morning, with the temperature dropping into the 20s in most locations.  Fairbanks airport reached 29°F, which is a little unusual in the modern climate for this late in the season; other than last year, it hasn't happened after May 10 since 2013.  The low temperature in Chicken was 19°F, which is the first time below 20°F this late in the season since 2014.

Farther north, accumulating snow fell in Fort Yukon yesterday morning.  The accumulation was probably less than an inch, given that the snowfall rate was light, but temperatures were below freezing, and it was falling for over three hours.  Snowfall measurements haven't been available from Fort Yukon for many years, but historical data shows an inch of snow as late as mid-June in 1971, and as early as late August in 1961.  It's a short summer in the northern interior.

Farther north still, the North Slope is still locked in with wintry weather.  The high temperature in Utqiaġvik yesterday was only 13°F, and that's a little more notable: we have to go back to 2007 to find a colder day this late in the season.  Utqiaġvik temperatures have been oscillating between above and below the 1991-2020 normal so far this year:

The contrast between North Slope winter and interior spring made me wonder when the seasonal contrast between the two regions is greatest.  It is indeed in spring: the following chart shows the difference between normal temperatures in Utqiaġvik versus the average of Bettles, McGrath, and Fairbanks.

The Arctic coast doesn't begin climbing out of winter cold until nearly April, but the interior has gained a lot of warmth by then, and the temperature contrast peaks in late April.  Of course, the interior cools off much more quickly in autumn, and the valley-level interior becomes briefly colder than Utqiaġvik in November.

Looking back at April, it was a cloudy and wet month for most areas except the North Slope and Southeast.  With precipitation well below normal in the usually wet Southeast, the state-average precipitation was not much above normal, but the southwest and southern interior were unusually damp for the time of year.

The broad swath of wet weather was caused by strong westerly flow to the north of a robust Gulf of Alaska ridge, and downstream of the ridge the dry anomaly was inevitable over Southeast Alaska.

Statewide mean temperatures in April were very close to the modern normal, with equally balanced ups and downs during the month.  The Aleutians were significantly warmer than normal, although not to the extent of some recent years (2016, 2020).

Cloud cover was well above normal for most of the state.  The combination of cloudy and damp weather certainly wasn't a bad thing in terms of wildfire risk as we advance quickly toward fire season.

Big El Niño Coming

Most readers are probably well aware that El Niño is on the way for next winter; the forecasts have been well-publicized.  Normally one would be wise to take El Niño forecasts with a grain of salt at this time of year, because there is a notorious "spring barrier" of predictability in Northern Hemisphere spring, but in this case the signals are so strong and coherent that there's little doubt of the outcome.

The most compelling evidence that a major El Niño is about to emerge can be seen in the profile of subsurface temperature anomalies in the equatorial Pacific.  The figure below is a couple of weeks out of date, but it shows a very intense warm anomaly below the surface, and this is migrating eastward and upward.  As the warmth continues to emerge at the surface, East Pacific equatorial surface water temperatures will become much warmer than normal, consistent with a strong El Niño signature.

The warmth produced by the subsurface anomaly will add to the substantial surface warming that has already occurred in the last few months, so it seems inevitable that the equatorial SST anomalies are going to become quite extreme in the next few months.  The canonical Niño3.4 SST index is already near 1°C above normal, although this includes a contribution of about 0.5°C from long-term warming in recent decades:

Here's a chart showing the recent progression of Niño3.4 SST anomaly forecasts from ECMWF's seasonal model.  The red line is the latest forecast; the model has been adjusting warmer with every run in recent months.

If the forecast is correct, the Niño3.4 anomaly may approach +3°C by autumn, with the modern record being +2.7°C in November 2015 (relative to a 1991-2020 baseline).

What does this mean for Alaska?  We'll discuss prospects for next winter another time, but for summer the history of past years with strong El Niño conditions (see below) suggests a heightened probability of above-normal temperatures, driven by relatively high pressure (ridging) from Alaska to the Canadian Arctic.  The top analog years also show dry summer conditions in Southeast Alaska and hints of dryness in southern interior and western Alaska.  Note that this is based simply on the 10 years since 1950 with the most positive Multivariate ENSO Index in June through August.

However, not all El Niños are created equal, and analog years show a lot of variability.  To illustrate this, here are temperature percentile maps for six summers when a robust El Niño developed by autumn following La Niña-like conditions the previous winter.  Where possible, I've used a preceding 30-year baseline for a fair comparison across the decades.

More recently, 2015 produced a very strong El Niño, but that episode didn't emerge from a cool phase in the previous winter.  There was also a robust El Niño in 2023, although its oceanic warm signature was inflated by unusual global warmth.

Not surprisingly, temperatures from the northern Gulf Coast to southwestern Alaska tend to be linked to the PDO phase, which was positive in 1957, 1986, (especially) 1997, and 2015, but negative in 1965, 1972, and 2023.  Elsewhere, variability is large among the analog years, although there isn't much cool to be found in northern and western - or southeastern - Alaska.

Similar maps for precipitation are included below.

It's difficult to pick out any consistent moisture signals on the maps, but interestingly ground-truth data from Fairbanks shows that none of the top 6 El Niño analog years had above-normal rainfall for May through September.  Click the figure below to enlarge (2014 is included only to show the record wettest summer):

Juneau also shows distinct indications of a (relatively) dry summer when El Niño develops in a similar way:

Summer rainfall can go either way in Anchorage, however:

For a more comprehensive look at all the seasonal climate guidance for this summer, I highly recommend tuning in to Rick Thoman's briefing on May 22:

https://uaf-accap.org/event/may2026-climate-outlook/

Arctic Sea Ice Update

Notwithstanding very abundant sea ice in the eastern Bering Sea this winter, Arctic-wide sea ice is now vying with 2019 for record-low extent at this time of year; it has been running near the bottom of the "pack" (pun intended) since November.  The following chart provides a comparison to 2019 (the previous record low at this date), 2012 (the year of the record low in September), and last year.

Deficits relative to the 1981-2010 normal are widespread across the marginal seas of the Arctic:

It's an interesting question as to whether the developing big El Niño episode will have a positive or negative effect on Arctic sea ice this summer.  Several of the strongest El Niño events of recent decades developed in parallel with substantial year-on-year declines in Arctic sea ice extent, so the risk of another major drawdown would seem to be elevated.  Here's a scatterplot of year-on-year change in September ice extent versus the June-August Niño3.4 SST anomaly; the notable recent years are 2023, 2015, 2002, and 1997 (lower right quadrant).

However, it's equally clear that there is actually not much correlation overall between year-on-year ice change and the phase of ENSO during summer.  In particular, pre-1997 El Niño events didn't have a consistent relationship with ice changes, and many stronger La Nina summers show modest ice declines rather than gains.

For a bigger picture, here's the correlation of June-August detrended SSTs with the year-on-year ice change:

The colors indicate the SST anomalies that would normally be associated with an increase of sea ice.  Clearly, the North Atlantic and North Pacific have more correlation than the tropical Pacific, but this doesn't imply causation; the SST patterns revealed here may simply reflect the impact of circulation patterns that affect sea ice extent.  Nevertheless, it's interesting to observe that a negative PDO pattern (like recent years) tends to favor higher sea ice extent, and that's confirmed in a scatterplot - although the correlation is not statistically significant:

The atmospheric anomalies themselves show a greater connection with sea ice change, of course; but those anomalies are relatively unpredictable at seasonal time scales.  A positive Arctic Oscillation, for example, slows the summer ice melt, whereas a negative AO phase tends to bring a reduction in ice extent.

ENSO does tend to be inversely correlated with the AO phase in summer, so it's more likely than not that we'll have a negative AO pattern this summer, i.e. more high pressure over the Arctic (and favoring warm and dry conditions in much of Alaska).  This points toward greater sea ice loss, as in other recent El Niño summers; but time will tell if this year fits the same pattern.

Snow Analysis

Fairbanks reached 50°F yesterday for the first time this season - a couple of weeks later than normal - and thaw season is now fully under way, with no sign of an imminent return to unusual cold.

Winter's last gasp earlier this week had a sharp bite to it in the interior and north, with overnight temperatures dropping well below zero at many locations.  Here are a few notable readings from Wednesday morning:

-24°F at the Birch Creek HADS

-16°F at the Livengood RAWS

-17°F in Bettles and Coldfoot

-27°F at the Dahl Creek HADS

-20°F at the CRN site to the east of Kaltag

It's pretty late for this kind of cold.  In Bettles, for instance, only one year (since 1951) has recorded a temperature below -15°F after April 20, and that was only marginally later in the month (April 22-25, 1966).

Accumulating snowfall is likely finished in Fairbanks, so the tentative total for the season fell just short of 100".  It was the snowiest winter since the triple-crown big snow winters of 1990-91 through 1992-93.

In view of the Fairbanks snow onslaught in February, and similar heavy snows in Anchorage and Juneau, I thought it would be worth looking at the contribution of the snowiest days and months to seasonal snow totals at all three sites.  For example, the top 10 snowiest days of each winter generally contribute about half of a winter's total snow accumulation in Fairbanks; but has this changed over time?  Fairbanks typically "nickels and dimes" its way to a substantial snowpack, but is this more true of recent years or of an earlier climate?

The answer is there hasn't been any significant change over the decades.  Click on the following chart to enlarge; it shows the seasonal snow totals in light blue, the contribution of each winter's 10 snowiest days in dark blue, and the ratio of the two totals in red.

There are slight increasing trends in snow totals in Fairbanks, and there's a hint of a decreasing trend in the "top-10-days to seasonal total" ratio, implying that heavy snow days used to contribute a bit more of the total snowfall - but the trends are nowhere near statistical significance.

Another perspective on the question is to look at the snowiest 30-day period each winter and calculate its contribution to the seasonal snow total.  Again, there's a slight decrease in the snowfall fraction attributable to the snowiest month, but it's not a significant change.

Anchorage also exhibits slight - but not significant - upward trends in snowfall, both in terms of seasonal totals and each component of the segmentation.  The snow fraction derived from top-10 days and the snowiest month haven't changed significantly over the 72 years of data in Anchorage.

A more striking result is found in Juneau, where both fractions have increased very significantly in the 82-year climate history.  The snow totals themselves haven't changed significantly, but there's clearly a substantial increase in the concentration of snowfall in the heaviest episodes, starting in 1976-77.

It seems to me that the change in Juneau must reflect the increase in winter average temperatures that also developed in the late 1970s, when the PDO phase turned positive.  In Juneau's mild climate, warmer winters have more days with rain rather than snow, and therefore the snow events make up a larger fraction of the winter's overall snowfall.  (To understand this, consider the extreme of a much warmer climate where snow is rare; 100% of the snow will occur in the top-10 snowiest days each winter.)

The surprise, perhaps, is that Juneau hasn't seen a more significant decline in total snowfall; there is a negative slope, but it's not close to statistical significance.  However, it seems possible that the inter-annual variance is increasing; many recent years have had low snow, but there have been very snowy winters too, culminating in this winter's all-time record.

Winter Storm Warnings

Winter storm warnings have been hoisted from the Yukon Delta and eastern Norton Sound to the western interior, as a decaying Bering Sea low joins forces with an Arctic trough to bring significant snowfall.  Here are snow forecast graphics from the NWS (click to enlarge):

Interestingly, however, the model forecasts have adjusted warmer for next week, with cold disappearing quickly by mid-week; warmth will quickly surge up from the southwest, replacing the cold air mass centered over northwestern Alaska.  Here are the latest 850mb temperature forecasts for late Tuesday, with cold poised to depart:

If warmer weather can stick around for the rest of the month, as seems possible, then thawing will start to make some headway.  Of course we don't want it to get too warm too quickly.

Here's a historical perspective on Fairbanks temperatures in previous years when La Niña transitioned rapidly into El Niño between winter and summer, as it is doing this year.

Interestingly, this set of "analog" years suggests that the odds most favor below-normal temperatures for 30-day periods ending late April through late May - or right around breakup season.  The median breakup date at Nenana for these 10 years is May 6, i.e. a few days later than normal.  However, not all ENSO transition years are cold around this time; two of the last three (1997 and 2009) were not, and Nenana breakup was on April 30 and May 1 in those years respectively.