Saturday, March 26, 2016

More on Howard Pass

In last week's discussion of Brooks Range wind chill, I mentioned the fact that temperature tends to decrease as wind speed increases at Howard Pass.  An inverse relationship between wind speed and temperature is very different from what most Alaska residents experience - in most locations it's coldest when the wind is calm - so I thought it would be worth taking a closer look at the Howard Pass data.  I obtained the hourly observations (2012-present) and looked at the data from several angles.

First, the distribution of wind speed in December through February (see below) does not show anything too unusual, although it's obviously a windy place; the wind speed exceeds 40 mph about 15% of the time in winter.  And as we know, very high wind speeds sometimes occur.  Last Saturday's sustained wind speed of 98 mph (during the 7-8 pm hour) was the highest on record for the station; this measurement probably wouldn't have been possible without the new robust anemometer that was installed last summer.

The distribution of hourly temperature measurements is more unusual, showing a double peak, with the temperature most often falling between -10 and -20°F.

If we look at the joint distribution of temperature and wind speed, we see very clearly that the highest wind speeds are invariably associated with below-normal temperatures, and in fact significantly below-normal temperatures hardly ever occur without strong winds.  For temperatures of -30°F or below, the wind speed is typically above 40 mph.  We also see that calm winds are uncommon when temperatures are well above normal, so there is a minimum in median wind speed for temperatures near the climatological normal.

Looking at hourly wind chill values, the distribution shows a pronounced double peak, and the most common wind chill is between -40 and -50°F.  This is quite remarkable for a location at only 2062' elevation (although it is above the Arctic Circle).  Bear in mind that wind chill is undefined for wind speeds less than 3mph, so quite a number of calm-wind observations are excluded from the distribution.

Another way to look at the joint distribution of temperature and wind is to plot a kind of wind rose diagram - see below.  The red line shows the median wind speed for each point on the compass, and the blue line shows the median temperature.  Interestingly there are two peaks in wind speed; the primary one occurs when the winds are from the north-northeast, through the pass, and these winds import cold air from the North Slope, leading to extreme wind chill - the median temperature for this wind direction is -16°F.  The secondary peak in wind speed occurs for east-northeasterly winds, and in this scenario warm air reaches Howard Pass; the median temperature is about +20°F.

Here are two images showing the location of the Howard Pass RAWS relative to the surrounding topography (click to enlarge); in the first image, the Howard Pass site is the one with the red "20" next to it, at the northeast edge of the Noatak River basin.

It would be very interesting to look at the other RAWS sites scattered around the western Brooks Range, to see if any others show the same kind of behavior.  I plan on doing this when I have time to obtain and parse all the hourly data.  For now, I just looked at the data from Anaktuvuk Pass - see below.  Surprisingly, Anaktuvuk Pass appears to be remarkably non-windy in winter; the absence of strong winds in this data almost looks unrealistic, although a topographic map reveals that the site is actually very sheltered from most directions.  There is a slight increase of wind speed with warmer temperatures and there's no evidence of the cold northerly gales that are observed at Howard Pass.

For comparison, the Umiat RAWS on the North Slope shows behavior that is more typical of low-elevation Alaska in winter; wind speeds are typically light, and higher wind speeds are associated with relatively warm temperatures.


  1. I've been nearby Howard Pass in summer and winter, but mainly on the larger lakes and tributaries to the Noatak. Not an expert here but I do have an appreciation for the location and topography.

    I suggest an elevation profile of the HP RAWS WX station and drainage through the pass might be interesting, especially if compared with Anaktuvuk Pass with the Anaktuvuk River flowing to the north, and the John River that flows to the south. Something like +-10 miles in either direction for example.

    Howard Pass is a relatively short constriction in the terrain, whereas Anaktuvuk is a more gradual river valley of some greater extent.

    When the wind blows N>S the POES satellite IR shots of Howard Pass show a visible marker stream compared with nearby mountain passes in the Northern Brooks Range. I suspect it's a function of extreme temperature and pressure differential over a short distance between the North Slope and the Noatak River drainage and nearby ocean to the SW.

    Anaktuvuk is located considerably further to the east and away from potentially lower pressure or warmer temperatures. Just a guess however.


    1. I agree, Gary: pressure and temperature gradients are to blame. The low-level temperature inversion presumably acts as a lid to create a vertical constriction as well.

      Good idea re: elevation profile; I'll see what I can do.

    2. I guess I'm curious about the profile because of the winds and any effect it may have in increasing them local to the station.

      There's two drainages (~N and NE) potentially driving the winds from those directions and perhaps combining (Fig. 7 above). If the elevation (2062) rises and falls quickly N>S then I can see something interesting happening, but am not sure if it's sloped katabatic or Bernoulli in nature. The vertical capping you note may be the real answer.

      No doubt it's a site worth investigating the physics behind the winds.


    3. By Bernoulli I meant that the constriction of Howard Pass due to local terrain (vertical and horizontal) in the N>S flow pattern typical of high winds creates an area of increased velocity (and presumably lower air pressure) that the station then reports.