Tuesday, October 7, 2014

First Sub-Zero Day

It appears that the first sub-zero observation of the season occurred today at several locations. The RAWS station at Norutak Lake (west of Bettles and south of the Brooks Range) reported a morning low of -2°F. While RAWS data is not always reliable, the Ruby 44 ESE USCRN station was also below zero (-1°F). This value is certainly bullet-proof. Figure 1 shows a screen capture of the low temperatures in the western interior this morning. The -2°F at Norutak Lake is circled. (The Stoney, Alaska, RAWS station may have been even colder but they are not reporting at the moment)

How does this date compare to other seasons? It is actually later than the long term average. Last year was the latest on record but the 1st of October is more typical. Figure 2 shows the date of the first occurrence of sub-zero temperatures in Alaska since 1950. Data extends prior to 1950 but due to the sparsity of stations, statewide assessments are difficult to make.

Finally, the area around the Brooks Range is a common location for the first sub-zero temperature to be measured. Figure 3 shows the location of the season's first sub-zero observation since 1950, If a tie occurred, the station with the lowest temperature was mapped.

Figure 1. Low temperatures in the western interior on the morning of October 7, 2014. Data courtesy of the University of Utah's Mesowest site.

Figure 2. Date of first sub-zero reading in Alaska from 1950-2014. Values obtained from GHCN v.3 and GSOD.

Figure 3. Location of first sub-zero reading of season between 1950 and 2014. If more than one station was below zero on a date, the lowest reading was selected.


  1. Very nice, Brian - I was curious about this myself. It's interesting how large the spatial variability is - one might expect a few locations to dominate. On the other hand, perhaps it's common for conditions to be favorable over a large area on that first cold night (like last night), and there are many possible candidates for coldest spot.

    1. Thank you Richard for putting the GSOD data together. I combined the GHCN and GSOD for this analysis. In two of the years, the GSOD data moved the first sub-zero date up by 20 days! However, about 2/3 of years the GHCN captured the low temp.

      As for the geographic distribution, the non–Brooks Range locations tend to be years where the occurrence is later. It might mean nothing more than it was cloudy for most of mid-September to Mid-October in the Brooks Range and so the temperature was never able to drop very low.

    2. I realized that the only sub-zero locations, including today, are either RAWS or new CRN locations, which were not included in the historical analysis, so it may be an unfair comparison with earlier years. I'm not aware that any COOP, city, or airport site has hit zero yet. Here are the sites that have, from my analysis:

      Ivotuk CRN
      Ruby 44ESE CRN
      Bettles Field SNOTEL
      Norutak Lake RAWS
      Stoney RAWS??

  2. The specified criteria for accuracy in determining air temperature are similar between the RAWS and CRN facilities:

    RAWS: +-0.6C @ +-50C: http://raws.fam.nwcg.gov/nfdrs/pms426-3.pdf

    CRN: +-0.3C @ +-50C: http://www1.ncdc.noaa.gov/pub/data/uscrn/documentation/program/X040_d0.pdf

    Whether the product reflects design criteria is another matter. I'd want at least three separate sensors (or watches, an old joke) if I wanted to better know the temp (or time).


    1. Gary, thanks, good information. My point was not to question the quality of RAWS or (heaven forbid) CRN, but to indicate that most of the historical record does not include these locations. As we've discussed before, many of these new remote sites are colder, so we have to be careful with historical comparisons.

      As an aside, this morning we did have sub-zero at a couple of sites with long histories: -8F at McKinley Park, -1F at Big Delta. The coldest I saw was -17F at Marguerite Creek RAWS near Healy 13NE; Denali Visitor Center was -10F. Not bad for the time of year.

  3. Perhaps Rick can chime in on the CRN site selection criteria. Clearly the goal of the project is to have a long-term (50 years) data history at a variety of locations that are not likely to experience urban or land use changes during those 50 years. I suspect that in Alaska they want to have a wide variety of settings; e.g., forested, tundra, permafrost, mountainous, interior, southeast, Aleutian, Arctic, upland, valley, cold, warm, etc. To a lesser degree the RAWS stations are also put in somewhat remote places to fill in data gaps (although I am not familiar with the selection criteria). There's probably a good reason why no settlements are at those locations that are especially cold.