Some Errors in the NCDC U.S.A. State Weather Extreme Records Data: Part 1

By: Christopher C. Burt , 8:21 PM GMT on March 18, 2011

Share this Blog
2
+

Some Errors in the NCDC U.S.A. State Weather Extreme Records Data: Part 1

Last year the National Climate Data Center (NCDC) created a new web site that neatly summarizes various state weather records: specifically what the extreme values for hottest and coldest temperatures, maximum 24-hour precipitation and 24-hour snowfall, and maximum recorded snow depth for each individual state are. This may be viewed at the NCDC web site.

The interactive map they provide and the consolidation of these records into a single user-friendly web site is a big improvement over the way they previously displayed this information (which was by listing separate tables and maps for each of the values in question). The NCDC also updated and revised the figures through the year 2009 using their collection of COOP station forms maintained in the NOAA database. These may be viewed here at the IPS web site.

Over the next few weeks I will post a series of blogs concerning mistakes, some obvious and others more subjective, that I and others have discovered in the course of fact-checking these records.

Some Errors with the Official State Records: Case of Hawaii

For some reason the extreme weather records for the state of Hawaii are fraught with errors. Both the ‘official’ record high temperature of 100° at Pahala, Big Island on April 27, 1931 AND the official low temperature of 12° at Mauna Kea Observatory on May 17, 1979 are obviously wrong.

HAWAII RECORD HIGH TEMPERATURE

Unfortunately, the COOP forms for Pahala have been digitized and the original form no longer appears on the IPS site mentioned above. However, the digitized version of the form is available for viewing at the excellent Utah State University Climate Center’s ‘Climate Database’ (see at the USU Climate Center's web site.)

Looking at the Pahala dataset for April 1931 we see that the maximum temperature for April 27 is, in fact, missing and represented with the figures “9999”, meaning missing data. For some reason this has been interpreted by the NCDC to mean 100°!




The digitized version of the Pahala, Hawaii COOP form for April 1931. Reproduced from the Utah State University Climate Center database.

We can see that the day before April 27th the maximum temperature was just 82°F and the day after just 80°F. Furthermore, the observation time temperature (likely made at 7a.m. local time) was a fairly normal 70° on April 27th. It is amazing that this bogus 100°F reading from Hawaii has been on the record books for some 80 years already!

So just what is the real record high temperature for the state of Hawaii? Unfortunately, this is very difficult to determine. No 99°F reading has ever been made but there have been six (6) different occasions that 98°F has been reported:

At Waianae, Oahu in July 1915

At Mahukona, Big Island on an undetermined date prior to 1941

At Kaanapali, Maui on an undetermined date prior to 1941

At Puunene, Maui on Aug. 19, 1951

At Puunene, Maui (again) on July 14, 1957

At Sea Mountain, Big Island on Sept. 1990

After extensive research and investigating the COOP forms (where available) for the above sites, it would seem the least suspicious figure from the above list is Puunene. The two 98°F readings from here were made at a small airport about 4 miles southeast of Kuhului on Maui. The inland and low elevation setting make this location a reasonable candidate for hot weather in the otherwise temperate Hawaiian Isles.



COOP form for Puunene, Maui for August 1951, one of the two months that the station reported 98°F temperatures.

Interestingly, this means that Hawaii is now the ‘coolest’ state in the Union so far as record high temperatures are concerned. Even Alaska has recorded a 100°F reading (at Fort Yukon on June 27, 1915).

HAWAII RECORD LOW TEMPERATURE

The coldest temperatures on record for Hawaii are, of course, reported from the several mountaintop observatories in the state. Mauna Kea Observatory located near the summit of said mountain at 13,796’ has had a potted history of weather observations. I say potted because for some reason the COOP forms from this site have been stamped as “unreliable” for several years (1976-1978) of observations during the 1970s. This includes the month of January 1976 when a 9°F reading was made on January 12th of that month and a 12°F reading on Feb. 22, 1977. The ‘official’ low of 12°F on May 17th, 1979 is obviously an error when one looks at the COOP form for that month:



COOP form for the Mauna Kea Observatory in May 1979. A new minimum thermometer was installed on the 5th of the month.

One can see the observer has circled the 12°F reading and replaced it with a 21°F note and question mark (see above). Furthermore, the low temperatures on both the day before and after May 17th were just 23°F lending further doubt to the 12°F reading.



The summit of Mauna Kea in the winter of 1971 with snow down to about 11,000’ (3,350m). Photo by Dale P. Cruikshank.

So what might be the actual lowest temperature for Hawaii? A 14°F reading from the summit station at Haleakala, Maui (elevation 10,023’) on Jan. 2, 1961 is the 2nd coldest reading from the state. This, however, is also unbelievable since the low temperature from the previous and following days was just 34°. Twenty-degree swings in minimum temperatures from one day to the next are highly improbable in tropical environments like Hawaii. This leaves a 15°F reading from Mauna Kea Observatory on Jan. 5, 1975 as the mostly likely reliably measured temperature in the state on record.

Next week I will look at some other glaring errors from the ‘official’ records in some other states.

KUDOS: Maximiliano Herrera, Howard Rainford, and Trent McCotter.

Reader Comments

Comments will take a few seconds to appear.

Post Your Comments

Please sign in to post comments.

or Join

Not only will you be able to leave comments on this blog, but you'll also have the ability to upload and share your photos in our Wunder Photos section.

Display: 0, 50, 100, 200 Sort: Newest First - Order Posted

Viewing: 4 - 1

Page: 1 — Blog Index

4. wxhuskie
9:17 PM GMT on November 04, 2013
If you take a look at the April 1931 CD ("Climatological Data"), published on August 1, 1931, you will see that the 100 degree Pahala temperature is mentioned. So, obviously it's not a NCDC misinterpretation of "9999."

On the original co-op form, the observer clearly wrote "100" for the MaxT on the 27th, and also mentioned in the comments that is was the hottest temperature "in nine years of record." However, someone else came along later and drew a faint line through the 100 and annotated it with an M to the left of the entry.

So, is the 100 wrong? Very possibly. Given the temps the day before and after plus the morning low on the 27th, and Hawaii's tropical atmosphere (meaning large temperature swings are very unusual), I'd guess that the 100 is probably incorrect (even though the observer believed it). Still, it has nothing to do with NCDC misinterpreting 9999, since the value was listed as truth in the April 1931 CD.
Member Since: November 4, 2013 Posts: 0 Comments: 0
3. liza1314
6:38 AM GMT on April 22, 2011
2. DocNDswamp
5:21 PM GMT on March 20, 2011
Hiya Chris,
Thanks, another excellent entry and you're findings sure ring familiar!
Bear with me as I contribute some of my own.

My thoughts on the high altitude Mauna Kea low teen reading have leaned being plausible, well, until we see the multi-day trend you've posted... and notations / corrections from the original data sheet!

Yes, does raise questions how apparent errors - a standard "9999" misinterpretation, serious typos, or failure to account the observer or Weather Bureau met's corrections, etc - can become installed into the historical record, but as I've researched into my local COOP station's records (and it's incarnations since 1893), comparing datasets for several years now, I'm not surprised you're finding them too! ... Obviously all-time state record data should be closely examined and reasonably verified, but since I've also found really screwy errors in the records - or the translation into databases - from my local station, it makes one wonder how badly rampant these temp record mistakes are.

Here's one example that I've mentioned several times before:
The advent of the IPS site has been a boon in my own efforts - so far I've downloaded 163 months worth, saving 89 of those original wx observers pdf's, discarding the other 74 after examination / notation. Along with adding insight on methodology, diligence, TO periods, seeing notations / corrections, (and along the way, respecting those diligent, dedicated souls even more!), having the original observers sheet sure clarified what I know is a glaring error in previous / current datasets that continue to show my location, Houma LA (HUML1 / LA 164407), as having recorded it's coldest all-time record 4F low on Dec 21 1960... Yet the pdf copy of wx observer's datasheet shows was clearly marked with a 34F low on that date! Was nothing ambiguous about his reportage, yet the 4F low - a whopping 30 degree departure from reality(!) - has been listed since the SRCC records, and this error (along with others) continues to populate the datasets, including Weather Underground's Houma records... Having lived thru 6 decades in the same SE LA location as a wx observer, I remember all our coldest outbreaks, and this isn't one of them!

That said, and despite our relatively mild N Gulf location, we HAVE gotten brutally cold into the single digits - our all-time record low is 5F, set like many others nearby, on Feb 13th 1899... And we recorded 10F on Dec 23 1989 (with outlying nearby areas reporting 9F), during what became locally the coldest arctic outbreak of the 20th century...

But it gets worse -
Conflicting datasets abound, adding to the confusion with various "start date" periods being used, especially viewing records via internet sites, including as mentioned, here at Weather Underground and The Weather Channel's data bases. Unfortunately, they have eliminated many notable older records in using 1930 start date for my area... So it fails to reveal Houma's astounding all-time record snowfall of 16" in Feb 1895, our true record low of 5F in Feb 1899, AND our all-time record high of 104F on June 22 1915 - ALL of which appear legit! Equally as bad, they're now comparing that 1930 start dataset - which came from the still fully operational HUML1 Ag Center COOP station used by state's LOSC, and the SRCC, NCDC and IPCC - with the KHUM airport data, which only recently updated to a full-time 24 hr AWOS station on May 11 2010! All previous KHUM data prior to that date was minus ~ 12 hrs of overnight obs... Yet the new KHUM AWOS still doesn't log detailed minute data or tally precip as the state-of-art HUML1 COOP does! Contributing more confusion since they reference airports, the New Orleans NWS office has offered same erroneous comparison of "new" KHUM data to "older" HUML1 data in local newspaper articles recently when we've approached near record temps... Sadly, you'd think the NWS mets should know better... Having personal knowledge of one's local weather history, quirks in reportage, station histories and such is immensely helpful in sorting out what is truly a mess! LOL, "trust the data", indeed... with a healthy dose of skepticism and prudence!

PS: Yikes, I just checked the IPS site, there is currently a SERIOUS glitch ongoing - only showing 4 more recent files available! With the header stating - "Notice: All publications have limited periods of record online. NCDC is working toward restoring full access as quickly as possible."

Let's hope!
;)

Member Since: September 21, 2005 Posts: 94 Comments: 4783
1. atmoaggie
3:06 AM GMT on March 19, 2011
Looking at the Pahala dataset for April 1931 we see that the maximum temperature for April 27 is, in fact, missing and represented with the figures “9999”, meaning missing data. For some reason this has been interpreted by the NCDC to mean 100°!

Whoa. Really? Are there other instances of the same issue in other places? (As a meteorologist and programmer this is a little painful to see.)

Clearly, some QC algorithm would remove 100 for, say, Thule, after the null value is read or interpreted as 100 F. Probably not for a clime where 90 F happens once a year, or so.
Member Since: August 16, 2007 Posts: 6 Comments: 12461

Viewing: 4 - 1

Page: 1 — Blog Index

Top of Page

About weatherhistorian

Christopher C. Burt is the author of 'Extreme Weather; A Guide and Record Book'. He studied meteorology at the Univ. of Wisconsin-Madison.