A couple of days ago, I located the “long lost” Honolulu Observatory GISS weather station on the Island of Oahu with just a couple of hours of digging. That one apparently got “lost” because the station name changed, and the inter-agency communications seemed to be the cause, and nobody at GISS bothered to look to see if there was still current data coming from the station.
Today I found one in under 5 minutes. I wasn’t even planning on looking for one, it happened by accident. I was watching the Discovery Channel TV show this afternoon “Deadliest Catch” where crab fishermen brave the worst imaginable weather to keep crab shacks running nationwide. They are based out of Dutch Harbor, Alaska.
While watching a scene where they were coming into the docks, I saw an ever so brief flash of what looked like a Stevenson Screen off in the distance near the docks. I hadn’t expected to see one and I wasn’t 100% sure, but I thought I’d check NCDC’s metadatabase (MMS) for Dutch Harbor, AK. Sure enough, they have a COOP station there with a Stevenson Screen there that is “current”.
When NCDC says “MAX-MIN THERMOMETERS”. that means mercury thermometers in a Stevenson Screen.
The MMS Location description tab had this:
Topographic Details: TOPO- STN NEAR WATERFRONT, ACROSS HARBOR FROM DUTCH HARBOR ARPT IN VILLAGE OF UNALASKA. MNTOUS ISLAND. STN EXPOSED TO STRONG TURBULENT WINDS AND TEMPS INFLUENCED BY SURROUNDING WAT
I assume the last word was to be “WATER” and they ran out of characters in the database field. The MMS database also mentions it to be located at REEVES TERMINAL, which I assume is the ship terminal/dock.
Here is where the lat/lon given by NCDC places it, I seem to recall my brief glimpse was closer to the docks visible near the top of the image, but the lat/lon given by NCDC is not always accurate:
So I was curious to see what the temperature record looked like in this very remote part of Alaska. I figured if it was an active weather station, GISS would have a plot of it. They did indeed but it was not what I expected to find:
Here is the link to the GISTEMP graph above.
Huh. Big data hole. But NCDC said it is current? WUWT?
So I decided to look at the COOP section of NCDC, and sure enough it WAS current to April 2009. I found the most current B91 form and downloaded it.
Here’s the PDF of the form: Dutch_Harbor_April2009
Seeing the big hole in the GISS data, I decided to look for the data forms backwards, and sure enough, it is current all the way back to 1985 where it picks up in January and appears to have every month through April 2009.
So why does GISS not have this data? Remembering the name change which happened in Oahu, names again come to mind. I can’t be sure, but it might have something to do with the station name spelling.
- I spell the station name this way: Dutch Harbor
- NCDC MMS spells it this way: Dutch Harbor
- NCDC B91 selector spells it this way: Dutch Harbor
- The station COOP observer spells it on the B91: Dutch Harbor
NASA GISS spells it on the graph header and web page station selector:
Harbour? WUWT? It’s an American port!
I checked this guide for differences in British, Canadian, and American spelling, and sure enough:
Note to Gavin: change the spelling in the GISTEMP database and the station data might automatically kick in on the next data pass.
Glad to help! Got any more lost stations and station data you need found? We’ll look for the last 20+ years of Dutch Harbor data to show up in GISTEMP real soon.
UPDATE:
In comments “timetochooseagain” writes-
I’m reminded of this from John Christy:
“the use of a few popular stations for which the data are easy to find, leads to too much warming when the averages are constructed. I have published research for North Alabama, Central California and in a few months East Africa, in which I went back to the original sources of data to augment the number of stations by roughly a factor of ten – indeed, ten times more stations. This effort requires significant time in searching for and manually digitizing the records for scientific purposes. In each case, I’ve found that the data sets based on a few popular stations overstate the warming by up to a factor of three.”
http://www.nsstc.uah.edu/atmos/christy/ChristyJR_WM_Written_090225.pdf
Update2: I’ve removed a sentence related to Gavin Schmidt and British spelling of the station name, since he’s not at fault, GHCN managed by NCDC is the source of error -Anthony
O/T, but I don’t know what else to do with it.
Anthony,
Chico might want to look into what San Jose is doing. This is proven technology (Germany is using a LOT of this,) and will be a cornerstone in San Jose’s bid for energy independence. Waste to Biomass
REPLY: OK that does it, I’m making a tip line – Anthony
I have no doubt that NASA GISS would delightedly welcome the funding required to chase up every suspicious-looking station, even though this data is produced and owned by another agency. Or maybe they could just accept that any database this size is bound to have some errors and take this into account when doing their analyses, which might just be a better use of resources.
Read Steve McIntyres comments. – Anthony
Well, I did, especially BTW I’ll bet that GISS just uses the GHCN spelling and that GHCN spells it Harbour (which is my natural way of spelling it BTW).
And he is correct, the GHCN Inventory file ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v2/v2.temperature.inv has the ‘Harbour’ spelling, as I surmised. On the GISTEMP Station Data home page we have …
In our analysis, we can only use stations with reasonably long, consistently measured time records. For a list of stations actually used click here, for the full list (copied from GHCN’s website and augmented from SCAR) click here.
So GHCN own the Station names, assuming the difference in Spelling is the cause of the data loss, Gavin’s rational response to a complaint is to contact GCHN and NCDC and ask them to get their act together. The Japanese have a proverb: ‘fix the problem, not the blame’, and the important thing is to restore the data feed and play catchup with the historical data. It is an embarrassing fowlup, but the embarrassment is not primarily NASA’s. The original comment about Gavin was misguided, but not snarky, I concede, but it did kinda open the floodgates in the comments….
REPLY: OK I hear you. The point being made is that GISS needs to do a better job of quality control. They should complain to NCDC when station data disappears and follow up on it. What is mind boggling is that one part of NOAA spells it one way ( in surface reports) and the other (NCDC who manages GHCN) spells it another, and is inconsistent in their own databases. MMS, COOP -vs- GHCN – Anthony
I called up Dutch Harbor on Google Earth, and searched the area for an automated weather station. The only thing I could find that I THINK is a weather site is between two buildings just to the east of the runway. I’ve deduced that this may be a weather facility by studying the shadow cast on the ground to the north-northwest of the actual tower. It’s situated between a hangar and what appears to be a fire station. Coordinates are: 53°53’50.39″N, 166°32’22.19″W.
REPLY: There are two weather stations in the area. 1) an AWOS station at the airport, used only for aviation. It is linked to the airport VHF radio system on 121.5 I believe so pilots can get the report in the air on approach. 2) The COOP station, with Stevenson Screen and mercury max-min thermometers, read daily, and manually. It is the one that climate data is logged from. – Anthony
You’re no longer “praying silently” if you are expressing these sentiments on a blog that gets 1M+ hits a month. 😉
Also, I think that this line of thought is not constructive, and is intellectually lazy. On my blog, I’ve ridiculed those on the alarmist side that are “rooting for global warming” ( http://algorelied.com/?p=2106 ), and I think that “praying silently for an ice age” is just as ridiculous.
Me? I practice the “Que sera, sera” philosophy. I do find the cooling temperatures amusing, but I’m certainly not “praying silently” for them.
* I spell the station name this way: Dutch Harbor
* NCDC MMS spells it this way: Dutch Harbor
* NCDC B91 selector spells it this way: Dutch Harbor
* The station COOP observer spells it on the B91: Dutch Harbor
They all seem to be spelled the same, am I missing something?
REPLY: Yes
http://www.stopepa.com/
I think this is worth commenting
Philip Johns (04:05:36
Philip, I won’t presume that you’ve done any data analysis of substance. If you watch Steve Mc or any of the other folks at CA, the first step is always checking your sources rather than trusting your sources. even if only to spot check or check for obvious mistakes. Much of the study of temperature data is really rather simple math done on a VERY COMPLEX data provenance problem. The math of calculating an area weighed average is mundane. The ARCHEOLOGY of reconstructing which sites reported what data when and where and under what conditions is the HARD problem and one that requires a disciplined attention to detail. Like surveying every site, like looking at all the instrument issues, Like reading B91 forms. Its not for the feint of heart.
Here is a plot of NCDC vs. GISS. I was only able to get NCDC data back to 1922. GISS goes back to 1881.
REPLY: Thanks! What was your source data from NCDC? do you have a link? – Anthony
Link to the data: http://dss.ucar.edu/datasets/ds560.0/
You can get data through 1991 from the state file for Alaska and the rest of the data from the individual yearly time series files.
REPLY: Thanks very much, this is valuable. I never would have thought to look at UCAR as they are not normally a surface data provider. – Anthony
As an Aussie, i consider myself at least tri-lingual. I can understand Australian, British and American English in both written and spoken forms. The wonders of English will never cease and may be summed up by a poem detailing the difficulties of English. The first two lines are:
We’ll begin with a box, and the plural is boxes,
But the plural of ox should be oxen, not oxes.
The full poem can be found here http://www.worldwidewords.org/articles/ar-eng1.htm
Personally, I vote for harbour 😉
One other thing-concerning Crabbers,I know a few and some are still alive some are dead.Lost a family friend last year running the Gold Beach (Rouge River) Bar. Even on the Oregon coast the Old man North Pacific can have his way with you…
I respect these people…
Philip, I won’t presume that you’ve done any data analysis of substance. If you watch Steve Mc or any of the other folks at CA, the first step is always checking your sources rather than trusting your sources. even if only to spot check or check for obvious mistakes. Much of the study of temperature data is really rather simple math done on a VERY COMPLEX data provenance problem.
Quite. As it happens, I was once responsible for the Data Warehouse at a global bank. We took feeds from around 150 subsidiary countries, rather fewer than NASA’s several thousand stations. We worked with all the suppliers to get the data as good as it could be, yet we knew that there were residual defects [I guess the equivalent to this would be a branch that was reported as closed, when in fact it had just changed name. ]. We could have diverted resources into investigating every last suspicious data dropout, only to find that the majority were legitimate, and that the effect of restoring the missing ones on the ‘fitness for purpose’ of the warehouse would be marginal. Such a proposal would not have got past the Board.
Another analogy would be a retailer who ordered several thousand live lobsters, inevitably a proprtion would be DOA. The business that ordered a postmortem into every last dead crustacean would not be in business very long.
Actually, given that a lot of the data is online, a systematic investigation into ‘missing’ NCDC data feeds might make an interesting volunteer project, as long as the expected result was not predetermined, naturally.
But I fear the relevance of Dr Schmidt’s nationality still escapes me, given that
– GISS does not own the Station Inventory File
– The data dropout dates from 11 years before Gavin joined NASA
😉
Philip Johns wrote:
“Another analogy would be a retailer who ordered several thousand live lobsters, inevitably a proprtion would be DOA. The business that ordered a postmortem into every last dead crustacean would not be in business very long.”
No disagreement there.
But in the case of GISS, they don’t even sort out the “DOA lobsters”, they leave them in place with the “live” ones.
So by your analogy, using GISS practices any lobster business that didn’t sort the DOA from the “live” would not be in business very long either.
I just don’t think it is too much to ask to either manually or programatically sort the dead stations from the live ones. There’s just no quality control of this type going on at GISS at all. – Anthony
How would GISS manage with Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch?
John Christy:
“…the use of a few popular stations for which the data are easy to find, leads
to too much warming when the averages are constructed.”
If enough “lost” stations are found, it would be interesting to compare their averages to the non-lost stations to see if there is any statistically significant difference.
I would suggest that someone propose to Gavin that he should determine a list of words spelled differently in the US than Britain and do a global search for them in the GISS database. I would bet he finds at least several dozen stations that have been lost in cyberspace.
There is a Stepheson’s screen type box over at the Mississqoi National Wildlife Refuge in Vermont. I havn’t seen it in the list. Maybe I have missed it, if it is on the list, I can survey it, since I drive by it several times a week. It would be located in Swanton, V
“REPLY: I sent him an email tonight to be sure. I wouldn’t want him to miss this one just in case he doesn’t read WUWT regularly, especially on weekends – Anthony”
— chuckle 🙂
Philip Johns,
As it happens I was once responsible for running a data system at a university that billed 15,000 students multiple times per year. Every set of transactions had to be correct or we heard about it. It was part of the system to do QC and nobody would ever think to accept errors. People weren’t “diverted” to get it right; it was part of their primary job. Should be the same with a dataset so crucial to world economies and policy making. No excuses.
The first rule of gathering and transmitting, and later archiving, weather observations was (and surely still is), if you can’t identify the reporting station plus the date and time of the observation, then you really don’t have a weather report.
Storing such data requires a bit of complex indexing — both on-line and off-line — but if you can’t find it, it may as well not exist.
So why is the storage of weather archives at NCDC so haphazard? These folks have had computers since sometime in the 60’s.
Anthony,
I am wondering if someone evaluated the historical locations in the country side, ie those locations with little urban heat effect, how much change they have seen on average over their life of reporting. Has anyone pulled the data on the stations with the correct level of maintenance and audited their change specifically? Ie I know you are working on a longer term larger project on the topic, my question is, has anyone built a smaller sample of the best stations and evaluated them yet?
If not, could I get a list from you, of the pool of best choices to look at? 100 give or take stations with long histories and a lack of urban heat island effects to choose from would be my choice. I personally would like to know, in the countryside, has global temps changed that much on average.
Jack
As it happens I was once responsible for running a data system at a university that billed 15,000 students multiple times per year. Every set of transactions had to be correct or we heard about it. It was part of the system to do QC and nobody would ever think to accept errors.
Yeah, but I bet you had some sort of control over your inputs, in the case of GISS, it is the NOAA that decide which monthly station data gets released, not NASA.
Oh, and Gavin is not concerned with the day-to-day running of GISTEMP.
http://www.realclimate.org/index.php/archives/2009/06/a-warning-from-copenhagen/#comment-127464
Bye for now.
Good grief Anthony, you eat drink and sleep this. Catching the screen on TV is truly remarkable. I hope your wife supports you 100%. I know how tough it can be if not.
All of these types of foul ups should be reported to Fox News. I’d suggest emailing them to Glen Beck. I think some of them will make his show. The more you can show the ineptitude of the AGW scientists the more you cast doubts on their claims.
Wow, two for two. Maybe it’s time you set up a project to try and find all of the missing stations. Seems like it might be worth it.
Part of the reason one may have to pay to get weather data (from the US, Australia, UK, etc) is because in the ardor to make the government more business-like, rules were passed that required these agencies to recover the costs of providing these data. In short, since business gives away nothing for free, the government shouldn’t either. Thus folks get charged for data.
If you don’t like that, then urge your representatives to change the rules regarding costing of data.
REPLY: Explained that way it makes sense. Thanks, Anthony