The NOAA website http://mi3.ncdc.noaa.gov/mi3qry/search.cfm? has a Map tab linking to local maps and Google Earth. I don’t recall seeing it before. Maybe someone can comment on this and also check the accuracy of the maps for stations shown there.
I was looking to see if their equipment reporting included whether stations had aspirated or unaspirated housings. In the old CDIAC forms, the form of housing was recorded and perhaps it is here as well (if so, I’d appreciate info on how to find it if anyone knows.)
Climate Audit 
33 Comments
I’ve looked at those maps quite a bit. They give a generalized location, but not a fully accurate one.
The lat/lon used is rounded in some cases. Also the red and blue markers indicate station moves but its often hard to tell which one is the current location. And sometimes the station has moved but that lat/lon has not been updated.
The only reliable way I have found of locating stations is to find a physical address. I do this with the “identity” and “location” tabs, then do some Google text searching to find the street address or a phone number to contact the observer.
I have also used these maps and found that they accurately place the markers where the coordinates say that they should be, but they don’t always place the markers where the stations actually are. Most of the maps will show several markers for each station with the blue marker being the most recent location and the orange markers showing previous locations. Twice, I found the actual station at a location shown as a previous site for the station. This happened in both Odessa, WA and Davenport, WA.
I documented the Odessa station but I have not put up the photos of the Davenport station because I have not yet contacted the individual listed as the caretaker. The Davenport station is in such poor condition that I’m not sure that it wasn’t abandoned.
In the case of the Odessa station the caretaker told me that the station had never been moved since it was first placed on his property yet the NOAA metadata website said that it had been moved from that location several years ago. The supposed new location was in an open field but I walked through that area for nearly an hour and found nothing.
In general, the most recent locations (since mid 1990s) are recorded at degrees/minutes/seconds resolution; earlier locations are only degrees/minutes. If you click through to the location detail screen for a specific location, it indicates DDMMSS or DDMM, as well as the actual values entered. The lat/lon displayed elsewhere seems to be represented as a floating point number, which as Anthony points out has rounding issues and minor display bugs such as displaying 60′ instead of incrementing the degrees value.
There are some more major bugs in the MMS display, chiefly the duplication of items such as relocation distances and location and topographical descriptions, and their association with the wrong date rows, in e.g. the location “grid” display. I have been in contact with the MMS system’s software team leader regarding this, and he has confirmed the problem, but reports it may take some months to implement fixes; this is understandable, as they have their plates full, and designing universal web-based access to a collection of essentially legacy data/systems is not easy.
Regarding aspirated housings, I only remember one aspirated housing/sensor from Anthony’s web site, but don’t remember where. If you find it, you could look up the Equipment listing for that station in the MMS and see what it says.
Do you know if there is any research being done in the UK on the matters you raise here? I read a report of your work in the Timesonline and then found this site. Congratulations on your work. Is there anything I can do to help with finding information in the UK?
If ASOS, or AWOS, is mentioned, it’s aspirated. Otherwise,
it usually would not be, U of A Tucson being an exception.
Kristen Byrnes found one attached to an air conditioned shack – in Maine, I believe.
Derek regarding the UK there some starter links from http://www.anenglishmanscastle.com/archives/004438.html where we are looking at the Central England Temperature CET record – the worlds longest series, except they have changed the stations a few times….
Any help most welcome.
Tim
Re 4 it was Redding
http://gallery.surfacestations.org/main.php?g2_itemId=1311
But this is not the sensor in question. The HO-83 is a hygrothermometer see these references:
http://www.tabacofamily.com/jtabaco/ispwx/swo/chapter_17Sect9.htm
http://www.crh.noaa.gov/crh/?n=arp17-05a
and this one in particular:
http://www.crh.noaa.gov/crh/?n=arp17-05b
which shows a study on the errors, apparently the error of the HO-83 is well known to the NWS.
Now the question is, how much of it’s errors went into the climate record?
to the The Englishman
England has a great gift, with the worlds longest running temperature series; however I would like to point out that any country in the world that suffers from Urban Heat Island, is would be England. We’ve urbanized more than any other country; which means we should expect the UK to show a larger trend.
Which has been proven; however, the Met-Office doesnt properly account for this. We desperately needed to understand and quantify that effect of urban head before we can predict climate change. And Climate change will happen with or without mankind interference.
PS i’m from south of the river; ie greenwich
RE 9…
Somebdy had to buy the sensors. A PO exists in the system.
RE: #10 – South of a line that runs roughly from Birmingham to Boston, and east of Cornwall, the UK is essentially one big exurb, with proper suburbs and cities popping up here and there. North of that line, to about a Liverpool – Newcastle line, it’s on par with much of the Eastern US. Interestingly, north of there (which, I point out, is a line not all that far south from Hadrian’s Wall – LOL!) it’s obviously much less dense – with only 10% of the entire UK population found in Scotland and some minor fraction in the Borders.
Steve,
The maps are usually off because the NWS marks the location on a topo then the NCDC approximate back at the office based on what is marked on the topo.
Re: #13
Actually, the most recent coordinates (at least for Coop stations) tend to be GPS-generated, as often noted on the Location detail screen (e.g. “lowrance GPS”). I’m not sure why, but Google Earth seems to be a bit off in plotting coords on the aerial/satellite imagery. When I set the datum to WGS84 (which is what my GPS uses, and what Google Earth uses) in http://www.topozone.com, the locations there were spot-on, while in Google Earth, they were off by 50-100 feet or more.
While auditing a site for Anthony recently, my wife and I had an interesting experience that directly relates to this very discussion.
We were traveling through Green River, Utah and used the Google .kml file available on Anthonys site to run down and locate the Green River USHCN station. As it turns out the coordinates led us to the corner of the local golf course and sure enough there was a weather station tucked in behind the grounds keepers metal buildings but it was all non-compliant equipment! I was excited to do the audit and as we talked to the people at the club house to get permission, it became apparent that the station was not a USHCN station and was used only by the grounds keeper for determining how much irrigation will be required and so on.
The young lady at the club house then informed us that there was a gentleman in town that did take temperature readings several times a day and called his reading into Salt Lake. She gave us directions that led us across the highway and down a couple blocks. From the street we could see the Stevenson Screen and rain gauges sitting in the back yard. Unfortunately no one was home at the time so we were unable to get permission to do the audit at that time. Our trip however did take us back through Green Rive on the way home and while the curator himself was not home we were able to speak with his wife and get permission to do the audit.
She told us that the readings at the site were take 11 times a day beginning at 4:30 AM! And that her husband had been doing it since 1986. She said fortunately he was the kind of guy that can get up and go right back to sleep.
When we returned home I checked Google and found that the reported coordinates listed the golf course site within about 100 to 200 feet from where the golf course station was but what we believe the USHCN station to be was around a half mile away.
This particular site did have an aspirated Stevenson screen which threw me for a little while, being a rank amateur and all. I did note that the electric running to the fan was unplugged though.
I was intending to talk to Anthony about the discrepancy in coordinates but time and internet access was somewhat limited in Green River. I hope to have the site audit posted up sometime this weekend (black rubber mats under the station and all!).
I found an interesting error which I commented on surfacestations.org:
Chris Dunn posted a photo taken recently (summer, 2007) of the Greenville-Spartanburg SC ASOS station at the airport. I noticed the photo he posted does not look at all like the ASOS station photos I pulled off the NOAA website for the G-S station at this ID. The NOAA ASOS photos were accompanied by notes taken by the original surveyor (posted in this album), and these notes include the coordinates of the site that was photographed. I punched the coordinates into Google Earth and one can clearly see an ASOS station on the north side. The one Chris found is on the south side and is also visible in Google Earth. The two stations are about 8200 feet apart.
The photos I found are clearly the one’s NOAA believes are from the station in question. However, MMS says the station is operated by the National Weather Service, and Chris’s ASOS is very close to the NWS building, and very close to the “official” GPS coordinates. Thus, an error appears to exist, but it is not clear where.
NOAA indicates the Greer station is “0.00” miles from the G-S station, and that it is a NEXRAD operated by the NWS. This would seem to indicate that Chris found the correct station and that the photos of the ASOS on the NOAA website are of the wrong station.
15, Did I hear that right? The fan was unplugged??? There’s no way in hades that can be producing meaningful numbers without the fan running.
#17, I wouldn’t think that even with the fan plugged in, the black rubber mats laid out under the station wouldn’t allow for very meaningful numbers either.
I have targeted 10 stations so far, and found 7, of which 6 were close to where the coordinates said. It has been hot here in the south, and I haven’t spent more than 2 hours searching for the stations with wrong coordinates. Anthony put me on to address checking and trying to contact the local responsible, but too late for the missing stations. After spending most of a day digging through NCDC data, checking google maps, recalculating coordinates etc., I’m pretty sure I could now find the 3 lost stations. However one of them had as the only useful reference “1.5 mi NW of PO”. Another is 5 miles from the coordinates which place it on a tongue of cultivated field surrounded by woods. The site is noted in MMS as at the water plant, and googling gives 4 water plants. About 4 moves down in the history there is an address, and Voila!, the WWTP and station.
In addition to wrong coordinates, there are wrong street names in Google maps. It’s kind of like a treasure hunt. I have never found anyone who has any knowledge of station history, but people have been friendly and helpful everywhere. Digging through MMS it is possible to reconstruct quite a bit of atation history, but double checking is required. eg One move was about 1 mi. north by coordinates, but “2.5 mi E” in the history notes. The coordinates were clearly more right in this case. The quality of information available is pitiful, but if volunteers look at the task as a challenge/adventure it is kind of fun. Murray
Errors when you use googlemaps/googlearth are normal. It is all to do with the projection system used to put spherical geographic coordinates onto a flat sheet (map). Try putting the coordinates for the Royal Observatory at Greenwich into googlearth (N 51° 28.677 W 000° 00.000) and you will see the effect nicely (the coordinates are over 100m “off” when plotted on the satellite photo). Depending on where on the earth you are plotting, errors of several hundred metres are possible. Googlemaps/googlearth is not a proper GIS (geographical information system), but is great despite these errors!
Re 20 Where I have been looking (30N, 81W) when the google placement is “right” it is within less than 100 yards of the site, and 3 times within 200 feet. Incorrect coordinates are a much bigger problem 0.2 mi to 5 mi. Murray
Re 20 Where I have been looking (30N, -82W) when the google placement is “right” it is within less than 100 yards of the site, and 3 times within 200 feet. Incorrect coordinates are a much bigger problem 0.2 mi to 5 mi. Murray
Re #21
It all depends on where you are, you could get lucky if the center of the projection zone happens to be where you are looking, distortion increases to the edges of the projection zone. Agreed that rounding errors can be equally large. The way to check is to take your hand held gps with you on the audit! I doubt that the owners of station data are being deliberately obscuring, just lazy probably.
I’ve had similar experiences with the 5 I’ve surveyed. Accuracy of the MMS coords varied from “spot on” to “pretty good” for the current locations, but even a 100′ can make a lot of difference for microsite issues. One station’s given coords were correct for the precip instruments, but I found that the MMTS was far away from them and right up against the house. I think this shows the value of the ground truthing being done with both photos and GPS.
I do NOT trust the coords for the previous locations, especially the ones from pre-GPS days. Some of don’t seem to make much sense as a location when you view them in satellite view. I think JS at #13 is right about those, and it’s difficult to know how accurate they are, except maybe for the few that include a street address in the location metadata.
I’ve had good luck with Google Earth using my own handheld’s coordinates. When I enter the numbers from my GPS into GE, the landmarks around the station show up right where they should.
Since this thread is attracting some surfacestations folks, I thought I’d post a few techniques I’ve found useful:
1) Since Google Earth’s photo coverage is often lower-resolution outside cities, try an alternate like maps.live.com (Microsoft’s Virtual Earth). I’ve found it to have good high-resolution photo coverage even in rural areas (at least for WA state). Use the “3D” mode for maximum zoom — the “2D” mode, though seemingly using the same photos, won’t let you zoom in quite so far.
2) http://www.topozone.com allows you to access USGS topographical maps for free. Although you can pay for an enhanced service, I’ve found the free service sufficient so far. They also sell access to USGS aerial photos; for our purposes, you can often get the info you need just by looking at the free preview/sample photo for an area.
3) As mentioned by Murray, the local PO is often used as a reference point. Both Google Earth and Microsoft Virtual Earth will plot the locations of post offices for you – just do a search for “post office” as the business type.
Re #17 and #18,
The Stevenson screen was designed not to use a fan.
Like others, I’ve found the listed locations and maps in the Map tab to not always correspond. I’ve tried to document inconsistencies in the survey reports. Furthermore, the locations found in the NCDC Radar Data Inventories ( http://www.ncdc.noaa.gov/oa/climate/stationlocator.html ) don’t malways match all that well against the MMS, either. I suspect all these issues documented in this thread are symptomatic of the larger QC problem that surfacestations.org has been documenting at the site level. I also have to wonder what coordinates are used for determining lights vs no lights.
The MMS data agrees with the Master Station List on their website. I built the kml file for Anthony that’s on the surfacestations website. As a check I plotted the locations for Petaluma of the MMS value, GPS from the survey form and the site location from the survey that is marked on the image and they all disagree. I haven’t tried Virtual Earth to see if the registration is any better. The problem could be map registration or error in the GPS reciever, probably some of both.
#28 Barry, please dont misunderstand my concern. My concern isnt with the .kml file. I was happy to use it and it was a wonderful guide pointing out the stations that we would be able to visit along our route.
My concern was that we found a weather station within what you would expect of the coordinates given. The only problem, was that the weather station we found had nothing to do with the USHCN network and the actual station was at least a half mile away as the crow flies. We were more concerned that maybe the official site was retired and nobody bothered to tell the curator of the site that data was now being acquired from another site with non-compliant equipment. I think its just another example of just how much detailed information NOAA has on its network sites.
Our experience with your .kml was nothing but positive for all the other sites we visited. At most I think we found the sites within an easy city block of what was listed in the file. We found sites that we couldnt audit because curators werent home, or they were behind locked gates and so on, but we found all the sites we set out to find! The file is definitely a helpful and worthwhile effort on you part. Good job and thank you!
Re #29
Thank you. No, I didn’t take your previous entry in a negative sense. I’m happy to see that someone has found the kml file useful. I just wanted to post an FYI so that people where aware that the locations NOAA has on the MMS are consistent with their other documentation (but not necessarily correct), but even with the GPS locations Goolgle Earth may not show the site at the correct spot in the image.
BarryW, the KML file is wonderful. It’s been very useful for helping to see at a glance which stations are near our travel routes, and how far away they are. Thanks very much for doing it.
rec31.
All you need is a surfacestation triptik.
You enter your starting point and destination and a route and directins are
calculated that allow you to visit every station within a user selected distance.
Surely, in a scientific data-collection project of this magnitude (not to say importance), a decades-long pattern of non-uniform apparatus, poorly sited, subject to innumerable reporting discrepancies which authorities refuse to admit or reconcile, cannot be taken seriously on any level.
snip