How not to measure Temperature, part 26 – counting A/C units

There’s been some recent discussion about how only rural stations have been used in the NASA GISS analysis, and those rural stations are qualified by looking at night time DOD satellite photos, and doing a count of visible streetlights within a radius to quantify UHI potential or lack thereof. The “best” stations are labeled “lights=0”

One of those stations is Happy Camp, California, population 2182, an old gold mining and logging town located in the rugged NW corner of the state, and about 100+ miles from any major city. NOAA MMS metadata website reports data back to 1931 with 3 small distance station moves, and no changes to equipment. NASA GISS reports data back to 1914.

It looked like a good candidate to look at for a lights=0 survey. The weather station is located at the Ranger Station:

Happy Camp Ranger Station - USHCN climate station of record

But what you can get from satellite images and databases can’t really prepare you for what you may find. I “expected” to find an old classic Stevenson Screen, probably near the Ranger Station office. Check on that. But what I didn’t expect to find was a “rural” station swimming in a sea of exhaust from 22 air conditioning units within 100 feet of the Stevenson Screen. Ridiculous, you are making this up you say? Well that would be my first reaction too.

But here they are, count them, I’ve labeled the A/C units for your convenience:

Happy Camp Ranger Station looking West from Stevenson Screen

Happy Camp Ranger Station looking West from Stevenson Screen


Happy Camp Ranger Station, looking NE, Stevenson Screen visible


Happy Camp Ranger Station, looking North towards courtyard

Happy Camp Ranger Station, looking southwest inside courtyard

Happy Camp Ranger Station, looking southwest inside courtyard

To help you get bearings on this walking tour, an aerial photo is available here

And the complete collection of photos is available on

In addition to the 22 A/C units within 100 feet there are other biases too. Granted, not all 22 may be introducing a bias, but since NASA’s Dr. James Hansen counts lights near stations, to asess UHI magnitude, we can count A/C’s. If each A/C unit was 2000 BTU, that would be 22×2000=44,000 BTU of waste heat dumped within 100 feet of the Stevenson Screen where the thermometer is located.

Additionally. for other biases, positive and negative there’s the buildings, the windows, the shade trees, the wind sheltering, and the lawn sprinkler. There’s also the big parking lot to the southwest, and the Stevenson Screen is at the top of a slope and there’s a parking lot downslope.

When I mentioned to the site curator about the A/C units she said “hmm, I never thought about that” but then added, “But I can tell you that when we water the lawn, my high temps are lower”. I asked the curator what the prevailing wind direction was, and she said from the “south to southwest usually”.

Now there doesn’t appear to be much of a trend according to the NASA GISS plot, but there are some large amplitude swings and discontinuities:


Unedited NASA GISS raw data plot for Happy Camp RS

So one has to wonder, with all the observed microsite biases, what is the data really showing? One also wonders what the plot might look like if this station was better sited.

And if a lights = 0 station like this one, far removed from urbanization, has so many such micro-site biases, could others have similar problems? It looks like more hands-on site surveys will have to be done to determine the true value of lights=0 USHCN sites.

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
August 2, 2007 5:51 am

Please stop posting duplicate content!
Note that the data is not available after 2002, and the annual is gone after 1995, so this station isn’t being used by any of the modern climate estimates.
*** NOTE fram Anthony: Ken I don’t tell you how to run your website, site, so please stop telling me how to run mine.
There are two different audiences that read this blog and CA.

August 2, 2007 7:17 am

I saw this on an MSNBC website and am shocked as to what we consider good data. Thanks so much for the information.

August 2, 2007 9:18 am

The existance of any, let alone many, grossly contaminated sites demonstrates that the QC procedures are at best inadequate and at worst biased.

August 2, 2007 10:53 am

I just reviewed the co2science historical data:
noticing a major jump in recorded temps at 1977, was that the time when the AC units were installed?

August 2, 2007 1:15 pm

Funny, this is still being used by Hansen at NASA as a lights = 0 site. I bet next year he will ignore the photographic evidence and still include it in his GISTEMP.

August 2, 2007 3:39 pm

“RE27 well the point is, We’re not out to measure mixing rates. We’re measuring site compliance. This site is clearly out of compliance by published and known NOAA standards. Thats what this project is about, to get a census of compliance. Those that keep trying to tell me that I need to do a bunch of other things at the same time should just either mind their own beeswax, or go do those studies themselves. Its a volunteer effort, done on my own time and systems and the generous time of others so funds for such suggested research diversions aren’t readily available.
We have one goal, completing a USHCN site compliance census.
Other studies will follow I’m sure….”
I just have a question about the future. It seems clear if you are just looking at site compliance.
however, if you do other analyses, you may get into “post-hoc” problems. Would it be a benefit to identify now the specific questions you might ask if you get the full data-set ? That way you would avoid accusations of data-dredging.

George M
August 2, 2007 5:22 pm

One of the photos shows the bottom of a radio tower, which may be a met. tower or may support the antenna for the Rangers’ two-way radio. In any event, another possible clutter factor is Radio Frequency Interference (RFI). If these sites are like many instrumentation sites I have seen, transmitting from a nearby radio transmitter could easily overload sensitive electronics and produce wild swings in the data. MIL-STD 462, I think, addresses RFI susceptibility, and AFIK, none of the weather measuring equipment is tested to that type of standard.
Might be very interesting to watch the indicators and have a ranger stand next to the shelter with a radio and transmit.

Max L
August 3, 2007 9:01 am

Notice in the first picture the “Incident Base” sign attached to the Ranger District sign. I suggest that not only was there a fire presently burning but the Ranger Station was used as a ICP in numerous past fires. This means that many fire engines historically have stopped at the ranger station for many different reasons ie: to check in/check out, use the flush rest rooms before going to base camp, record travel time, use the Forest Service email system/phone system, get ice/water/coffee etc. The whole time their large diesel engines on idle in the parking lot. The heat and exhaust from these engines would most assuradly skew the readings.

Steve Moore
August 3, 2007 9:08 am

George M:
Good Point! And one that was obvious to me once you made it. I spent 32 years in a pulp mill, and one of our constant annoyances was sporadic ClO2 alarms. Figured out that if someone simply keyed the mike of his radio within 8 feet of a sensor, it would go off.

Allen jennings
August 5, 2007 11:49 am

Is there a study done by a statistics expert (mathematician)on the global “warming” trend simply focusing on the math and the statistical relationship and if weather reporting stations are within a range of statistical significance?

Frank Ch. Eigler
August 12, 2007 11:34 am

Can you help focus the auditing on those ~210 “rural” stations?

%d bloggers like this: