How not to measure temperature, part 16

This photo comes come to me from NOAA’s Weather Service Forecast Office in Monterey.

This is the official USHCN climate station of record for Livermore, CA. USHCN # 44997 The temperature sensor is located in a backyard of a residence within six feet of the swimming pool.


Here is the temperature trend from NASA GISS:


The question is: can an unbiased and accurate reading of temperature be obtained in somebody’s backyard next to their pool? With NOAA siting requirements saying a minimum of 100 feet from buildings, I would assume this would apply to pools too.

I couldn’t make this up if I tried.

You can see the picture without the annotations on the NWS website with this direct link:

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
June 30, 2007 8:27 pm

OK, so it is near a pool. What is the big deal? Also, it has been recording temperature since 1880. What period of warming does the current location invalidate?

Anthony Watts
June 30, 2007 10:23 pm

The pool acts like a giant heat sink, dampening and regulating temperature variations nearby, especially on calm days and nights.
For nights that have air temperature are cooler than the pool water, the pool will radiate heat near the sensor. The humidty from the pool, also makes overnight lows warmer.
If the pool is heated, the problem is even greater.
As for invalidating warming, periods, it depends, the sensor has been moved several times during the last 100+ years, and used different instrumentation.
The point is, the measurement environment is well outside of NOAA published specifications. You can read them here:

July 1, 2007 3:23 pm

“Please, can somebody tell me how a scientifically accurate reading of climate, especially temperature trends, can be obtained in somebody’s backyard next to their pool?
Ok, I’m game. I’ll try. I can think of two instances where it would be valid. The first is if the entire valley was populated with swimming pools. (What is the population there anyway?) In this instance, the temperature would be fairly representative. The second case would be if it were part of a larger network, say 1000 stations, that uniformly sampled the terrain.
There are three issues that come to my mind immediately that you are addressing with the site surveys. These are the primary issues that the community seems to be struggling with. The first is the low density of weather measurements.
The second landscape changes over the past 100-150 years. A long record with thin sampling would be best with little landscape changes over the time. Clearly this is not true in this picture nor in many other locations–even with the best datasets, because the landscape has changed. I purposely use term landscape instead of the current term in favor, “land use”, because I find the latter term too anthropomorphic. Landscape changes would cause temperature drifts with or without the presence of humans.
The third issue I can think of off the top of my head is the attribution of the observed surface warming to carbon dioxide. Whilst watching this station situation unfold, it seems clear the a case can be made that the landscape correction done on the historical data may need to be more thoroughly reexamined.
Keep up the good work Anthony.

July 1, 2007 7:18 pm

Why is the temperature trend roughly linear? I don’t see a step function when the sensor was moved/updated, nor do I see a step function when the pool was built. Do these things have the effects you claim the do? I would think they do. I agree with you. But why can’t you see it in the data?

John G
July 1, 2007 7:42 pm

#1: The pool and fence have not been there since 1880.

July 2, 2007 5:37 am

#5: Exactly. So there is a long warming trend of ‘valid’ data. When was the pool built? 1980? That means there is a full century of warming in the record. And why isn’t there a noticeable change when it was built?

July 2, 2007 9:42 am

The issue with the pool is that it makes the overnight lows higher due to humidity (From John Christy’s research) and also makes the daytime highs higher due to the concrete surface surrounding the sensor.
This would have the effect of biasing the trend upwards. The overall upwards trend reflects the UHI of the SFO Bay area, which has undergone steady growth for 100 years.
Unfortunately, there are no USHCN or GISS stations west of Livermore that have not been surrounded by Bay Area growth, but east of Livermore about 100 miles is Electra powerhouse, up in the foothills which has a sensor location unchanged and has not been encroached upon. It shows a downward trend.

Paul G
July 2, 2007 11:54 am

== Post # 4 by James ==
=”Why is the temperature trend roughly linear?”=
Going by the graph alone, without access to the raw data and site history, means one can only guess at it’s linear trend.
=”I don’t see a step function when the sensor was moved/updated, nor do I see a step function when the pool was built. Do these things have the effects you claim the do?”=
When was the sensor moved? How many times? When was the pool built? When was the dark fence added?
=”But why can’t you see it in the data?”=
“Data” may be a loose term in this case. All we have is a graph. Has it been “adjusted” with software or other methods to smooth it out?
A photo like this raises a hundred questions. Questions which professional climate experts seem remarkably incurious about looking at.

July 2, 2007 3:02 pm

I have to agree with James. The graph clearly shows the temperature increasing at a fairly uniform rate, regardless of whether the censor was moved or a pool was placed adjacent to it. While the placement may not meet the standards laid out by NOAA we clearly see a warming trend has been present for 100+ years.
Paul G: Do you know for a fact that one or more professional climate scientists are not looking at this issue (sensor placement)? Your statement seems rather overreaching to say the least.

No Junk Science Please
July 2, 2007 4:13 pm

I live in Florida, and I have a pool.
During the summer months, my pool area is hot during the day. I don’t need an expensive thermometer or government-issued device to tell me that both the pool water and the surrounding concrete DO radiate heat at night.
All I need to do is walk barefoot on the concrete and witness the steam rising off the pool (which coincides with the steam rising off the lake behind that).
Neither my pool nor the lake are heated by any direct manmade mechanisms. No solar panels, no natgas device, no heat-exchanger.
To those of you looking at the graph and observing that there still appears to be a warming trend of 100+ years, you should consider this possibility:
Maybe now in recent sampling periods, there is actually a COOLING trend, but maybe it’s being masked by the presence of the (relatively new) pool, concrete, and fence.
How can we say that this has NOT happened?
If anything, we should note that the CONDITIONS have changed, so we cannot say for certain what is true and what is not…
My point: You cannot change the conditions under which the test is being done without chancing that you are also inadvertently altering the results. Just because the results “seem to follow a linear line” does not necessarily mean that they are representative of “what is.”
Ignoring this possibility is like saying that “the ends justify the means.”

July 2, 2007 10:24 pm

James- As Paul G pointed out, you are not looking at raw data. The raw data is not readily available. I suggest contacting NOAA or NASA GISS if you wish to access the raw, un-adjusted dataset. Also, you can ask them for all of the metadata for this site. That way, you can start to figure out how many changes happened and when, like when the lead paint was overcoated with latex paint, when the new MMS sensor replaced the old screen, when the sensor was placed next to the pool, what changes in fence color occurred when, etc. Good luck.
Anthony, does the forecast office have other photos available for your project?

July 3, 2007 9:39 am

Re: No Junk Science,
What you say is technically true and possible, but not necessarily logical. What do you think is more likely? That the past 100 years sans pool continue, and there is a slight contribution from the new setup, or that there really is a cooling trend that is getting masked? Just because the data has been tainted does not mean it is likely that the Science is Junk.
Also, if you read the latest RealClimate post, you will learn that NASA GISS is a user of this data, not a creator. They are making do with the best that they have available. They are trying to understand one of the most complex systems around, using what tools they have. This is not Junk Science, this is science. It is never perfect, even in sterile labs.

July 3, 2007 11:33 am

James- If as you say there are problems with the data, then the error bars claimed by the IPCC should be adjusted to reflect the problems. Do you really believe the IPCC or USHCN error bars of +/-0.05 degrees C after glancing through Anthony’s collection of photos? Let’s amend that to +/- 2 degrees C.
RealClimate? What’s a RealClimate?
This whole surface temperature measurement effort looks a lot like a circular argument. Adjust the GCM output to replicate recent past temperature trends, then go back and correct the surface temperature records so that the ‘measured’ data looks consistent with expectations. Then repeat every four or five years, and eventually a perfect convergence between models and ‘measurements’ occurs.
I still think it is possible to find 20-30 excellent stations distributed throughout the US (excellent being completely rural, with no land-use changes, using well-maintained equipment, reliable observers and accurate metadata of site changes, with no or small adjustments to the record) that will represent the surface temperature in the US over the past 100 years. I applaud Anthony for making the effort to go out and find those excellent stations.

July 3, 2007 5:36 pm

You can find those stations with the help of this tool:

July 4, 2007 7:04 am

James- Nice website, but no, that is not raw data from each site. It has already been adjusted with USHCN corrections. Plus, there is no metadata available. In other words, all of the changes, missing data and offsets have already been removed.

No Junk Science Please
July 4, 2007 1:23 pm

Hello, James. I don’t mean to “pile on” with the others here, but I really must respond to your first paragraph, just to set the record straight.
You said, “What you say is technically true and possible, but not necessarily logical. What do you think is more likely?”
I would respond that what I “think” is irrelevant. What the scientists “think” is irrelevant. And what anybody “thinks” is irrelevant.
We are sooo often wrong in our “thinking,” with sometimes disastrous results. That is EXACTLY the reason that we perform scientific experiments, and that is exactly the reason that we do those experiments with control studies; so that we can compare “A” with “B” and measure the difference.
If what we are doing here is collecting and analyzing data, with the possibility that we might one day use that analysis to establish environmental policy, tax policy, and laws to support them, then we should make EVERY POSSIBLE effort to ensure that the data is faithful and true.
You also said, “Just because the data has been tainted does not mean it is likely that the Science is Junk.”
I am saying precisely that untrue input will result in untrue output, which will in turn result in incorrect analysis. Making decisions on such analysis can have catastrophic consequences.
So yes, it IS bad science!

%d bloggers like this: