Fun with Thermometers

A guest post by David Smith

Recently I completed my tenth survey for These surveys are fun, almost like treasure hunts where the clues are good but not always great, thus requiring some ingenuity. Also, the surveyor gets to see areas which may otherwise never be visited. And, they’re for a good cause.

While I found no “poster-child” poor quality sites I did observe an array of siting problems. Some thermometers were near the drip-lines of trees, some next to buildings, one was near a concrete patio, one at a sewage plant, several sat above poorly-drained soil and so forth.

These conditions are less than ideal, obviously. Perhaps more importantly, these conditions can change over time. Trees and shrubs grow and die, ground cover changes, concrete is added (and tends to darken over time), drainage may improve or deteriorate, fences and other construction are added or removed, and so forth. Each of these can subtly change the local temperature, a situation which is especially important if one is looking for changes of a fraction of a degree.

To what extent do these imperfections affect local temperature?  Well, we really don’t know (or if anyone knows they’re not talking!).

So, to make a small and imperfect step in that direction, I’m running a few local experiments. My goal is to examine, at least qualitatively, how local microclimate factors like trees and concrete affect temperature. As you’ll see, my methods are too crude to allow fractions of a degree determinations but I should be able to quantify the magnitudes of the impacts of trees, concrete, etc. Or at least that is my goal.

First, my instruments:

I’m using several temperature detector/recorders (”USB1″) like the gray object shown in the photo. These electronic devices measure and log the temperature to the nearest degree F and allow sampling on various schedules. I use 30-minute sampling.

Note: Interested readers can buy these at:

At this point I’m testing the hardware and developing my experimental plan. But, I have made a few (literally) backyard tests and I’d like to share one of those. This is to help illustrate the approach and, I hope, stimulate helpful comments from other readers.

This initial run (sort of a beta test) was made in my backyard. It involved two extremes. One is near my garage, above a dark-soil flower bed and landscape bricks. This is near a wooden deck and walkway gravel. This spot gets direct sunlight about 50% of the day.

The second extreme is deep shade, beneath low-tree (crepe myrtle) cover and above thick, semi-tropical shrubbery.This is about twenty feet from sunlight. A photo of the backyard is below, with red boxes marking the two locations:

I also use the temperature readings from an airport/airbase located four miles west of my house. This airport provides professional-grade open-field temperature readings which should reasonably approximate regional ambient conditions.

A representative backyard temperature time series is below:

This shows pretty good agreement between the deep-shade max/min and the local airport open-field max/min, which frankly surprised me. I’d expected the deep-shade readings to show less variability (lower highs and higher lows).

More importantly is the contrast between #1 (sunlight and plant beds) and #2 (deep shade). The #1 spot stayed 5 to 10F hotter at midday than #2 (deep shade) less than 50 feet away (and, as a matter of fact, #1 was 5 to 10 F warmer than the high-quality nearby airport).

Why does this matter? well, suppose a co-op station had slowly drifted, over several decades, from open-field conditions to those found at site #1. What would that do to the apparent trend?  That’s an important question which is at the heart of the surfacestation effort.

This backyard demonstration involved convoluted conditions. There is little chance to untangle the relative contributions of so many variables (bricks, soil, tomato plants, trees, etc). So, my plan is to reduce the number of variables in the tests such that we might be able to make broad conclusions about the relative impacts of trees, concrete, drainage and other factors which may change over time.

This should be fun! Suggestions welcome.

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
Pamela Gray
April 27, 2008 6:39 am

You would need to somehow isolate any temperature problems with the dark notebook computer. They heat up as they are being used. They also may heat faster than the surrounding environment and cool slower when you move into the shade. Is the sensor in the plugin or is it remote?

steven mosher
April 27, 2008 6:40 am

buy geigers book climate near the ground. that should give you some
good ideas

Pamela Gray
April 27, 2008 6:51 am

How about using your IR camera to test the relative heat of the notebook in sun and in shade?

April 27, 2008 7:04 am

Good idea, would be interested in doing some observations down here in Florida. Perhaps you could come up with a generic list of tests that people could do. Being a gadget nut I will definitely invest in one of those USB data loggers,. Glad you made it home safe and the trip was both informative and productive.

April 27, 2008 7:21 am

Nicely done. Remember Anthony’s effort started with a study of the changes in different paint on Stevenson cabinets.
CoRev, editor

April 27, 2008 7:50 am

I am so completely not a climate scientist so the following may be full of holes. Anyway…
I’m guessing one important difference might be the wind. Your backyard looks pretty sheltered. If there’s any breeze at all local heat source effects will be averaged around.
Or perhaps think of it this way:
On a still day your sensors and the airport sensors are both measuring microclimates. On a very windy day your sensors and the airport sensors might both be measuring macroclimates. Somewhere in between your backyard sensors will be measuring microclimates whereas the airport will be measuring a macroclimate.

Carl Smith
April 27, 2008 8:22 am

David, it may be a good idea to record humidity with these tests, if possible, as varying humidity could influence the results and how you interpret them.

Mike Bryant
April 27, 2008 8:22 am

Glad you didn’t put a humorous illustration on this subject…

Retired Engineer
April 27, 2008 8:51 am

Suggestions: Try putting all loggers in the same place and see how well they track each other. I’ve used this guy for a while, with what look like reasonable results, but no way to calibrate it. I think it’s good for measuring variations, absolute is unknown. I’m a bit surprised that it tracks the local weather station. A lot can happen in 4 miles. Another unknown: does it drift with time ? To measure effects of changing environment, tree leaves, grass, etc., it should be repeatable. Another unknown. If it continues to track the weather station, a good sign.
All that said, a great beginning. Keep us posted.
Tim Conrad
35 years of measuring stuff and catching h— if the numbers were wrong.

Eric Gamberg
April 27, 2008 8:56 am

Instrument radiation shelters (beyond the plastic housing)?

April 27, 2008 9:17 am

very good work, we need more of these experiments!

April 27, 2008 9:35 am

You are doing a commendable task in seeking reliability in our nations climate data base. You mentioned measuring to tenth of a degree? I seriously doubt you can with the equipment you are working with.
My concern is calibration. My personal background is in metrology (often confused with the other ‘ology’ here) for more than twenty years covering solid, mechanical, and electronics at levels from bench tech to supervisor.
I have seen few reference to design accuracy/tolerances or calibration reports, data, corrections, schedules, traceability and correlation between stations.
Having calibrated thousands of ‘new’ electronic instruments, I can remember only a few that came out of the box and met the manufacturers stated accuracy.
Liquid-in-glass thermometers are the most reliable with the largest error is the human reader- recorder, and the remote electronic are the least with too many variable loops in the data stream.
Unfortunately assurances are costly but errors should be known and stated or they will be exploited by those who need to support some issue other than the science.

David Smith
April 27, 2008 11:58 am

Thanks for all the feedback, it’s much appreciated.
* the temperature sensor/recorder isn’t connected to the computer other than for data download.
* I have access to very nice IR cameras and will be using those on occasion
* my conjecture is that wind (mixing) plays a major role in microclimate. I’ll be logging nearby (airport) wind conditions and perhaps that will be useful when examining the data.
* Carl, will do. Here on the US Gulf Coast the humidity in the coming months will be persistent!
* Mike, my wife had the same comment…
* Eric, when the devices are in use they’re in a naturally-ventilated white PVC shelter, away from sunlight. The initial shelters, which I built, look remarkably like the three devices Anthony photographed at the climate research. While I’m comfortable that mine would provide good ventilation and protection against raditation I’ll probably obtain some conventional shelters so as to minimize any controversy.
* Doug, to clarify, my search is for larger differences, sort of screening tests. The devices lack the sensitivity to detect tenths of a degree variation. I have tested the devices side-by-side and got almost identical tracking of temperature, which frankly surprised me.
I’ll likely have six or so devices and stands which should allow me to develop area profiles. Among the things that come to mind for testing are the effects of
1 distance (and direction) from tree cover
2.concrete and asphalt
3. grass cover vs bare soil
4. proximity to air conditioners
5. sewer plants, if I can get permission to do so.
More ideas are welcome

April 27, 2008 12:24 pm

The consistency between “deep shade” and “airport” is spurrious or repeatable ?
I can’t believe the match is anything else than chance.

April 27, 2008 12:26 pm

BTW, as long as you don’t have proper shielding for sensors #1 and #2, you can only compare #1 and #2 but not #2 and airport.

April 27, 2008 2:20 pm

Demesure, a short table of this weeks temperature max/min is here:
My conjecture is that the dark, exposed soil, and the presence of a nearby stained deck and garage wall, contribute to heat accumulation.
Regarding shielding, in my judgment my shielding is at least as good as the standard MMTS type but, to avoid the inevitable questions I’ll likely use standard. I like what I take to be be the climate data folks’ sensors, as shown in Anthony’s photo, where an overhead shield, probably a polymer, keeps direct sunlight from the sensor housing entirely and which is similar to my construction.

Tom in Florida
April 27, 2008 3:15 pm

I am curious about the sudden decrease in temp on sensor #1 around noon. Can I attribute that to a short period of cloud cover? If so, it says volumes about how much energy a sensor absorbs while in direct sunlight as opposed to what the real temperature of the surrounding air actually is. I always get a chuckle when I see a glass enclosed thermometer exposed to direct sunlight on a football field as the announcer tells us the temperature on the field is 120 degrees. Can I also assume that the airport sensor is shielded from direct sunlight therefore making the temperature recording closer to your shaded sensor?

April 27, 2008 6:41 pm

Tom in Florida said:
“I am curious about the sudden decrease in temp on sensor #1 around noon. Can I attribute that to a short period of cloud cover? If so, it says volumes about how much energy a sensor absorbs while in direct sunlight as opposed to what the real temperature of the surrounding air actually is. I always get a chuckle when I see a glass enclosed thermometer exposed to direct sunlight on a football field as the announcer tells us the temperature on the field is 120 degrees. Can I also assume that the airport sensor is shielded from direct sunlight therefore making the temperature recording closer to your shaded sensor?”
The only way to check that is to either 1) shield both from direct sun, or 2) put a light meter to check for sudden cloud cover.

April 27, 2008 6:53 pm

The graph says it was a windy day (the nighttime trace suggests it was not a dark and stormy night).
I would expect that there would be enough air motion to keep it at a pretty homogeneous temperature throughout the area, and your temperature trace at #2 matches what I see frequently at . The exponential decay at night until dawn then quick rise in the morning says you had a temperature inversion and probably very little wind.
Around 0900, the air near the ground became warm enough to rise and the air column begins to mix. Air flows up, air flows down, and winds aloft mix down to the surface. The warming rate slows way down because now you have to heat the entire air column instead of the layer right next to the ground. The air temperature becomes a little ragged, I suspect due to the air not being perfectly neutrally buoyant and momentum brings parcels all the way down to the ground that ought to be floating. (Perhaps convection cells would be the better analog.)
With the wind, I would expect that #1, #2, and the airport all track well. That #1 doesn’t suggests to me that your enclosure could look fine in visible light, but may be translucent or darker at longer wavelengths or somehow isn’t letting enough air inside.
What’s immediately upwind of #1? I wonder if there is something with enough surface area that the sun can heat it and warmed air is blowing on the enclosure and thermometer.
One thing I’ve wanted to do with several temp. loggers is to mount them in a vertical string to measure the effect of radiational cooling. E.g. at 1, 2, 3, 4, 6, 8 10 feet. You may be able to see a wave of cold gradually rise over night. I’ve often seen frost on my car despite my temperature sensor, about 10′ above the ground, not reaching freezing overnight.
I remember one night in Pittsburgh walking between Schenley Park and Panther Hollow. Cool, rather chilly air was flowing down the park, over me, and into the valley. Meanwhile the warmth from the buildings on the other side of the valley helped keep my that side of me more comfortable.

David Smith
April 27, 2008 7:49 pm

Both were shielded from direct sun.
The shield type I’ll be using for the “official” runs is this one:

April 27, 2008 8:25 pm

Interesting experiment, Dave. I looked at the USB data loggers, and they are VERY interesting. I would buy one if I could find a practical use for it around the house.
There is an almost infinite possibility of products to be developed from this concept.
How long will it be before we have data loggers measuring temp, humidity, Cosmic Rays, TSI, CO2 levels, methane levels, complete with Bristle Cone Pine proxies for comparisons.
These next level data loggers will be programmed to pre-fudge the data, saving NASA time and money in their data adjustments.
I love technology.

Pierre Gosselin
April 28, 2008 3:28 am

Couldn’t resist this gadget – and ordered one.
Wonder how long it’ll take them to ship to Europe.
REPLY: Customs is usually the bugaboo, figure 2 weeks

Retired Engineer
April 28, 2008 8:01 am

One thing to check: My logger looks like yours, came from Extech. Can be set to C or F, gives 0.1 deg resolution when in C mode. Not as good as 0.1 deg F, but better than whole degrees.
Another comment: With many data points, you can get a good average temp. Compare this to the midpoint of Tmin and Tmax. That never seemed to be a good way of determining average, now you can check it.

Sam Urbinto
April 28, 2008 8:35 am

The idea is that if you have the thermometer in the same conditions, even if it’s “hotter” or “colder” than it “should be” you’re only getting an anomaly, so point X over time is the same for the area as point Y over time or point Z over time in that same area.
That should be true, as an offset over time, but rests on what seems to be a fatal flaw; that a thermometer in a location where plants, buildings, artificial watering et al change over seasons, years, decades and multiple decades can stay in “the same conditions”.
It seems obvious that they don’t. So what is the anomaly telling us? Which is why I refer to the anomaly or its trend. It’s not temperature. Temperature is a measurement, not a derived monthly average of mean daily min/max calculations compared to some other period. Especially one that probably doesn’t reflect anything physical.
As one example, I put in a shed close by one day. Not enough to “directly” influence things, but what about wind patterns? Then a fence or few start going up in the area. Bigger houses. And so on and so forth.

Sam Urbinto
April 28, 2008 8:48 am

Oh, and it might be more appropriate to get the USB2 model, as it also tracks relative humidity and dew point.
The interesting thing though is the level of accuracy of both.
The temperature accuracy is +/- 1 F Which is also the internal resolution. That’s the USB2.
The USB1 is a resolution of 1 F also, but the accuracy is +/- 2 F
So today’s mean of 55 and tomorrow’s mean of 56 might actually be backwards or the same or ?
REPLY: The inexpensive solid state sensor has it’s limits.

April 28, 2008 10:01 am

I find it curious that the deep shade was a degree warmer at night, was this an artifact of calibration? Were both sensors reading the same temperature before you placed them? Otherwise this might be an example of radiational cooling at night, where the overhead trees are acting like cloud cover reflecting the IR back to the ground or possibly reducing convection.

April 28, 2008 10:02 am

I agree with Doug on the subject of calibration as I too have an extensive background in metrology.
When error bars are reported for surface temperatures, it makes me cringe. Our climate controlled labs don’t have the uncertainty claimed by climate “scientists”.
Doug is also correct in saying:
“Unfortunately assurances are costly but errors should be known and stated or they will be exploited by those who need to support some issue other than the science.”
All sources of error must be identified before before reporting the uncertainty. It is not an easy task, but the integrity of the data relies on correct calculations of the uncertainty. And true as well, the data can be exploited. Honestly, the idea that wind balances out the UHI/microsite issues (Parker et al) is completely laughable to those of us in the real world of the science of measurement (metrology). It seems Pielke and a few others (obviously Anthony Watts) are the only ones who fully understand this concept. Common sense goes a long way for certain things in this world, and it doesn’t take a PhD to see them.
Now we have NOAA on the one hand announcing a renovation of the surface station network, which is good, but on the other they say all the previous records have been “statistically corrected”. Surely they must have been blushing when making that statement.
I have little doubt Doug and Retired Engineer would agree if in our fields we turned in a study using the same sloppy methods used by climate “science”, we’d be out on the street.
There are lies, damned lies, and temperature records 🙂

Pierre Gosselin
April 28, 2008 10:55 am

Thanks for the delivery time David.
2 weeks! Darn it!
So I won’t get it for my birthday tomorrow. 😉

Jeff B.
April 28, 2008 12:03 pm
Doug Taylor
April 28, 2008 2:03 pm

Doug and DR–Excellent brief exposition of measurement theory and near earth temperature measurements. We should continue this discussion, and how it affects climate model predictions.

Eduardo Carreras
April 28, 2008 2:08 pm

Re: Very Damaging to the AGW crowd.
All periods of low solar activity have been named (Sporer, Maunder, Dalton, Kristen…). How about we name the current one “The Gore Minimum.”

April 28, 2008 5:53 pm

For those interested in uncertainty, a link to free software to play with:

April 28, 2008 6:08 pm

Today we had a north wind and full sunshine. The two backyard thermometers recorded only a 1F difference in maximum temperature, versus the 5F to nearly 10F of last week.
The difference is that last week the wind was from the southeast. Why does that matter?
Well, the layout of adjacent parts of my backyard, which can’t be seen from the camera, is here:
What this shows is that, when the wind is southeast, the breeze travels over about 50 feet of sunlit concrete, through a breezeway and across a brown wooden deck in a manner that probably swirls to the right. Pass that warmed, weakened breeze across the darker compost and bricks beneath the sensor and I suspect that a +5F local temperature rise resulted.
On the other hand, when the wind is northerly it travels across my neighbor’s grass backyard, through about 20 feet feet of shade and then across 10 feet of low vegetation. That’s a zero-concrete “cool” path.
This backyard comparison is simply an exercise, but it suggests the contamination possible by placing temperature sensors too close to human habitation.

Harold Vance
April 28, 2008 9:16 pm

This is off-topic for your post, but today in the Houston Chronicle, Eric Berger posted a story to the effect that CSU is thinking about ending support for William Gray’s hurricane forecasts. The excuse proffered by CSU is that “handling the media inquiries… requires too much time and detracts from efforts to promote other professors’ work.”
Here is the URL:
It’s just hard to imagine a university actually complaining about getting too much national and world recognition. lol.

Pierre Gosselin
April 29, 2008 12:48 am

Jeff B.
I’ve read it. He was quite general about it. Just another guy with nothing really new chiming in with the Ice Age scare.
The more I think about all this, the more I believe we have to watch what the oceans do. How the oceans react will tell us what’s in the pipeline, and will confirm solar correlations. Temperatures have been dropping slowly, but not alarmingly.
Many climate signals are still mixed. The AGW alarmists still have enough melting, weather anomalies, changes, warm records at their hands to keep their story going for quite some time.

Pierre Gosselin
April 29, 2008 12:49 am

For every cold, snowy incident us skeptics point to, the alarmists can still always come with “Yeah, but look at the blah blah blah”.

April 29, 2008 5:46 am

One would think that over time the sum total vectors of wind direction would equal zero on an annual basis since each season has a prevailing wind direction. Which means seasonal temperature is dependent also on prevailing wind direction. I believe this is an important point missed in the discussion since the atmospheric temperature is dependent on wind to convect heat away from the surface. The slower the wind, the greater the heat build up.

May 1, 2008 5:03 pm

You have heard of evaporative cooling from plants, right? The point is not so much the shading from large trees, its the cooling in the evening hours as plants release water from photosynthesis. It’s quite obvious is drier locales that are heavily shaded by large older trees versus open areas with lots of concrete/blacktop or rock (granite/basalts/gravels/sand). Human habitation (as a source of thermal anomaly as the shift from ”natural’ vegetation density/type to locale.
Best place to measure would be nonperturbed environments; ain’t gonna be easy to find. And it is worthwhile to measure thermal trends in non-natural environments. Consider the difference in evening desert thermal regimes in say, downtown Las Vegas (roof top or street paved surfaces) and fifty miles away in rural desert setting.

May 26, 2008 10:49 am

[…] is an update on recent field tests with remote thermometers (see the ”Fun with Thermometers” post for  […]

%d bloggers like this: