This may be partially due to low sun angles associated with latitude and the 24 hour day that Antarctica experiences in summer, along with snow albedo. I’m looking into the full paper and will have a follow up post later. See the map below, there are a lot of these AWS. As I’ve always said, if you don’t fully understand the micro-site conditions in which the measurements are made, and the instrumentation limitations, you can’t be certain of the data it produces.
Atmospheric temperature measurements biases on the Antarctic plateau
Christophe Genthon, Delphine Six, Vincent FavierLaboratoire de Glaciologie et Géophysique de l’Environnement, CNRS/UJF, Saint Martin, d’Hères, France
Matthew Lazzara, Linda KellerAntarctic Meteorological Research Center, University of Wisconsin-Madison, Madison, USA
Abstract
Observations of atmospheric temperature made on the Antarctic plateau with thermistors housed in naturally (wind) ventilated radiation shields are shown to be significantly warm biased by solar radiation. High incoming solar flux and high surface albedo result in radiation biases in Gill (multiplate) styled shields that can occasionally exceed 10°C in summer in case of low wind speed.
Although stronger and more frequent when incoming solar radiation is high, biases exceeding 8°C are found even when solar is less than 200 Wm−2. Comparing with sonic thermometers, which are not affected by radiation but which are too complex to be routinely used for mean temperature monitoring, commercially available aspirated shields are shown to efficiently protect thermistor measurements from solar radiation biases. Most of the available in situ reports of atmospheric temperature on the Antarctic plateau are from automatic weather stations that use passive shields and are thus likely warm biased in the summer. In spite of low power consumption, deploying aspirated shields at remote locations in such a difficult environment may be a challenge. Bias correction formulae are not easily derived and are obviously shield dependent. On the other hand, because of a strong dependence of bias to wind speed, filtering out temperature reports for wind speed less than a given threshold (about 4–6 ms−1 for the shields tested here) may be an efficient way to quality control the data, albeit at the cost of significant data loss and records biased towards high wind speed cases.
Received: May 13, 2011; Revised: August 01, 2011
=================================================================
h/t to the Hockey Schtick
Can one assume that the warnings about global warning may be false
As others have said, there’s a big difference between “warming” (trend) and “warm” (bias). The first paragraph above also mentions winds, so my question would be if there’s been a decrease in average wind speeds over the last N years. That would turn a bias into a trend.
RHS,
I would assume that UHI works just as well in winter as it does in summer. At least there shouldn’t be a 3F degree difference between winter and summer trends. That is a pretty significant difference.
Another question I would have is if Stevenson screens are still being replaced with the new sensors or has all of that work been done already? The reason I ask this is that it would account for the summer trend staying flat while the winter trend is in steep decline. So lets say the summer trend is actually declining too, but as Stevenson screen stations are replaced by the new sensors, the increase in erroneous summer high temperatures would accumulate over time and compensate for the decline in actual temperature.
This has also brought to mind something E.M. Smith noted in his blog sometime back when he was looking at temperatures reported at stations around the SF Bay area. That is, some stations seemed to show a very significant and sharp step up in temperature at a certain time of the morning. If this is due to low sun angle or reflected sun penetrating the shade vanes, that might account for that.
I’d have to go back and research a bit, but there seems to be an awful lot more data points on that map than were referenced in Steig et al, especially towards the center of the continent.
Nail after nail after nail driven into the AGW coffin.
But they are not going to give it up.
Well this should be a new chapter for the IPCC to flog. I mean really, the world’s best scientists, doing the best research and gathering the most accurate data . . . what’s a small 8-10 degree discrepancy.
Climate Scientology marches on.
What will they say at RC?
REPLY: They’ll either ignore it or say it doesn’t matter, just like with all the issues raised with the hockey stick. One they use the word “robust” to describe anything, there is no turning back with them. – Anthony
I am at a loss to understand why Steig et al used only a handful of thermometers, weighted especially on the western peninsular area. The map shows Antarctica to be a veritable spotted dick of thermometers.
Before snipping the metaphor please know that spotted dick is an english dessert:
“Spotted dick is an English steamed suet pudding containing dried fruit (usually currants) commonly served with custard” (Wiki)
Well it seems the Antarctic goose has an Arctic gander.
http://tinyurl.com/3waovy8
Al Gore was recently in Aspen, Colorado, overheard saying, “these thermometer thingies, they’re B*LL$h!+!!!”
Gore also remarked on the fact that it was “Big AWuhl Wat made muh carbon credit sacrament payments not work so wayull, but buhtwean me & Karl Marx, we dun made tha planet stop warmin, although it ain’t stopped warmin.”
“Also, my wife is a liar. Oh. Wait my wife ain’t no liar, cuz that book ain’t out yet.
But it was BIG AWULL WAT made muh WIFE get fed up and leave me. See how ‘fed up’ has FED in it, and how them BIG AWULLz iz always a’ palloootn muh FED with rePugNaKunz?”
And with that Gore proceeded to launch his platform of paying poor and disillusioned urban youth to stand outdoors in antarctica and estimate the temperature in their neighborhood slang lingos, saying the program would provide “much needed relief for waterless urinal businesses previously LOCKED OUT of the MARKETs in Antarctica by “creationists who couldn’t tell climate math from useless doodles.”
Wayne, you are correct about changes in wind speed affecting reported temperature, especially in the Antarctic winter. There are both decadal trends in wind speed and wind speed changes due to (seemingly slight) changes in sensor location which must be accounted for in any temperture study. My impression, however, is that the wind effects are usually ignored in the Antarctic studies.
For Crosspatch, who says: “I would assume that UHI works just as well in winter as it does in summer.”
Assume the UHI is caused by two major but different sources of heating — (A) solar and (B) various urban sources of heat such vehicle emissions. The amount of blacktop and development would be the same summer and winter. If solar heating in summer is greater than in winter, due to the sun being more directly above in summer but with fewer hours of light and sunlight at a lower angle in winter, then UHI might be higher in summer.
Also, possibly more important, the recent study by NASA, hilighted on WUWT, showed summertime UHI effects higher by 7 to 9 degrees C on average in the eastern US compared to neighboring areas, with the largest UHI occurring when neighboring areas were forested instead of farmland. In winter, forested areas don’t have green vegetation and water retention and don’t cool the land the way they do in summer.
This is a qualitative argument that UHI may well be higher in summer. I don’t know of quantitative measurements of the difference between summer vs. winter UHI.
Link:
http://wattsupwiththat.com/2010/12/14/image-the-urban-heat-islands-in-the-northeast/
John
Well, it should be easy enough to tell. Look at the winter lows for downtown Indianapolis oe Denver compared to a small town outside of the city. Then look at the summer lows.
This is starting to make sense to me. It seemed that the Antarctic sea ice should not have been steady/increasing while W Antarctica warmed up, because W A should have a disproportionately high effect on sea ice.
If, instead, the thermometers were being incorrectly heated by the sun, my thinking goes like this:
Henrik Svensmark’s theory is that clouds are driven by GCRs, and in turn drive global temperature. The late 20thC warming was driven by decreasing cloud cover (confirmed by Earthshine). The Antarctic does not warm in parallel, because ice albedo is stronger than cloud albedo. But .. the reduced cloud cover lets in more sunlight, hitting the thermometers where it shouldn’t, and bingo! W A gets a spurious warming trend.
The observed behaviour of Antarctic sea ice would, to my mind, confirm that the W A temperature rise could well have been spurious. Further confirmation would come if the spurious warming is now declining again.
Reasonable?
“August 10, 2011 at 12:25 pm
“up to 10°C” – compared to what? Presumably true air temperature. If the old thermometers had a similar bias, then this wouldn’t contribute to a false warming, just that the temperature records are poor and too warm from sunny periods.”
If the sensor has a bias and you never change the sensor, then taking an anomaly effectively removes the bias because what you are concerned about is the trend and not the absolute temperature. The other thing is with the bias only happening under certain conditions, you will end up with a more noisy anomaly. If you change sensors from one that has a too warm bias to one that is correct you will induce an artificial cooling trend.
Considering that temperatures fall in Antarctica in sunny (cloud free) weather, it seems extremely important not to measure irradiation.
steven mosher says:
August 10, 2011 at 2:53 pm
“August 10, 2011 at 12:25 pm
“up to 10°C” – compared to what? Presumably true air temperature. If the old thermometers had a similar bias, then this wouldn’t contribute to a false warming, just that the temperature records are poor and too warm from sunny periods.”
If the sensor has a bias and you never change the sensor, then taking an anomaly effectively removes the bias because what you are concerned about is the trend and not the absolute temperature. The other thing is with the bias only happening under certain conditions, you will end up with a more noisy anomaly. If you change sensors from one that has a too warm bias to one that is correct you will induce an artificial cooling trend.
***************************************************************************************************
Can you tell me how the anomaly is calculated. Is it by calculating an average for a previous period, (perhaps) before the “warmer” shield was in place.
I don’t know whether the sensor has been changed, but the shielding has. If this has changed the temperature reading upward, then the comparison with the previous temperature period(with older, colder shield) will also change disproportionately and affect the trend.
Likewise, as you point out, if you change from an old shield/sensor which is correct, to a new, warmer sensor/shield you will induce a warming trend. This is actually what appears to have happened.
The IPCC statements about “settled science” and “certainty” of outcome (of CO2 increases) depends, it seems clear to me, on PRECISE data but INACCURATE data. Temperatures that are multiples of one sensor in one place at one time may be averaged and its proximity to “truth” become better as a result, but data grouped each day at different places from different measuring devices and (in the pre-electronic days) by different people in different frames of mind cannot, it seems to me, merged together and, by averaging, have a result that is more accurate than the general reading. If 100 readings are done at one place, one time, by one person, of one event, then the actual number probably does have an error equal to a square root function, for example, but 100 readings of different places by different people must have, it seems to me no less an error of the majority reading. All errors must add, not become reduced, through manipulation. The IPCC/Hansen data, it strikes me, is less accurate than portrayed, though the statistics used to get it give many decimal places for precision.
Certainty of outcome is model driven and may be very small. But if models are matching to observations with a large error due to inaccuracy, not precision, what happens to the certainty that results? Does not certainty fail to improve beyond whatever the accuracy, not precision, of the data is?
Hansen and HadCru both say they describe the global temperature – not the temperature of their dataset, but the global temperature. The global temperature trends are important for accuracy more than for precision. But Hansen, with his :’superior’ data of the Arctic, has about a 0.12C warmer view of the world in a world that has warmed about 0.55C since their (same?) reference point, a 18% or so difference. Since the models are using this data to determine their fit, then the accuracy of the model fil cannot, it seems to me, fit better than the accuracy of the data being used. This does not leave us with 95% or 99% certainty, it seems to me.
Am I wrong? Is the work on measurement, equipment, UHIE and interpolation not showing us that the accuracy of the trend is greater than the accuracy possible as described by IPCC “certainty”? Would a statement that the “settled” science produces a certainty in the probable 82% range not be more correct than the 95 – 99% range?
And that doesn’t include the certainty in the modelling: what IS the certainty of outcome when data quality (not quantity) and modelling assumption errors are considered?
What can we say is the accuracy of any reading? I know the error bars, but I still hear that the temperature has risen 0.05C worldwide: if none of the readings are better than 0.15C in reflecting reality, can you say that 0.05C has any meaning? And if not, what does that do to the various, finely tuned modelling results? Are they really just a smear?
crosspatch says:
August 10, 2011 at 11:31 am
Well, if this is the case, it should be easily seen in the data that most of the recent “warming” would occur in the summer months. CO2 induced warming would have its greatest impact in winter months when the air is very dry (little water vapor) and CO2 accounts for a greater portion of the total greenhouse effect of the atmosphere. Is most (all?) of the recent “warming” occurring in the summer months?
This work which uses spectral analysis of DLR to quantify the contribution of the various GHGs to total DLR suggests CO2 accounts for a full third year round.
http://journals.ametsoc.org/doi/abs/10.1175/JCLI3525.1
The study was done at the South Pole and indicates that location should be the area of CO2’s peak impact on the GH effect, which is interesting if you are at all familiar with what observations there show in regard to temperature trends in the last fifty years
This is exactly the sort of systematic sensor measurement error Hubbard and Lin warned about in their 2002 paper, “Realtime data filtering models for air temperature measurements“. In 2005, they also explicitly warned about the systematic sensor measurement errors due to snow albedo: “Surface air temperature records biased by snow-covered surface”
Steve Mosher’s demurral about anomalies is true only when the bias error is constant and the systematic measurement error is i.i.d. However, since the size and the distribution of the systematic measurement error is local weather dependent, the error will also vary systematically in magnitude and the distribution of error may skew differently over time. Bias errors are very likely to vary among any given set of mean temperatures, bias errors will vary among various data sets such as from month to month, and the standard deviation of the systematic bias errors will also vary with the data set.
In short, unless they have been measured to be constant and i.i.d., systematic sensor measurement bias errors do not fully subtract away, cannot be assumed constant over time and cannot be assumed to be i.i.d. Systematic bias errors cannot be assumed to subtract away by taking anomalies. And even if the mean bias happens to subtract away in the difference between favored data sets, the standard deviation of the bias error does not.
When were these things installed….and what was the temp jump when they were?
I “think” I remember reading that cooling summers in Antarctica were consistent with irritable climate syndrome…..can’t keep up, everything is consistent
Would even more cooling be even more consistent?
Trend – schmind – bullhollocks
10 degree too warm – is 10 degrees too warm – period
Kevin Kilty said;
“Of the five dimensions of measurement, temperature is the most challenging.”
Well, yes and no, I’m not sure which five dimensions you refer to, but I agree that measuring temperature is more difficult than length, time, voltage, weight etc.
However measuring optical radiation accurately is much more difficult. According to NIST if you follow all of their procedures you might get an absolute radiometric measurement in the visible light region of about +/- 2%. This involves a lengthy process that involves buying a “standard” light bulb from them and using careful spectral measurements taken every few nanometers across the spectrum. The international standard for optical radiation is the flux emitted by pure platinum in its liquid state. Even NIST can only afford to perform this measurement every few decades. They measure some light bulbs against this standard and then transfer this data to other light bulbs that you can purchase. Each transfer of data adds about 1/2 % of error.
Absolute optical radiation measurements in the infrared are even less well developed and errors of several percent are to be expected. Relative measurements are of course quite a bit easier.
I have done temperature and optical measurements and if I had a choice I would much rather determine the absolute temperature of a material.
Inferring the absolute temperature of a surface by measuring the total optical radiation (satellite measurements) is even more difficult again. It involves knowledge of the emissivity (across the whole spectrum) and a bunch of other error sources. I think that the absolute accuracy of the satellite measurements are AT BEST +/- 1/2 degree F.
Cheers, Kevin.
steven mosher,
Your comment is valid for a single station over the life of that specific instrument. Maybe. Maybe not. The instrument error may not be constant over time. As mentioned in the subject paper, low wind during summer resulted in maximum error. OK, you may say, on a year to year basis the bias is the same. That would hold true if windiness and cloudiness are constant year over year. Are you willing to assume that much?
Of course, we’re not only interested in the single site, single instrument temperature measurements. We want to incorporate this ‘biased’ set of data into our existing network of observations. Do we just slap in the data without regard to the sometimes 0 C, sometimes up to 10 C, errors? I should hope not. How does one quantify the uncertainty we introduce? Or should we give it CRN 5 “DEFECTIVE” sticker and toss it on the junk pile?
A solar bias could be critical. If I remember correctly we just passed the peak of summer time exposure due to the interaction of planetary tilt and the earth’s orbit.
It seems that the shield shown here is representative of Gill-style multi-plate radiation shields: http://www.youngusa.com/products/2/11.html
If so, the angle of the louvers or plates makes it easy to see how solar radiation could warm the core from highly reflective ground cover (snow), particularly if the sun is low in the sky (arctic condition).
For comparison, an aspirated shield like that they seem to use as a reference for the temperature comparison is shown here:
http://www.youngusa.com/products/2/16.html