Bishop Hill Australian temperatures
Nov 4, 2011 Climate: Surface This is a guest post by Philip Bradley. Please note that all graphs can be seen full size by clicking on the image.
An analysis of Australian temperature data recorded at fixed times, and the implications for the 'global average land surface temperature' derived from minimum and maximum temperatures
Jonathan Lowe, an Australian statistician, has performed extensive analysis of weather data recorded at fixed times by Australia's Bureau of Meteorology (BoM). This analysis is available at his blog, A Gust of Hot Air. The data comes from 21 weather stations manned by professional meteorologists.
This work needs to be brought to a wider audience because it paints a very different climate picture to the global land datasets based on minimum and maximum temperatures – GISS, HadCRUT and the recent BEST analysis.
I'll present his discoveries in three parts, as there appear to be three significant elements in his work.
-
Using a minimum and maximum temperature dataset exaggerates the increase in the global average land surface temperature over the last 60 years by approximately 45%
-
Almost all the warming over the last 60 years occurred between 6am and 12 noon
-
Warming is strongly correlated with decreasing cloud cover during the daytime and is therefore caused by increased solar insolation
-
Reduced anthropogenic aerosols (and clouds seeded by anthropogenic aerosols) are the cause of most the observed warming over the last 60 years
I'll then add a part 4 covering some additional analysis of mine
Part 1 - Using a minimum and maximum temperature dataset exaggerates the increase in the global average land surface temperature over the last 60 years by approximately 45%
The BoM records temperature, and other values such at cloud cover, at three-hourly intervals for its main surface weather stations. This dataset goes back 60 years covering the same time period as the recent highly publicized BEST project.
Jonathan's analyses of the fixed time temperature data show that days have become significantly warmer. However, nights have warmed substantially less (only about a third of the day time warming).
Contrast this with the analyses based on minimum and maximum temperatures (Tmin, Tmax) which show warmer days and even warmer nights, based on the assumption Tmin represents nighttime temperatures.
In fact, Tmin usually occurs after dawn at the point when incoming solar insolation exceeds outgoing longwave radiation.
GISS, HadCRUT and BEST wrongly assume the mean of Tmin+Tmax represents average daily (24 hour) temperatures.
At best, the mean of Tmin+Tmax represents average daytime temperatures, and Jonathan's work confirms this.
Note at this point, that any day time warming that does not persist through the following night time is irrelevant to climate warming, as it is not heat that is retained in the climate system for more than 24 hours. So even a correctly derived average temperature is a poor indicator of the amount of warming.
In summary, calculating average temperature from the 3 hourly data instead of the Tmin+Tmax/2 method results in ~45% less warming since 1950.
The graph below illustrates this.
;)
GISS, HadCRUT and Best all refer to their results based on Tmin+Tmax/2 as 'global average temperature'. Clearly, daily temperatures are a range and picking the mid-point of a range (which is what the mean of Tmin and Tmax is) is a poor way to estimate an average. Averaging measurements taken at fixed intervals will provided a more accurate estimate of the daily average.
The ~ 45% warming bias introduced into the 'global average temperature' by using Tmin+Tmax/2 is in itself is a very significant discovery, but there is considerably more to Jonathan's discoveries.
Part 2 – Almost all of the warming over the last 60 years occurred between 6am and 12 noon
Jonathan uses the simple but elegant method of subtracting the temperature at one three-hour interval from the temperature at the next three-hour interval to determine at what times of day the warming over the last 60 years has occurred. This breaks down the total warming over the last 60 years into three-hour slices of the day.
You can see his graphs here - http://gustofhotair.blogspot.com/2009/04/analysis-of-australian-temperature-part_16.html
He finds that warming has occurred in only three of these time slices; 6am to 9am, 9am to 12 noon and 6pm to 9pm. The other five three-hour time slices show either a cooling trend or no significant trend.
In parts 3 and 4, I discuss physical mechanisms that would cause temperature rises to be restricted to these time periods.
Jonathan's analysis shows that there has been no warming occurring at nighttime at all since 1950. Although nighttime temperatures have risen due to residual heat from warmer days.
There are numerous published papers that uncritically state that nighttime temperatures are warming based on the fact daily minimum temperature is increasing.
In 1990, Kukla and Karl published a paper in Environmental Science and Technology that contained the following,
Almost all of the observed increase in mean daily temperature over land over the past 40 years appear to be a result of the increase in early morning minimum (temperature). One is tempted to reason that the efficient blanket of the CO2 rich atmosphere keeps the nighttime temperatures high.
I find it surprising to say the least, that Tom Karl, Director Of The National Climate Data Center hasn't in the last 20 years investigated whether minimum temperatures (which he correctly states occurs in the early morning) do indeed measure nighttime temperatures, as this is perhaps the most important assumption underlying the evidence for AGW.
I could find only one published paper that investigates the relationship between minimum temperatures and measured nighttime temperatures and that was based on hourly temperature measurements taken at one location, the Manua Loa Observatory, and apparently made by the same team that measures CO2 levels. They found that hourly nighttime temperatures did indeed closely correspond with minimum temperatures. Interestingly, they found a slight decrease in daytime temperatures measured at 12 noon, The authors attribute their results to GHG warming. Although they concede other effects could be at work, including clouds.
http://www.clim-past.net/7/975/2011/cp-7-975-2011.pdf
A study using Anthony Watts' Surface Stations data reaches a couple of interesting, albeit tentative conclusions. Firstly, it found that for the best sited stations, diurnal temperature range (DTR) has not decreased and perhaps even increased over the last century. Secondly, it found that the PDO has a cyclical influence on DTR.
Jonathan's work indicates a marked increase in the range of nighttime to daytime temperatures using the three-hourly data, even though this data doesn't directly measure DTR. Decreasing DTR is considered the signature of GHG warming (more nighttime warming than daytime warming), and this study concludes decreasing DTR at the poorer sited stations may well result from data quality issues.
The Manua Loa study used data from 1977 to 2006, which closely corresponds with the warm phase of the PDO, and this may account for the findings.
Fixed time temperature data is available in the USA. The American Meteorological Society states at its website.
The National Weather Service (NWS) and its predecessor, the U.S. Weather Bureau, have operated a network of weather observation stations and offices in or near many of the large cities in every state, commonwealth, and territory under its jurisdiction. At many of these nearly 300 "first-order stations," systematic measurements of numerous weather elements are made by professional observers. Some of these weather data are recorded hourly,
I assume temperature measurements at fixed times is available in other countries.
Climate science collectively having wrongly assumed minimum temperatures measure nighttime temperatures then goes on to misinterpret the role of clouds, which is the next part of Jonathan's analysis.
Part 3 -Warming is strongly correlated with decreasing cloud cover during the daytime and is therefore caused by increased solar insolation
The BoM includes cloud cover in its three-hourly dataset.
Note that the BoM's website says that cloud cover is still measured by the traditional method of an observer estimating how many eighths of the sky is cloud covered. While I find this surprising, without evidence to the contrary I'll assume this information is correct.
The BoM's three-hourly cloud cover data shows that there has been a marked decrease in cloud cover in the daytime, with a marked increase in cloud cover in the early part of the nighttime (measured at 9pm).
;)
Jonathan then correlates the cloud cover trend with the temperature trend for each three-hourly set of data. Shown in the graphic below.
;)
Unsurprisingly, he finds decreased daytime cloud cover strongly correlated with increased daytime temperatures. There is a weaker correlation between cloud cover and overnight temperatures.
Surprisingly, he finds that the increased cloud cover at 9pm and midnight doesn't correlate with increased warming, which is directly contrary to what is generally accepted as the effect of clouds on nightime temperatures. In part 4, I propose that these clouds are condensation from nighttime radiative cooling, and therefore thin limiting their capacity to reflect longwave radiation. However, clouds and LWR is a complex subject.
The significance of the cloud cover temperature correlation is that it occurs in the daytime and accounts for both increased minimum and maximum temperatures, as decreased clouds result in increased solar insolation and warmer days.
The effect of clouds on minimum and maximum temperature seems to be widely misunderstood in the climate science community. For example,
From 1950 to at least the mid-1990s, minimum temperatures on land have increased about twice as fast as maximum temperatures [Easterling et al., 1997]. This may be attributable in part to increasing cloudiness, which reduces daytime warming by reflection of sunlight while retarding the nighttime loss of heat [Karl et al., 1997].
Karl has got things exactly the wrong way round.
Part 4 - Reduced anthropogenic aerosols (and clouds seeded by anthropogenic aerosols) are the cause of most the observed warming in the land surface over the last 60 years
While Jonathan correctly concludes that increased solar insolation is the primary cause of both increased minimum and increased maximum land surface temperatures, I'd like to extend that conclusion with some analysis of my own.
I mentioned earlier that the daily minimum temperature usually occurs in the early morning at the point when (incoming) solar insolation exceeds net outgoing longwave radiation. In the tropics this is a few minutes after dawn. At higher latitudes in winter it could be an hour or more after dawn.
When the sun rises, solar irradiance initially travels at a low angle through the atmosphere and as a result traverses far more of the atmosphere than it will for most of the day.
Any atmospheric effect that blocks or scatters solar irradiance and hence reduces solar insolation will have its maximum effect just after dawn around the time the minimum temperature occurs. Making the minimum temperature particularly sensitive to atmospheric effects that reduce insolation.
There are two effects at work reducing (especially early morning) solar insolation. One is clouds, particularly low level clouds. The other is aerosol pollution (smoke and haze), particularly low level aerosol pollution
Bear in mind that aerosols act as cloud seeding nuclei. So clouds and aerosols don't vary independently.
A great deal of work has been done on the effect of aerosols on the climate and atmospheric temperatures. However, I am concerned with one specific effect that of aerosols on daily minimum temperature.
Which makes me focus on the effect of black carbon aerosols (smoke to most people).
The Daily University Science News reported,
Black Carbon Aerosol Pollution Cools, Heats, Confuses
New research based on NASA satellite data and a multinational field experiment shows that black carbon aerosol pollution produced by humans can impact global climate as well as seasonal cycles of rainfall.
Because aerosols that contain black carbon both absorb and reflect incoming sunlight, these particles can exert a regional cooling influence on Earth's surface that is about 3 times greater than the warming effect of greenhouse gases.
But even as these aerosols reduce by as much as 10 percent the amount of sunlight reaching the surface, they increase the solar energy absorbed in the atmosphere by 50 percent -- thus making it possible to both cool the surface and warm the atmosphere.
Scientists are concerned that this heating may perturb atmospheric circulation and rainfall patterns.
"When we combined the satellite measurements with surface measurements, we found that the reduction of sunlight reaching the surface was three times larger than the amount of sunlight reflected to space," says V. Ramanathan, director of the Center for Clouds, Chemistry, and Climate at Scripps Institution of Oceanography at the University of California, San Diego. "Averaged over the entire northern Indian Ocean, the man-made pollutants reflected more solar radiation back to space (than pristine skies), but they absorbed up to twice as much radiation in the atmosphere."
The key phrase here is ' both cool the surface and warm the atmosphere' and recall that the daily minimum temperature occurs at the time when solar radiation is traversing the atmosphere at a low angle and any blocking or scattering by aerosols will have its maximum effect.
I discussed this with Jonathan and he confirmed that domestic burning of coal and wood in open fires, agricultural burning and the burning of garden waste was widespread in Australia 50 to 60 years but has since been almost completely eliminated. I grew up in the UK and can confirm the same thing occurred there.
Anyone under the age of 50 will have difficulty appreciating how much smoke an open hearth coal fire produces. I remember in the 1950s and 60s the winter morning ritual of lighting a coal fire. It produced a solid column of smoke up the chimney for 10 to 20 minutes before the coal got properly alight. Multiply that by tens of millions of homes and there was a lot of smoke within a few hundred feet of the ground when the sun rose.
Then there was burning of agricultural and garden waste in the summer and autumn. Burning a pile of wet leaves produces a terrific amount of smoke.
The smoke from a fire is generally pictured as going more or less straight up. It doesn't. Initially it goes upwards but as the smoke cools to the ambient air temperature, it drifts with the wind almost horizontally, forming a layer typically a few hundred feet above the ground.
Starting as early as the 1950s, clean air legislation in the developed world progressively reduced black carbon emissions until almost all routine emissions were eliminated. Which fits with the timing of increasing minimum and maximum temperatures in the global datasets
But what happened in the developing world, particularly rapidly industrializing China and India?
Some studies of temperature trends in China show a larger warming trend than the global average (derived from Tmin+Tmax/2), especially minimum temperature.
http://journals.ametsoc.org/doi/pdf/10.1175/3230.1
On investigation, I found that domestic coal burning which was the ubiquitous form of home heating 50 years is now banned in urban areas. So black carbon emissions will have substantially reduced in and around urban areas in China despite continuing high levels of other aerosol pollutants.
Other studies show substantially less surface warming in China than the global average in recent decades.
http://shadow.eas.gatech.edu/~jean/monsoon/Menonmonsoon.pdf
The situation in India is complicated by the monsoon, which 'washes' aerosol pollution out the atmosphere. There is no doubt that domestic burning of coal, wood, agricultural waste and dung has substantially increased black carbon aerosol levels outside the monsoon season in recent decades. A study from Pune concludes.
The analysis reveals significant decrease in mean annual and mean maximum temperature (since 1900). This decrease in temperature is more pronounced during the winter season, which can be ascribed to a significant increase in the amount of suspended particulate matter (SPM) in the ambient air during the last decade. On the contrary, monsoon season shows warming. This warming can be attributed to a significant increase in the low cloud amount.
The other area of the world that's of relevance is the old Soviet Block, where pollution controls were minimal. When the collapse occurred, almost all of the highly polluting Soviet era industry was shut down within a few years, and former Eastern Bloc countries rapidly implemented the kinds of clean air initiatives already in place in the West, rapidly reducing aerosol levels toward those in the West.
Overall, the global picture is of substantially reduced aerosols, particularly black carbon, over most of the world since 1950. Although with substantial regional and national variations.
It is generally accepted that black carbon aerosols block sunlight, resulting in less solar insolation but at the same time warming the atmosphere. Decreasing black carbon will increase solar insolation, increasing minimum temperatures, while at the same time cooling the troposphere.
If changes in black carbon aerosol levels are as large as I believe, using land surface minimum temperature to measure global temperatures is giving an entirely false picture of the actual atmospheric warming over land. Increasing minimum temperatures is in fact evidence the atmosphere is cooling (and visa versa).
One final point before I wrap things up.
It is well accepted that aerosols, including black carbon, seed cloud formation. Decrease aerosols and you will decrease clouds. Therefore, I argue that the primary driver of changes in clouds found by Jonathan, and which I infer are primarily low level clouds, is changes in aerosol levels, particularly black carbon. Other studies support this conclusion, including the Pune study referred to above.
The reduced anthropogenic black carbon aerosols hypothesis neatly explains Jonathan's discovery of a sharp increase in cloud cover at 9pm.
If reduced anthropogenic aerosols are reducing cloud seeding and leading to less daytime clouds, more water vapour will remain in the atmosphere. This increased daytime water vapour will be more likely to condensate out as clouds when nightime radiative cooling occurs.
Summary
In overall summary, it is clear that increased solar insolation, caused by a combination of decreased clouds and decreased anthropogenic aerosols (and I argue particularly black carbon), is the primary cause of increasing minimum and increasing maximum temperatures, and hence the increase in the land surface 'global average temperature' calculated from these values.
Jonathan's analysis shows deriving 'average temperature' from Tmin+Tmax/2 over-estimates the landsurface average temperature by ~45%. He further shows that most of the remaining warming in the 'average temperature' since 1950 is due to increased solar insolation.
Increased solar insolation from decreased aerosol and aerosol seeded clouds also explains why warming measured by Tmin+Tmax/2 has levelled off in recent years, because in the developed world, parts of the developing world and the ex-Soviet Bloc there is limited remaining black carbon (and other forms of) aerosol pollution left to clean up.
While Jonathan's used Australia-only data, there appears to be no comparable study in any other part of the world and we should assume similar effects exist in the rest of the world.
In conclusion, it is troubling that climate science, including the recent BEST analysis, persists with the minimum and maximum temperature derivation of global land surface average temperature, when the methodology is flawed and the results misleading.
Note that Jonathan also analyses seasonal and tropical versus temperate zone differences. Space considerations made me omit them as they would require considerable discussion, but overall they are supportive of the conclusions above.
Reader Comments (54)
Interesting.
If we're using "temperature anomalies" (ie, the difference compared to a baseline), does this make a differemce, assuming the same measurement criteria are used?
The bit about cloud cover is the most interesting, with the aerosols a close second.
Since our start dates with the instrument record, are well into the "Coal Era", that's going to make quite a difference.
Needs to be published, to stop the "It's not peer-revued" brigade, from ignoring the results.
Absolutely fascinating, Philip - full marks. We still have bonfires whenever possible, so I'm doing my bit!
It's definitely a compliment to this site's popularity that we're starting to get commercial spam
[Now removed. BH]
Adam,
I believe Jonathan's data from the BoM was monthly averages of actual 3 hourly readings, not anomalies.
Where Jonathan shows anomalies he has calculated them himself.
Unfortunately, Jonathan hasn't responded to my attempts to contact him and he is not available to answer these kinds of questions.
This looks like very good analysis, and very important indeed. Calculating a mean as (Max + Min)/2 is crude, being very vulrnerable to brief excursions. I look forward to this being replicated with other data sources, and to the additional insights it promises to bring. Well done Philip Bradley for bringing this to a wider audience, and well done Jonathan Lowe for doing the investigation. Unsettling stuff for those who prefer their science settled!
I'm still not sure I know what 'global temperature' is anyway (even less so now!) and it would be seem to be a necessary starting point for discussions on GW, let alone AGW or CAGW.
How meaningful is any historical global temperature construct when the inputs from Africa, S America and much of Asia are largely worthless?
I remember there was a documentary a while back which reported on data from evaporation pans which measured directly heat from the sun. It must have been after 9/11 because there was some discussion about contrails, and the effect of grounding of all air traffic over the USA.
If these data are available somewhere, I should have thought they might be a useful contributor to the evaluation of insolation effects on temperature.
Many thanks Philip, interesting and thought provoking read.
Have you found an explanation anywhere of how the 21 stations were selected? I have had a quick scan of Jonathan Lowe's "A Gust of Hot Air" but cannot see how they were chosen.
TIA
Green sand,
The 21 stations are all those for which 3 hourly data is available from the BoM. No selection performed by Jonathan.
I assume that is the number of stations staffed 24/7 by meteorologists.
I notice that this was on Jonathan's website in 2009. What I found surprising, for someone claiming to be a statistician, is that he presents pages and pages of regressions, but not an error estimate anywhere that I could see. Since he has a sample of only 21 stations, that is a serious problem.
I live in Sydney. It's bloody cold and been getting colder for as long as I can remember.
I'm a bit concerned that there appears to be confusion in the headpost about whether warming has been exaggerated by 45% or whether there has been 45% less of it.
If what is meant is the former then there has in fact been 31% less of it.
These results are very provocative. Is the data set accessible? Daily data would be more useful than monthly data, given the likely explanatory mechanism.
Justice4Rinka: "exaggerated by 45% or whether there has been 45% less of it"
The first figure shows regression slope changing from 0.0117 (degrees/year, it appears) to 0.0066 this new way. That is about a 45% reduction in the warming trend (66/117 = 0.564, approx. 0.55).
For what it is worth, the burning of wood and organic waste typically produces whiter, or gray smoke, while incomplete diesel fuel burning produces dark, black smoke. The latter contains black, elemental carbon, in the form of molecules of stacked, polynuclear aromatic hydrocarbons (PAH; Ebert, 1990, Science 247, 1468-1471), which in my analysis of remote U.S. aerosols (Huffman, 1996, Atm. Env. Vol. 30-1, 73-99) appeared to be substantially oxygenated, by being internally mixed with sulfate, while wood burning smoke appears to be rather organics-substituted, rather than oxygenated, PAH. The difference is that the typical "black carbon" strongly absorbs light, while the organics from wood/organic waste burning mostly just scatter it. That is, really black smoke, as from diesel fuel burning, is the villain.
OK, so the bit that says
Using a minimum and maximum temperature dataset exaggerates the increase in the global average land surface temperature over the last 60 years by approximately 45%
should in fact read
Using a minimum and maximum temperature dataset exaggerates the increase in the global average land surface temperature over the last 60 years by approximately 77%
Right?
How amusing - an apologist for CAGW actually asks for error bars? I thought the *consensus* was to abolish them. After all, if you're an alarmist psyentist fitting them to regressions with "acceptable" R2 values of, say, 0.00003, then you'd need to pulp the Amazon rainforest to get a big enough piece of paper to show their full extent.
Am I missing something here? The message coming through to me is that a careful analysis of existing data routinely collected by BOM in the course of weather monitoring has revealed (gasp!) that
(a) it's warmer without clouds and hence
(b) the inference is that cloud albedo dominates 'radiative feedback' causing net NEGATIVE feedback by clouds?
Next question: When are other nations going to analyse their data in similar fashion? AND
How WRONG are the IPCC models? And why did we pay millions of dollars to prop up IPCC?
Once the taxes are in place the erroneousness of the science won't matter; we'll be told abolishing the taxes is unaffordable.
"Once the taxes are in place.."
Indeed. Explosive force (both literal and figurative) will be needed to dislodge them.
Harry Dale Huffman,
Thanks. In future I'll refer to black carbon and hydrocarbon aerosols.
What I'd really like to find is a dataset that records exact time of day of the daily minimum temperature.
Brilliant study, thanks!
More general recent anaylsis of Australian temperatures:
http://hidethedecline.eu/pages/ruti/australia.php
K.R. Frank
John Shade "This looks like very good analysis, and very important indeed. Calculating a mean as (Max + Min)/2 is crude, being very vulrnerable to brief excursions. I look forward to this being replicated with other data sources, and to the additional insights it promises to bring."
If other data sources record the 3-hourly data like this one, wouldn't there be scope for a crowd-sourcing project, if someone with the requisite scientific/statistical knowledge was prepared to manage it? I'm sure there'd be plenty of volunteers on sites like this one who would be willing to do the donkey work.
Philip,
Thanks for an illuminating post.
All this emphasizes the silliness of relying on a daily mean that is (Min plus Max)/2 in characterizing surface temperature and just how informative it is to have more detailed data. Hourly would be a good standard now that cheap and reliable electronic devices are available and we have the computing power to handle the data that is produced.
It's great to know that we can learn so much more if we have the required detail of just where in the daily cycle the temperature increased/decreased.
Elementary stuff one would think.
Many years ago I was surprised to discover that the average of twenty four hourly readings for a particular site could be two or more degrees either side of the mean. The mean is not a statistic that should be relied upon by men of science.
Justice4Rinka
Exactly.
So both maximum and minimum daily temperatures occur during the day! This is the kind of unexpected fact our geography teacher used to try and excite us with. Perhaps Phil Jones and James Hansen never did O-level geography? And nobody, anywhere, has been looking at nighttime temperatures at all. As for Dr Muller... forgetting threequarters of the globe is unfortunate, forgetting also half the clockface looks like carelessness.
Just a minor point, but the percentages quoted are a bit disingenuous.
For example, if there is a "10% increase" in temperature from 10°C to 11°C, you are ignoring the fact that you are not using a true zero point to judge that ratio. You should be using 283.15°K to 284.15°K
or about 0.3%.
I can't find a list of the weather stations that were used so their location is unknown for now.
Australia is a large sparsely populated country outside of the major population centres, which are mostly on the East coast.
As the weather patterns move from West to East it is hard to envisage anthropogenic aerosol production being involved in this study. For example, Melbourne, where I live ,is continually subjected to what must be considered as pristine air from the Southern ocean. Other East coast cities are mostly subjected to air from the interior which is very sparsely populated, any aerosols from these centres being dispersed into the Pacific.
Justice4Rinka Once the taxes are in place the erroneousness of the science won't matter; we'll be told abolishing the taxes is unaffordable.
And anyway, as Matt Ridley pointed out, debunking is like water off a duck’s back to pseudoscience.
I don't see any reason to doubt that the CAGW religion will still be going strong 1000 years from now.
Nov 4, 2011 at 12:27 PM | Unregistered Commenter
At last.
(max + min)/2 is not the mean. You have to integrate over time to get the mean.
And S-B radiation is not proportional to mean T^4. You have to integrate over time to get that too.
These things are so very basic they could only happen if the statisticians were kept safely out of the picture.
GGGRRRRR
"if the statisticians were kept safely out of the picture"
As indeed they were. I thought it was an error of omission, but apparently not!
Having had a bit longer to digest all this, I'm astonished that a crude (max+min)/2 is used for serious recording. How on earth is that a representative average? And acceptable for scientific analysis?
Back to the drawing board, chaps...
geoffchambers (et al):
Max and Min USUALLY occur in daylight hours, but not always. It all depends on weather conditions. Still clear conditions Min is shortly after sunup, Max at 2-3pm. That's our (northern Australian) winter usually. But add in a thunderstorm, a cold change, monsoonal rain, a wind from the land or from the sea, and just about anything can happen. Recording (then averaging) daily Tmin and Tmax tells you very little about temperature- a running 12 hour mean of 5 minute data will give you better information. I had always assumed more clouds led to warmer nights, but it might just be humidity. Certainly dry years have higher Tmax. This study provides welcome new insight.
@ Martin A
Yes, I think so too. In 100 years' time, there will still be a hard core of green activist sociopaths masquerading as journalists and telling us CAGW is still imminent, it's worse than we thought and the only cure is to place them in power over us so they can inflict misery on us all.
The analogy is Galileo versus the Church. Although Galileo was right and the Church was wrong, there are still Catholics today: more, in fact, than there were then. The Church's egregious error has not held it back too much. It's not widely listened to but it's still there and it's still wealthy.
The green movement is exactly similar, promising endless scares of which nothing ever comes except new scares.
Hmmm. Global warming theory says warming would be distributed unevenly and result in warmer nights, warmer winters, and warmer northern latitudes. I got into a dispute about this with ChiefIO 2 years ago.
Something strange about your blog software. I am posting this at 7:55 PM EST, just like my previous comment which ended up in the middle.
The use of the average in such a way reminds me of a story about 2 hunters watching a prize bird some distance away. The first hunter takes aim and fires, and the bullet passes an inch to the right of the bird. The second hunter takes aim and fires, the bullet passs an inch to the left of the bird. The hunters look at each other and one says"well, if we take the average we hit it dead on".
You probably would not be surprised to hear that using average like this is very common in assessing performance in the delivery of public sector services rather than gathering and using data to build a true picture of what actually happens and real service user experience.
I was unaware that climate science used the same approach. Thanks for this Philip, enjoyed it.
I have to assume that the reason daily averages were/are calculated by averaging the maximum and minimum temperatures was that originally they were measured by maximum-minimum thermometers and were recorded manually once a day, hopefully at a reasonably consistent time, and the thermometers reset. There are three problems with this methodology: firstly, as has been pointed out, they are point measurements and are not weighted in any way (as time integration does); secondly it is almost certain that they do not occur on the same day (or whatever fixed 24-hour period is used); and thirdly it is most unlikely that either measurement occurs at the same time each day. To be fair, before the advent of continuous recording methodologies it was the only method available. What is disturbing is that despite modern technology, the method is still apparently in use.
J4R
You and H L Mencken seem to be in agreement (as am I):
"The whole aim of practical politics is to keep the populace alarmed (and hence clamorous to be led to safety) by menacing it with an endless series of hobgoblins, all of them imaginary."
Gary Kerkin,
What prompted me to write this was that the recent BEST analysis persisted with the flawed min/max methodology even though much better time based datasets are available for the 60 year period Best covers.
I have avoided suggestions that this is orchestrated deliberate misinformation, but the almost complete absence of published papers using time based temperature data raises some serious questions.
Wow, this is great. I downloaded all of Environment Canada's data for some 1300 stations covering dates from 1900. I only got the daily TMin/Tmax and TMean. However, I did want to check if the average TMin+TMax/2 was in fact an average for the day or just the mean for the day. So I tested that by getting a few hourly temps for one day from their site, which is available. Sure enough, when I added all the hourly temps and divided by 24 I got a smaller average than the mean. Which would seem to fit what you have noted here. Looks like I'm going to have to check into this further.
I posted my results here http://cdnsurfacetemps.wordpress.com/2010/11/24/why-mean-temp-is-meaningless/
A thought-provoking essay. I will read it again, maybe more than once, to fully understand it. I had wondered about the "average" temperature being used for calculations, and it seemed to me that one should take hourly readings (or even continuous readings) and divide by the appropriate number of units.
My editor's eye picked out a couple of things that I would suggest looking at:
*"Manua Loa" (two times) is presumably "Mauna Loa".
*Part of one sentence reads ".....will be more likely to condensate out as clouds....." Water <condenses> out.
Best regards.
IanM
Philip
"absence of published papers using time based temperature data raises some serious questions"
Indeed, but maybe it's a 'wood for trees' thing - that's how it's always been done and few people have thought to query it. I would be interested to hear Judith Curry's take on this.
An important, pioneering study that needs to be published more widely and extended to other countries.
The discovery that taking the average of a maximum and minimum causes a bias is something I have suspected for a long while. It should go further. Night and day should be separate as they are diifferent, so maximum and minimum should have separate studies.
The importance of black carbon is valuable, but it is not the only possible anthropogenic effect. The earth is cooled by day by convection, and by evaporation of water . Human activity strives to interfere with both of these processes, with buildings, wind breaks, shelter belts, even windmills, greenhouses, reservoirs and flood and weather protection.
Glad to see some one "down under" coming up with similar results--and conclusions--that are all too apparent in USA data when analyzed from a more rigorous perspective than found in the simplistic methods of attention-grabbing climate "gurus." Understanding the variations of the diurnal cycle is absolutely fundamental to meaningful analysis of station data on climatic time-scales. Indeed, local insolation, even when recorded crudely as sunshine hours, is invariably coherent at multi-decadal scales with the temperature in well-kept records. After all, it's the sun, not the radiant absorbers, that drives the system.
Glad to see some one "down under" coming up with similar results--and conclusions--that are all too apparent in USA data when analyzed from a more rigorous perspective than found in the simplistic methods of attention-grabbing climate "gurus." Understanding the variations of the diurnal cycle is absolutely fundamental to meaningful analysis of station data on climatic time-scales. Indeed, local insolation, even when recorded crudely as sunshine hours, is invariably coherent at multi-decadal scales with the temperature in well-kept records. After all, it's the sun, not the radiant absorbers, that drives the climate system.
"GISS, HadCRUT and BEST wrongly assume the mean of Tmin+Tmax represents average daily (24 hour) temperatures."
The average daily temperature should provide a means of comparing specific energy readings over time. Using Tmin and Tmax (and assuming dry-bulb measurements) defeats that objective.
Usually the coldest part of 24 hours (day) is at about 3 a.m. and hottest is at 3 pm but I'd say its definitely not a symetrical pattern of heating and cooling afterall land is heated primarily by direct radiation and air is heated by land and sea at a different location to convect to land and there are days where the night is so warm that its hotter than daytime. Temperature measuring centres don't measure land temperature (very few measures of land temperature) - they measure air above land.
As an engineer, I don't understand why they don't use all the data- must be laziness.
hi, Philip Bradley - great post too by the way!
not sure why i didn't get your email, try again at jlowe@sportpunter.com, hopefully it wont get into the spam. And yes, I used monthly averages of 3-hourly data, and created the anomolies myself.