Out of Africa: A new paper by Christy on surface temperature issues

Mozal - Aluminium Smelter - Maputo, Mozambique -1998
Weather station, Mozal - Aluminum Smelter - Maputo, Mozambique - installed in 1998

More weather station photos from Africa here.

These stations shown and linked above are not GHCN stations as far as I can tell, but the siting was interesting nonetheless.

This new paper by John Christy, who works with Dr. Roy Spencer on the UAH dataset, points out that Tmin seems to have a signal in Africa where Tmax does not. Land use changes and aersols that affect the boundary layer at night are theorized and possible reasons. – Anthony

Surface Temperature Variations in East Africa and Possible Causes


Earth System Science Center, University of Alabama in Huntsville, Huntsville, Alabama


Surface temperatures have been observed in East Africa for more than 100 yr, but heretofore have not been subject to a rigorous climate analysis. To pursue this goal monthly averages of maximum (TMax), minimum (TMin), and mean (TMean) temperatures were obtained for Kenya and Tanzania from several sources. After the data were organized into time series for specific sites (60 in Kenya and 58 in Tanzania), the series were adjusted for break points and merged into individual gridcell squares of 1.258, 2.58, and 5.08.

Results for the most data-rich 58 cell, which includes Nairobi, Mount Kilimanjaro, and Mount Kenya, indicate that since 1905, and even recently, the trend of TMax is not significantly different from zero. However, TMin results suggest an accelerating temperature rise.

Uncertainty estimates indicate that the trend of the difference time series (TMax2 TMin) is significantly less than zero for 1946–2004, the period with the highest density of observations. This trend difference continues in the most recent period (1979–2004), in contrast with findings in recent periods for global datasets, which

generally have sparse coverage of East Africa.

The differences between TMax and TMin trends, especially recently, may reflect a response to complex changes in the boundary layer dynamics; TMax represents the significantly greater daytime vertical connection to the deep atmosphere, whereas TMin often represents only a shallow layer whose temperature is more dependent on the turbulent state than on the temperature aloft.

Because the turbulent state in the stable boundary layer is highly dependent on local land use and perhaps locally produced aerosols, the significant human development of the surface may be responsible for the rising TMin while having little impact on TMax in East Africa. This indicates that time series of TMax and TMin should

become separate variables in the study of long-term changes.

Some excerpts from the paper:


c. Possible causes for TMax and TMin differences The fact that the trends in the two temperature measurements (TMax and TMin) are likely significantly different encourages an examination of the causes for the warming of TMin and the significance of trends in TMin in the context of tracking global climate change.  Given a lack of detail on station siting and uncertainties in specifics on the boundary layer in East Africa, definitive reasons for the trends may not be available.

However, general aspects of boundary layer behavior may provide some guide for interpreting the trends.Thus, the following should be viewed as a context and hypothesis for the trend differences that deserve discussion and further attention.

From the conclusion:

For the 100-yr period from 1905 to 2004 in this grid cell, the trends were near zero for both TMax and TMin, but confidence in these results is low because of the relatively sparse data in the years before 1946. Beginning with 1946 and ending in 2004, near-zero trends were found for TMax. The TMin trends were more positive, and significantly so based on both measurement error and temporal sampling error. It is difficult to assess the

measurement error of these trends, but using the spread of 20 realizations in which the construction parameters

were varied, the range of 60.108C decade21 is plausible. The fact that the difference in trends in TMax and TMin continues, and in fact accelerates, in the period of 1979–2004 in East Africa may be important in interpreting the results of Vose et al. (2005).

While it is possible that East Africa difference trends are indeed different than that of the globe as provided by Vose et al. (2005), there is concern that the reduced number of stations in the 1979–2004 GHCN dataset may not be sampling many of the areas of the globe that are behaving like East Africa. Thus, it is important that the GHCN dataset be expanded to include more stations distributed around the globe.

The noticeable difference in trends of TMax and TMin implies that daytime and nighttime temperatures are responding differently to environmental factors. Changes in the surface characteristics and the boundary layer atmospheric constituents may be responsible for the relatively recent and rapid rise in TMin. There appears to be little change in East Africa’s TMax, and if TMax is a suitable proxy for climate changes affecting the deep atmosphere, there has been little impact in the past half-century.

The investigation of the surface temperature record as an indicator of human-induced climate change involves understanding the complex behavior of boundary layer processes (where surface temperatures are actually measured) and how temperatures within it are affected by the numerous changes that occur. This is an area of research open for considerable inquiry because it raises new questions concerning the types of

data indices now used to detect climate change.

At the least, the time series of both TMax and TMin should become separate variables to be studied for long-term changes.

Full paper is available here as a PDF 1.9 MB

0 0 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
July 18, 2009 8:30 am

I would say the reason of discrepancy is clear.
Coverage of Africa: http://www.climateaudit.org/wp-content/uploads/2008/05/ghcn_giss_250km_anom04_2008_2008_1951_1980.gif
Comparision of GISS vs UAH trend (by Bob Tisdale): http://i40.tinypic.com/1hb5sm.png
How to create desired tropical hotspot from non-existing data: http://www.ncdc.noaa.gov/img/climate/research/2009/jun/map-blended-mntp-200906-pg.gif

J. Bob
July 18, 2009 8:41 am

In looking at the latest report by NOAA on global temps., it seems central Asia is in a heat wave. Checking some of the stations, in that area, at http://www.rimfrost.no, many stations seem to be having normal temps. Ditto for Australia, and New Zealand.

July 18, 2009 8:52 am

What is Tmin? If this is nighttime temperatures, isn’t that the expected behavior when temperature rises? Most of the warming is at night and in cold places.

July 18, 2009 9:06 am

Increased Tmin is a good indicator of UHI. Based on GH theory, there should be not much warming in the tropics, but in polar regions where the prevalent GHG – water vapor – is much reduced compared to tropics.

July 18, 2009 9:08 am

MikeN: Tmin is the low temperature, which usually is at night. Tmin is expected to rise if the night sky is trapping heat, whether due to greenhouse gases or cloud cover. It also can rise if the thermometers are all on concrete which keeps warmer at night, or if a warm city grows up around the thermometer.

Stephen Wilde
July 18, 2009 9:21 am

I spy, a nice concrete plinth beneath and around the sensor.
When did that become a site standards requirement ?
And under a tropical sun too.
I thought it had to be above a grassy surface.

July 18, 2009 9:56 am

“At the least, the time series of both TMax and TMin should become separate variables to be studied for long-term changes.”

July 18, 2009 9:59 am

It would seem the Hadley and GISS anomalies for most of Africa are vapadata as calculated by scareware. The funny thing is, they can’t send the boys round to tell John Christy to get off their turf, since they don’t have any stations there.

July 18, 2009 10:07 am

Given the challenges we are having in accurately measuring US surface temperatures, I think all global surface temperature readings should receive some intense scrutiny. Looking at this NCDC Map:
it seems odd that the entirety of Africa and a good part of the oceans show positive anomalies. These seem to be areas where our measurement capabilities, especially in terms of accurate historical baselines, are most limited. And what the hell is going on in Siberia?
If the NCDC map was an election map I’d think it had been gerrymandered to maximize global temperatures, while minimizing the impact of the Surfacestations report. The temperature distribution just looks too calculated, like it is the work of man, versus nature. I think Surfacestations.org should go global…

July 18, 2009 10:07 am

J. Bob (08:41:16) :
In looking at the latest report by NOAA on global temps., it seems central Asia is in a heat wave. Checking some of the stations, in that area, at http://www.rimfrost.no, many stations seem to be having normal temps.
The rimfrost data appear to be from NOAA too. Am I wrong?

July 18, 2009 10:50 am

SIberia, like Alaska, have a summer roughly once every decade.
In dry, smiearid places, like Central Asia and East Africa, there isn’t much water vapor to act with the CO2. So, by AGW theory, no global warming should be taking place. Tmin should have a very high separation from Tmax, not a closing gap.
It’s the concrete for the station we see.
The rest of it is fabricated station polluting with the aim of “correcting” all good stations to the rotten egg station.
Just like they did for June in my area.
I know what the condition are here, it’s where I live.

July 18, 2009 11:53 am

rbateman (10:50:43) :
“I know what the condition are here, it’s where I live.”
I noticed that myself. This is why I have been asking if GISS’ outliers are temperatures that are eliminated. We have had a few ‘outliers’ through June and July. The high temp in Ohio for a date in the middle of July? 70ºF. Low is 52. Would these be dropped as outliers? I don’t know for sure, but they are the real temperatures I have today.

Bill Illis
July 18, 2009 12:11 pm

I guess you could call this a “test” of the accuracy of GISS, Hadcrut and the NCDC (and potentially the adjustments done to the temperature record by these agencies).
This test says the trend from these agencies from 1905 and from 1979 are more than twice too high (although the lower trend from 1946 seems to be good).
In a way, it is also a test of whether the lower trends from the satellite measurements of the lower troposphere are accurate as well. Seems to be a passing grade for this test.
I think there needs to be several more of these papers covering other areas for additional confirmation. Of course, this will need to be done in double-blind / objective way like medical studies use so that confirmation bias does not creep in (ie. someone other than Phil Jones, James Hansen, Thomas Karl and even John Christy needs to do them).

Allan M
July 18, 2009 12:57 pm

J. Bob (08:41:16) :
“In looking at the latest report by NOAA on global temps., it seems central Asia is in a heat wave. Checking some of the stations, in that area, at http://www.rimfrost.no, many stations seem to be having normal temps. Ditto for Australia, and New Zealand.
I wonder when a top manager in this organisation last used the phrase “Bring me the figures to prove…”

Richard Hill
July 18, 2009 1:26 pm

earlier a commenter said Tmin was at night. Not always. Sometimes it is after first light. There are interesting results from a study of resords from Mount Gambier in South Australia by Jonathan Lowe.

July 18, 2009 1:31 pm

J. Bob (08:41:16) : Temperatures here in New Zealand are indeed about normal for the time of year. For the region where I live (Waikato, central North Island), June was the second coldest on record (coldest was 1972) – mean temperature 7C, which is 2 degrees below normal. More snow today for southern and alpine parts of the country. The skifields are having yet another bumper year, after being told back in the 90s that they had no future.

July 18, 2009 1:31 pm

Bill Illis (12:11:33) : So far Christy has painstakingly put together studies likes this for: ALbama, California, East Africa…More needs to be done but it is a LOT of work just collecting the data. Recall that Christy has stated that he often collects ten times as much data as the popular datasets use.
If only more people would be interested in doing this research.
MikeN (08:52:52) : If this WERE true, then you would tend to still see significant warming in Tmax, which is connected better to the upper air. And of course there is a great danger of mixing up the signals.
What helps here is that models tells us the expected change in the diurnal temperature range from CO2 AGW-it turns out to be much LESS change than actually observed:
Seems like a landuse signal to me!
BTW and geniuses out there able to explain what the difference between Tavg and Tmean is? I’m confused by that…

Douglas DC
July 18, 2009 1:33 pm

I remember when I helped set up a Sawrs station in SE Washington-Richland. We had a
heck of a time finding an appropriate area for the Stevenson Screen! you see the old
Hanford Airport was mostly asphalt and concrete. We eneded up putting it on a small sand dune,as there was only sand and sage there at the time.
SAWRS means supplemental aviation reporting station-I had to take the same test as the Weather service folks to do this…
It appears that no one cares what is automatically reported…

July 18, 2009 2:11 pm

Richard Hill
As a keen wildfowler with ageing bones, out on the merse (marshes) of the Moray Firth (Scotland) in the depths of winter and well before the dawn, I can verify that the coldest time is very often about half an hour after first light, and that on dawns when no frost is present, frost will often
form on the ground at that time.

July 18, 2009 4:24 pm

Night time temperature inversions do shield the lower boundary layer from the “dynamics” aloft. The strength of the inversion for those areas in the tropics is dependent upon the absolute humidity. In a Maritime Tropical (Mt) airmass, there is less radiative cooling as compared to a Continental Tropical (Ct) airmass. Desert inversions can be fairly extreme with as much as 40 deg F of radiative cooling at night. In the more maritime regions of the tropics, the radiative cooling can be as little as 10 deg F. As a consequence, the Maritime Tropical regions see higher nocturnal winds and more turbulent mixing.
An interesting thing to examine in relation to East Africa is subtle changes in the Southern Oscillation over the last 40-60 years that may have affected humidity there. Obviously land changes also would have a more direct impact upon humidity in East Africa. Mt Kilomanjaro has seen its high level glaciers recede not because of AGW, but because of less precipitation in the higher altitudes. Could changes in farming in East Africa have lowered the abolute humidity there, and thus lead to lower nighttime lows?

July 18, 2009 4:40 pm

On average Tmin occurs after dawn when solar heating overcomes the effect of radiative cooling. This is true for all lattitudes. The time of Tmin after dawn is longest at high lattitudes in winter.
The Indian monsoon is late this year. This monsoon is caused by a summer heat low in central Asia. The monsoon typically reaches central Asia around this time of year.
So rather paradoxically, unusually high temperatures in central Asia are due to a lack of monsoon clouds and rain, which in turn are due to an unusually cool summer to date.

July 18, 2009 4:55 pm

This is an area of research open for considerable inquiry because it raises new questions concerning the types of
data indices now used to detect climate change.

I suspect they know what the problem is, but can’t come out and say it, because it is a rather large nail in the coffin of AGW.
Increasing Tmin isn’t caused by warmer nights. It is caused by increased early morning sunlight. Thus reducing the period after dawn when radiative cooling exceeds warming from the sun.
The cause may be reduced aerosols, reduced low level cloud or some other cause. But it can’t possibly be GHGs.
* There may be a UHI signal affecting Tmin as well.

Paul Vaughan
July 18, 2009 6:02 pm

*Awesome* story – with key focus *exactly* where it needs to be. I’m not sure if Christy is on-the-mark with (all of) his speculation (for example I see no mention of SSTs and coastal-continental gradients in Anthony’s excerpts), but he has my full support for raising the issue of TMin, TMax, TMean=(TMax+TMin)/2, & TRange=TMax-TMin. I would also throw in XTMax, XTMin, & XTR=XTMax-XTMin, where X denotes eXtreme-monthly. I spent a considerable amount of time looking into this issue last year (including the sparse literature on the topic) – and no matter how much I look empirically, I keep noticing more things [that do not seem to be on the mainstream radar]. Time to read Christy’s paper to glean more insights…

J. Bob
July 18, 2009 6:19 pm

WAZUP – The best I can tell is that they get them from a variety of sources. So far I have been using them to get long term temperature data ( Uppsala-LAN, Paris, Basel-Binn, etc. ), along with the Central England data. I’m trying to put the 1850 onward HadCrut3 data into a better perspective. The Central England data is good, but only one source.
Allen M. Check out the indicators from Australia on the Mar-May temp anomaly graph from NOAA. Go to Rimfrost , “raw data” and look at Perth, Alice Springs and Melbourne. The “raw data” looks more normal then high.

July 18, 2009 7:53 pm

Just an update from the weather is not climate, and we may not count it anyway department. Didn’t even hit 70 today. Makes me wonder if the low of 52 could drop into the 40s. Also makes me wonder if that will even make it into the official climate record.

July 18, 2009 7:58 pm

Paul Vaughan (18:02:44) :
I endorse your comment.
T Mean tells us very little about the thermal dynamics that drive biological systems. Consider plant growth, e.g. the grape vine. Temperatures below 10°C in the presence of light can bring about photo-inhibition whereby the leaves will not photosynthesise even when temperatures become more favourable. Carbohydrate generation is slight at temperatures below 15°C, peaks at about 25°C and is gradually inhibited as temperatures rise above 30°C.
The Rolls Royce situation for data analaysis is spot readings every 20 minutes. A passable job can be done in predicting plant performance with hourly data.
T Mean calculated as (TMax +T Min)/2 can be two to three degrees either side of a true average of 24 hourly readings. I suspect that in twenty years or less, peer reviewers will be throwing out papers that rely on T Mean on the basis that this statistic provides no indication of the thermal conditions driving biological systems.
Christy’s paper points us in the right direction.

Paul Vaughan
July 19, 2009 2:29 am

Philip_B (16:55:26) “I suspect they know what the problem is, but can’t come out and say it, because it is a rather large nail in the coffin of AGW.”
When people ask me what I like about climate research, I tell them, “Everything: The psychology, the politics, the complexity of nature.” People intuitively understand that the complexity extends into the stakeholder dimension. Interesting comment Philip_B.
– –
Re: Erl Happ (19:58:09)
The authorities in my region had to backtrack after they predicted TMin was going to overtake TMax — D’Oh! I asked around about the analyst; one reply came back, “Yeah, she’s not too bright.”

Bill D
July 19, 2009 5:03 am

I have not studied the science of diurnal temperature change. However, I had assumed that GHG would lead to less radiative cooling at night and warmer dailly minimal temperatures. Some of you are saying that cooler minimal temperatures could not be due to GHG. What is the basis for predicting tht GHG would have a greater effect on the maximum, rather than the minimum temperature?

July 19, 2009 6:04 am

At last, someone realizing that “averages hide more than they reveal – emsmith” and that min and max need to be looked at separately to have any clue what is really going on. I love it.

July 20, 2009 3:37 am

For more information on Southern Africa, 100 year + records should exist for South Africa and Zimbabwe. The more inquiring can follow up:
“Meteorological Services in Zimbabwe were initially established as an Office under the Ministry of Agriculture through a Parliamentary decree in 1925. Earlier on, from 1897 to 1920, the collection of meteorological statistics was a part-time duty the “Statistic” and later “Agricultural Engineer”. The original concept was the establishment of about a dozen climatological stations equipped to record pressure, temperature, humidity and rainfall. The duty of observation was to fall to Civil Commissioners. The first stations to be established were at the then Salisbury (Harare), Bulawayo, Gwelo (Gweru), Fort Victoria (Masvingo), Tuli, Belingwe (Mberengwa), Umtali (Mutare), Hope Fountain, Rusape and Inyanga (Nyanga). In Bulawayo the instruments were issued to St George’s School and later transferred to the Lawley Road site in 1903 with the arrival of Father Goetz. At about the same time, a policy was adopted of supplying rain gauges to Police and other outposts thereby increasing the number in use considerably.
Daily rainfall reports, synoptic charts and experimental weather forecasts started in 1922. Regular weather forecasts for the country started in 1924 through the press and by telegraph to all Post Offices. Scheduled aviation services commenced in January 1932 resulting in the opening of new Meteorological Offices at or near aerodromes at Bulawayo and Salisbury (Harare) in July 1936. At this stage, most of the 30-odd synoptic stations were at Police Stations issuing one to three reports per day. Following the commissioning of Lake Kariba in the late 1950s, a network of Seismic monitoring stations was established in Bulawayo, Karoi, Mt Darwin and Chiredzi.
Currently the Zimbabwe Meteorological Services has a network of about 64 full time and part-time meteorological observing stations around the country and more than one thousand volunteer rainfall-observing sites.
On 10 September 2003, the Meteorological Services Bill passed through parliament and got the Presidential Assent in the same month. The Meteorological Services Act (chapter 13:21) established the Meteorological Services Department, which had always existed under the Ministry of Transport and Communications before the Act as a Statutory Body with a legal framework to provide certain products and services on a Cost Recovery Basis and Free or Tax-payer funded Services.”
I recall the Mutare station was sited at the fire station and had a couple of Stevenson screens although I can’t vouch for it’s current state! There would be very little UHI effect due to the lack of recent (20 years) development. ( lat=-18.9718258998, lon=32.6654821465).

Ron de Haan
July 20, 2009 9:28 am
%d bloggers like this:
Verified by MonsterInsights