Skip to content

How Reliable Are Temperature Records In Africa?

April 11, 2014

By Paul Homewood




According to GISS, Africa is one of the fastest warming places on Earth, with last year being anywhere between 0.5C and 2C warmer than the 1951-80 baseline.

But just how reliable is the data there? Let’s take a closer look at East Africa, and, in particular, Tanzania.


There are just four GHCN stations there currently reporting, i.e. the ones used by GISS. These are listed below.


Location Population
x 1000
Airport Brightness
Date Range
Mwanza 700   10 1923-2014
Tabora 227 Y 0 1893-2014
Dar es Salaam 4364 Y 13 1949-2014
Songea 130 Y 0 1949-2014


So straightaway we find that there are no rural sites, using population as an indicator, and three of the sites are airport based.

To allow for UHI effect, GISS rely on Brightness to determine if a site is rural or not, with Dark = Rural. The Brightness Index runs from 0 = Dark to 256 = Bright. The cut off point for rural is <11.

The trend of the urban stations is then adjusted to that of the nearest rural stations.

This sounds fine in theory, but there is a fundamental problem with this approach in Tanzania – both Tabora and Songea show up as dark, but are airport sites, and therefore highly likely to have a UHI bias.

The fact that they are “dark” simply indicates that they do not operate at night, and not that they are rural.


But it gets worse! Mwanza is the second largest city in Tanzania, with a population, according to Wiki, of over 700,000. Yet, because, it is relatively dark at night, it is counted as rural.


Mwanza city Centre


This tendency for large cities in Africa to be dark at night seems to be a widespread one, if the map, used by GISS, below is anything to go by. Even Dar es Salaam, with a population of over 4 million, only has a brightness index of 13!






And just how much UHI effect is there likely to be? The important thing is to look at the change over a period of time, rather than the absolute amount, and the change in these cities is likely to be huge over the last 50 years or so.

Take a look, for instance, at how Dar es Salaam’s population has grown over the last century. Both in terms of population and infrastructure, the city would be unrecognisable from even 40 years ago, and the same population explosion has occurred across much of the continent.




So how does all this work out in practice? Let’s start with the GISS temperature record at Songea.





And what about Tabora?




How much of these temperature increases have been due to UHI bias? We don’t know what conditions were like at Songea or Tabora back, say, in the 1970’s, and at both sites there are big gaps and suspicious sudden jumps.

It seems highly likely though that with economic development, both nationally and locally, the airports will be busier than before and more developed. For instance, Tabora’s population has increased from 127,000 in 2002 to 227,000 in 2012.


Then, we have Mwanza!




Here we see a common plot in Africa – basically a hotch potch of incomplete records, that no serious scientist would dream about using – but it would probably be good enough for climate scientists to use to “prove” that it had warmed up since 1970.


If there really was no UHI bias at the above three sites, there would be an adjustment at Dar es Salaam, where there most certainly will be, but is there?

The first graph is of the actual temperatures, and the second after the GISS homogenisation adjustment for UHI.






The early part of the record is now omitted after homogenisation, which makes direct comparison difficult, but the only adjustment for UHI is 0.2C between the start of the record in 1949 and 1959. Since 1960, there has been no adjustment at all.

I repeat, despite a 30-fold increase in population at Dar es Salaam, and siting at a major international airport handling some 2 million passengers annually, the raw temperatures are not adjusted for UHI.

And the reason? Quite simply that the so called “rural sites”, against which Dar es Salaam is compared, are also affected by exactly the same sort of UHI bias. As the GISS algorithm assumes they are rural sites, with no UHI, when it finds a similar trend at Dar es Salaam it decides there is no UHI there either.



How typical is the case of Tanzania? Let’s look at two other East African states, Kenya and Uganda.

There are no stations currently reporting in Uganda, not a sausage. The last one, at Entebbe, stopped reporting in 1975. Across the border in Kenya, we find just two, Lodwar and Mandera, shown below.

Both are small towns with populations of 48,000 and 30,000 respectively, and both are sited at airports. But, more importantly, their records are so intermittent to be worthless.






With such big gaps in the record, there is no way of knowing whether the thermometers are even sited in the same place.

Furthermore, even recent records are woefully inadequate. For instance, at Lodwar there were only three months with temperature records last year.



We find that across three countries, there are just six current sites, all at airports or in cities. Of these, arguably, only three have full enough records to be usable.

One of these is in Dar es Salaam, where the UHI effect must have increased phenomenally over the last 50 years, but where there has been no allowance for UHI during that time.



Dar es Salaam Airport


The other two exhibit similar trends to Dar es Salaam, implying UHI effects there as well.

As GISS use smoothing over a 1200km radius, the influence of this very small number of stations, may be amplified way beyond their own little region. As it seems likely that many other parts of Africa are affected in a similar way to East Africa, temperatures over a large part of the continent may be substantially overstated.

There are probably so few reliable stations in most of Africa that we simply do not know what temperature trends have been, even in the last few decades. To pretend that we do is not science.

  1. Joe Public permalink
    April 11, 2014 5:53 pm

    Dar es Salaam: Maybe that’s why the Beeb blithely repeated the allegations of seaweed collectors, when two weeks ago they stated that climate change was killing seaweed across the strait in Zanzibar.

  2. April 11, 2014 5:55 pm

    Which adds more uncertainty to the global temperature record.

    It would be interesting to know what proportion of recording stations globally are at airports.

  3. April 11, 2014 6:33 pm

    Nice observation. Keep it up. –AGF

  4. A C Osborn permalink
    April 11, 2014 7:03 pm

    As you say “Jim it is Science but not as we know it”.
    But of course it suits their purpose.

  5. April 11, 2014 7:38 pm

    The ludicrous claims about Africa make more sense now.

  6. April 11, 2014 7:38 pm

    Reblogged this on CraigM350.

  7. Gary H permalink
    April 11, 2014 11:15 pm

    Was just reading the crapola from Desmond Tutu, here: . .

    . . went to the GISS site to look at some scattered plots.

    The first thing I noticed is that there are few records here available prior to the onset of the supposed onset of AGW, in the mid 1900’s.

  8. tom0mason permalink
    April 12, 2014 12:44 am

    Is there any temperature data, from any satellite that over flies this region, that shows nigh time temperatures ? MODIS ?
    Such data should highlight obvious UHIs and other non-solar originating heat sources.

  9. GregO permalink
    April 12, 2014 5:13 am

    Africa, yet another of the fastest warming places on earth. Here we see again the fallacy of sparse data with no (I would add no possible) uncertainty analysis at the temperature resolutions they are positing.

    Anyone, who has measured anything, knows there is a tolerance of uncertainty integral to any measurement of anything. So we are going to measure the temperature of Africa, the continent of Africa. And it is Warming. Or cooling. Based upon this data set, accompanied by precisely what uncertainty analysis? Please.

    It is an insult to any intelligent, cogent, individual trained in any science or even any precise craft trade, that we can know fractions of a degree delta T from such scattered, sparse, for what of a better term, crude measurements.

    If indeed Man-Made Global Warming is the crisis of our times, shouldn’t we be paying a bit more attention to the actual data collection and data quality control and perhaps even placing a few more thermometers?

    Just wondering.

  10. April 12, 2014 5:30 am

    The situation actually worse than this blog suggests for two reasons:

    1. The urban areas are underbounded, which means that the population overspills into surrounding census areas, which means the actual urban population is greater than listed.

    2. Maintenance is poor in most developing countries and worse in many parts of Africa because civil aviation administrations lack budgets, trained staff and effective managers.

  11. c777 permalink
    April 12, 2014 10:29 am

    Was this deliberate?

  12. April 12, 2014 2:04 pm

    Thanks, Paul. Another interesting article bringing some truth out of the purposefully created confusion.
    I would like to show the “2013 Lower Troposphere Anomaly Map”, by Dr. John Christy and Dr. Roy Spencer, Principal Research Scientists at the University of Alabama in Huntsville – UAH) , at
    I cannot understand the credibility that thermometer temperatures records still seem to have. The thermometers have been proved unreliable and warm-biased for reasons of placement and urban heat island effects.

Comments are closed.

%d bloggers like this: