Skip to content

African Temperatures According To GISS

November 19, 2011
tags: ,

 

image

 

According to GISS, Africa is one of the fastest warming places on the planet. But how reliable are their temperature records?

The first thing to notice is that the anomalies are based on the period 1950-1980, but if we look at a couple of long running South African stations, we immediately see the familiar story of a much cooler interlude between about 1940 and 1980.

 

 

image

 

image

 

So it would hardly be surprising to see that current temperatures are warmer than 1950-1980. Nevertheless we should surely be able to trust the temperature record in recent years. After all there are 538 African GHCN stations listed in the GISSTEMP record, so there ought to be a pretty good coverage across the continent. However, if you thought that you would be wrong.

I have analysed all of these stations to see which ones fulfil the following criteria :-

1) A  record from 1941 or earlier up to 2011. ( There are many stations with gaps in their record, but I have not excluded these).

2) No more than 20% missing data between 2001 and 2011.

Out of 538 stations, only 40 stations actually meet these two conditions. Furthermore, of these , 39 are urban / small towns, leaving just 1 rural station. Bear in mind, that according to GISS methodology, urban stations are only included in the GISS temperature analysis if there exists a nearby rural station so as to make adjustments for UHI. Therefore it would appear that the temperature analysis for most of Africa is based either on stations with a record not long enough to give proper long term trends or on stations with large gaps in their records in the last decade (or both!)

For interest the only rural station to meet the criteria is Calvinia, a small town in South Africa with a population of 8146. The temperature graph for Calvinia shows no apparent long term trend.

 

image

 

What of the rest?

So what about the other 498 stations? I wondered if I was being too selective in omitting stations which were only set up after 1941. To check this out, I carried out a fuller analysis on 50 stations from one page of the GISS subset, which in their own words “ contribute to the final products”. (This particular batch seems to centre around the area from Angola/Congo/Gabon in the west across to Kenya in the east – in other words a fairly typical swathe across Central Africa).

The analysis showed that :-

1) There were 29 urban/small towns and 21 rural stations.

2) Of the 21 rural stations, only 7 had records up to 2011. None of the others extended past 1991.

3) Of the 7 current stations, 6 had no records prior to 1987.

4) Amongst these 6 stations, at least 67% of the monthly temperature readings were missing for the period 2001-10. The missing records were even worse for 2006-10 with at least 75% missing in all cases.

5) To cap it all, the other station (the only rural station with records prior to 1987) is Garissa in Kenya. Although GISS list this as “rural”, it is in fact a large town with a population of 65000, and the weather station is at an airport. Even here, a full 50% of monthly records between 2006 and 2010 are missing.

It is abundantly clear that in most of Africa, the temperature record simply does not exist to allow continent wide analysis. Temperatures may have risen there in the last 30 years, but that should not allow GISS to make up numbers, if the actual data does not exist to prove it.

7 Comments
  1. Mike Davis permalink
    November 19, 2011 6:34 pm

    The simple answer is to discard all current surface temperature records as being corrupted beyond value. Then continue archiving satellite records while attempting to find errors in the methods used and when a minimum of two hundred years of “Good” data are available they can try to better understand weather patterns. 200 years is probably not even enough to display all the variability in regional climate!
    GHCN is bad enough but using that as an input for GISS Mixomatic, the result can only be finer shredded garbage with a side order of garbage!

  2. Douglas Hoyt permalink
    November 20, 2011 4:02 pm

    Perhaps looking at the satellite records (UAH or RSS) over Africa and comparing them to the surface observations would be productive.

  3. November 20, 2011 7:46 pm

    Thanks to Douglas’s tip I have checked and compared satellite temperatures with the GHCN record. (GISS figures are not available in Latitude bands, but are based on the GHCN network anyway).

    Taking the box [10W – 40E longitude] by [0 – 20N latitude] , and using the period 1979-2008, we get the following results :-

    MSU satellite – Trend increase of 0.14C

    GHCN – Trend increase of 0.68C

    Remember this box represents a large chunk of Africa where GISS show the greatest warming (from Kenya up through Sudan and across to Niger and Cameroon in the west).

    Food for thought!!

    I’ll run a few more boxes tomorrow and post up the results.

  4. Mike Davis permalink
    November 21, 2011 9:38 pm

    From GISS:
    Q. If the reported SATs are not the true SATs, why are they still useful ?
    A. The reported temperature is truly meaningful only to a person who happens to visit the weather station at the precise moment when the reported temperature is measured, in other words, to nobody. However, in addition to the SAT the reports usually also mention whether the current temperature is unusually high or unusually low, how much it differs from the normal temperature, and that information (the anomaly) is meaningful for the whole region. Also, if we hear a temperature (say 70°F), we instinctively translate it into hot or cold, but our translation key depends on the season and region, the same temperature may be ‘hot’ in winter and ‘cold’ in July, since by ‘hot’ we always mean ‘hotter than normal’, i.e. we all translate absolute temperatures automatically into anomalies whether we are aware of it or not.
    http://data.giss.nasa.gov/gistemp/abs_temp.html

    I translate that to say they make stuff up!

Comments are closed.

%d bloggers like this: