39. Namibia - temperature trends

When it comes to temperature data Namibia is not much better than Botswana. In total, there have only ever been 34 weather stations in Namibia (compared to 20 in Botswana). Of these only five are medium stations with more than 480 months of data, although nine have more than 450 months of data. This includes only two with more than 900 months of data and only four with any data before 1960. There is only one long station with over 1200 months of data (Windhoek), and only Walfisch Bay has any data before 1900. Unfortunately, both of these stations have portions of data that are clearly erroneous. The most likely explanation is that these data segments have either been incorrectly converted from Fahrenheit to Celsius when they were already in Celsius, or were not converted when they should have been.

 

Fig. 39.1: The temperature trend for Windhoek since 1911 according to Berkeley Earth (BE).

For the case of the Windhoek data, it is the data for the period 1911-1920 that is in question, as illustrated in Fig. 39.1 above. There is clearly an offset of more than 25 °C between the data before 1920 and the data after that date. What appears to have happened here is that some data that was correctly already in Celsius was assumed incorrectly to be in Fahrenheit. So when an unwanted correction of (x - 32)÷1.8 was applied to the data, the data was offset in a negative direction. Reversing this correction appears to remove the offset, as illustrated in Fig. 39.2 below.

Fig. 39.2: The temperature trend for Windhoek since 1911. The best fit is applied to the interval 1944-2001 and has a gradient of +2.21 ± 0.26 °C per century. The monthly temperature changes are defined relative to the 1971-1990 monthly averages.

The resulting temperature trend for Windhoek is strongly positive. However, the opposite is true for the only other significant temperature record that pre-dates 1920. This is the temperature record from Walfisch Bay, which is located at sea level on the coast about 200 km to the west of Windhoek (which is at altitude).

Fig. 39.3: The temperature trend for Walfisch Bay since 1885 according to Berkeley Earth (BE).

Like the Windhoek temperature record, the one for Walfisch Bay contains a significant amount of erroneous data (see Fig. 39.3 above). In this case, most of the questionable data have values that are forty degrees too large. This is almost certainly because these data values are recorded in Fahrenheit not Celsius. If we apply the appropriate correction, then the resulting data exhibits a strong negative trend as shown in Fig. 39.4 below. There are, however, ten data readings in Fig. 39.3 (from October 1941 to July 1942) where the offset is only about twenty degrees Celsius. The reason for this error is harder to ascertain and so these data points have been excluded from Fig. 39.4.

Fig. 39.4: The temperature trend for Walfisch Bay since 1885. The best fit is applied to all the data and has a negative gradient of -2.23 ± 0.16 °C per century. The monthly temperature changes are defined relative to the 1971-1990 monthly averages.

These are not the only station records with a significant amount of data before 1970, though. The graph below (Fig. 39.5) indicates that there are at least three other stations with such data. One is at J. G. H. Van Der Wath Airport (Berkeley Earth ID: 156958) which is at altitude near Keetmanshoop, while the other two are at Lüderitz (Berkeley Earth ID: 139074 and 156957) on the coast and about 200 km to the west of Keetmanshoop.

Fig. 39.5: The number of sets of station data included each month in the temperature trend for Namibia when the MRT interval is 1971-1990.

If we look at the data for J. G. H. Van Der Wath Airport (Berkeley Earth ID: 156958) we see that it exhibits a weak negative trend before 1980 of -0.43 ± 0.41 °C per century, but then shows a strong positive trend after 1980 of 3.3 ± 0.7 °C per century (see Fig. 39.6 below). Given the proximity of this station to both South Africa and Botswana, it is perhaps not surprising that the temperature trend resembles each of the trends seen in both those countries (see Fig. 37.2 and Fig. 38.1).

Fig. 39.6: The temperature trend for J. G. H. Van Der Wath Airport since 1933. The best fit is applied to the interval 1934-1978 and has a negative gradient of -0.43 ± 0.41 °C per century. The monthly temperature changes are defined relative to the 1971-1990 monthly averages.

The only other temperature data in Namibia from before 1970 comes from two stations at Lüderitz, a small coastal town (population: 12,500) in the south of the country on a similar latitude to Keetmanshoop. The two stations have different time frames that overlap between 1973 and 1986. The earliest data comes from Lüderitz Bay (Berkeley Earth ID: 156957) and extends from 1941 to 1986. The later data is for Lüderitz Diaz Point (Berkeley Earth ID: 139074) and extends from 1973 to 2013.

 

Fig. 39.7: The temperature trend for two stations in Lüderitz since 1941. The anomalies up to and including 1980 are for Lüderitz Bay (Berkeley Earth ID: 156957) while those from 1981 onwards are for Lüderitz Diaz Point (Berkeley Earth ID: 139074). The best fit is applied to the interval 1941-1995 and has a positive gradient of 0.27 ± 0.20 °C per century. The monthly temperature changes are defined relative to the 1971-1990 monthly averages.

 

If we combine the anomalies from the two Lüderitz stations we get the temperature trend shown above in Fig. 39.7. This has a very weak upward trend before 1990 which is followed by a sudden jump in temperature just before the year 2000. This temperature jump is similar in size to those seen after 1980 in data from Botswana (see Fig. 38.3), but it occurs about ten years later.  On the one hand this suggests that the temperature jump is a real phenomenon as it is seen in multiple station records, not just in Namibia, but also in Botswana, and to a lesser extent in South Africa. However, the variation in its timing across the different countries is a concern and means that we cannot completely trust its authenticity. What we cannot do is just ignore it.

Fig. 39.8: The temperature trend for Namibia since 1885. The best fit is applied to the interval 1944-2001 and has a positive gradient of +2.28 ± 0.17 °C per century. The monthly temperature changes are defined relative to the 1971-1990 monthly averages.

In addition to the five stations mentioned so far, there are another four stations with more than 450 months of data, as indicated in Fig. 39.5 above. Combining and averaging the anomalies from these nine stations yields the overall temperature trend shown in Fig. 39.8 above. This trend shares many features with those seen for Botswana (see Fig. 38.3) and South Africa (see Fig. 37.2). However, it also hints at the possibility of higher temperatures in the 19th century that would contradict the accepted conventional view of global warming in the 20th century. What is clear, though, is that once again there are significant differences between the actual raw data presented here and the temperature trend constructed by Berkeley Earth (see Fig. 39.9 below). The two most obvious differences are the temperature rises after 1980 and before 1920, both of which have been adjusted down by Berkeley Earth.

Fig. 39.9: The temperature trend for Namibia since 1860 according to Berkeley Earth.

Conclusions

The lack of high quality data makes definitive conclusions for the temperature trend in Namibia difficult. However, by comparing the Namibia data with that from neighbouring countries we can detect commonalities that allow some conclusions to be drawn.

1. For the majority of the 20th century little or no warming has occurred in Namibia, just as the same can be said for Botswana and South Africa.
2. There appears to have been a significant warming period after 1990. However, it is unclear what the cause of this is, and how long term it might be. It appears to be too abrupt and too large to be solely due to carbon dioxide.
   
3. There is weak evidence that 19th century temperatures in Namibia may have been much higher than those in the 20th century, just as we have seen previously in South America, Australia, New Zealand and the South Pacific. While the data before 1900 in each of these regions is scarce, the fact that there appears to be a consistent pattern across these regions for the temperature data that does exist would imply that the data is probably sound.