The striking thing about the temperature data for Sri Lanka is the lack of medium length weather station records (i.e. those with over 480 months of data). It only has two. Yet the country has ten long weather station records with over 1200 months of data, which considering its area is fifty times less than that of India, means it has over twelve times as many long stations per square kilometre as does India. Even if we combine the number of long and medium stations, Sri Lanka still has six times as many per square kilometre as India. Unfortunately, with only twelve temperature records in total (for a full list see here), any average temperature trend derived from them could potentially be less accurate than that calculated for India in the previous post, despite the higher station density. This is because, as I pointed out in a previous post, in order for regression towards the mean to be able to eradicate random data errors that are present in all temperature series, it appears that there needs to be at least twenty datasets in the average. That said, the temperature trend for Sri Lanka is still informative, not least because of its similarities to those of India and Pakistan.
Fig. 73.1: The (approximate) locations of the long and medium temperature records in Sri Lanka. Those stations with a high warming trend between 1876 and 1975 are marked in red while those with cooling or stable trends are marked in blue. Those denoted with squares are long stations with over 1200 months of data.
The locations of the long and medium stations in Sri Lanka are shown in Fig. 73.1 above. Their geographical spread is fairly even, although most are on the coast. Nevertheless, this suggests any simple average of their temperature anomalies should yield a fairly accurate approximation to the true mean temperature anomaly for the region. It can also be seen that the vast majority of stations have stable or cooling trends for the 100 years up to 1975.
Just as for previous regional analyses, the monthly temperature anomalies for each station were calculated by subtracting the monthly reference temperature (MRT) for the relevant month from the unadjusted mean temperature for that month. The MRTs are different and specific to each station, but were always calculated using data from the same time period (in this case 1951-1980) for each station using the method described previously in Post 47. This ensures that the baseline reference temperature for each station is consistent so that all temperature changes over time are measured relative to the same point in time.
Fig. 73.2: The temperature trend for Sri Lanka based on an average of anomalies from all long and medium stations. The best fit is applied to the monthly mean data from 1876 to 1975 and has a positive gradient of +0.11 ± 0.03 °C per century. The monthly temperature changes are defined relative to the 1951-1980 monthly averages.
If we calculate the monthly temperature anomalies for the twelve stations in Fig. 73.1 and then average them, the resulting times series exhibits a trend that is slightly warming, as shown in Fig. 72.2 above. However, this warming occurs in two distinct phases. Before 1975 the warming is negligible with temperatures rising barely more than 0.1°C in over a century. This is much less than the natural variation in the 5-year average temperature. Then, after 1975, the temperature appears to jump abruptly by about 0.4°C. Similar jumps have been seen in the trends for South Africa, Botswana, Europe, India and Pakistan.
I have yet to determine what is causing these jumps. Are they natural or are they man-made? Are they the result of climate changes, economic changes, or changes to data collection methods? In the case of Sri Lanka, one possible cause is the civil war that raged from 1983 to 2009. This may be why there is a large dip between 1980 and 2010 in the station frequency plot in Fig. 73.3 below. And as so many of the station records are consequently discontinuous between 1983 to 2009, this may have led to bad data being generated after 1983 from station moves and equipment changes. That, though, is pure speculation.
Fig. 73.3: The number of station records included each month in the mean temperature trend for Sri Lanka in Fig. 73.2.
Of course the general temperature stability seen in Fig. 73.2 before 1975, and the temperature jump that occurs after 1975, are not what is claimed by climate scientists. Below in Fig. 73.4 is the trend for Sri Lanka according to Berkeley Earth which shows a more or less continuous warming trend since 1830 of about 1.5°C in total. Clearly this is at odds with the trend based on the raw data in Fig. 73.2, and the reasons for the differences are the same as they were for Pakistan. They are mainly due to the adjustments made to the original raw data by Berkeley Earth through the use of breakpoints and homogenization.
Fig. 73.4: The temperature trend for Sri Lanka since 1800 according to Berkeley Earth.
The temperature anomalies used to calculate the trend in Fig. 73.2 were determined using the raw data from each station, using the method that I have used throughout this blog. However, if we average the anomaly data for the same twelve long and medium stations using not the original raw data, but the adjusted data that Berkeley Earth create from the raw data (both adjusted and unadjusted data sets are listed in the data files on their website), the result is the temperature time series shown below in Fig. 73.5. These trends are very similar to the official Berkeley Earth trends shown in Fig. 73.4 above. This once again suggests that gridding and homogenization may be unnecessary steps in creating regional trends. Simple averages of all station times series appear to work just as well in most cases, particularly if the stations are fairly evenly spread.
Fig. 73.5: Temperature trends for Sri Lanka based on the average of anomalies for all long and medium stations using Berkeley Earth adjusted data. The best fit linear trend line (in red) is for the period 1876-1975 and has a gradient of +0.28 ± 0.02°C/century.
The difference between the Berkeley Earth temperature trend in Fig. 73.5 that is based on adjusted data, and the trend derived solely from raw data can be calculated by subtraction. The result is shown is shown in Fig. 73.6 below (blue curve). Also shown are the breakpoint adjustments (orange curve) that are added to the data by Berkeley Earth. It can be seen that most of the difference between the data in Fig. 73.5 and that in Fig. 73.2 is due to the breakpoint adjustments in this case. In fact they add over 0.5°C of warming between 1890 and 2010. This explains why the official Berkeley Earth trends in Fig. 73.4 are so different from the trends that result from the actual raw data in Fig. 73.2.
Fig. 73.6: The contribution of Berkeley Earth (BE) adjustments to the anomaly data in Fig. 73.5 after smoothing with a 12-month moving average. The blue curve represents the total BE adjustments including those from homogenization. The linear best fit (red line) to these adjustments for the period 1876-1975 has a positive gradient of +0.171 ± 0.007 °C per century. The orange curve shows the contribution just from breakpoint adjustments.
Finally, a word of caution. It is also clear that the data in Fig. 73.2 before 1875 is much less reliable as an indicator of the true mean temperature in the region due to insufficicient stations in the average. If we ignore that data, then the trend for Sri Lanka will be as shown in Fig. 73.7 below.
Fig. 73.7: A rescaled plot of the data in Fig. 73.2. The best fit is applied to the monthly mean data from 1876 to 1975 and has a positive gradient of +0.11 ± 0.03 °C per century. The monthly temperature changes are defined relative to the 1951-1980 monthly averages.
Summary
It is clear from the raw data that there was no climate change in Sri Lanka before 1975. Temperatures were actually stable for over 100 years prior to 1975 (see Fig. 73.7), just as in India.
Since 1975 there has been a modest temperature rise of about 0.4°C (see Fig. 73.7), but this is a long way short of the almost 1.5°C claimed by Berkeley Earth (see Fig. 73.4) or the similar value currently being touted by climate science for the mean temperature change for the entire Northern Hemisphere.
No comments:
Post a Comment