62. Chile - temperature trends COOLING

At over 3,600 km in length, the land border between Chile and Argentina is probably the longest continuous land border between the same two countries anywhere in the world (part of the USA-Canada border is through the Great Lakes). You might think, therefore, that the climates of these two countries should be very similar, and that their experiences of climate change should be the same. Except they are not. The reason, of course is that they are separated by the Andes mountains. So while the climate of Argentina over the last 100 years has been stable but with a sudden temperature rise of about 0.45°C in 1967 (see Post 61), the climate of Chile has been cooling steadily, just as most of the South Pacific has as well (see Post 34).

Overall there were 50 stations in Chile with over 300 months of data up until the end of 2013. The longest of these is Santiago, which with 1835 months of data is the second longest temperature record in South America behind Rio de Janeiro. There are also two other long stations with over 1200 months of data, and another 31 medium stations with over 480 months of data. It should be noted also that of the 50 stations being considered here, 20 have little or no data after 1960 and 16 have little or no data before 1960. This makes the choice of MRT interval problematic (monthly reference temperatures or MRTs are explained in Post 47).

Fig. 62.1: The temperature trend for Chile since 1860. The best fit is applied to the interval 1891-2010 and has a negative gradient of -0.30 ± 0.06 °C per century. The monthly temperature changes are defined relative to the 1961-1990 monthly averages.

The temperature trend in Fig. 62.1 above was derived by averaging the temperature anomalies from all the stations with more than 300 months of data which also had at least twelve years of data within the interval of 1961-1990. This amounted to 94 stations in total (for a list see here). The interval of 1961-1990 was used to determine the monthly reference temperatures (MRTs) against which the temperature anomalies are determined, as explained in Post 47. 

The trend in Fig. 62.1 is clearly strongly negative as indicated by the red best fit line. However, the temperature change is not uniform and there is considerable variability. The trend from 1960 onwards is the result of averaging over 20 different sets of temperature data, as shown in Fig. 62.2 below. This suggests the trend after 1960 is highly reliable, as I explained in Post 57 previously, while that before 1930 will probably be much less so.

Fig. 62.2: The number of station records included each month in the mean temperature trend for Chile when the MRT interval is 1961-1990.

The geographical distribution of the long and medium stations in Chile is illustrated in Fig. 62.3 below. These are classed as either warming stations (in red) or stable/cooling stations in blue. The criteria for determining if a station is warming are two-fold. First, the temperature trend must exceed twice the error in the trend in order to be statistically significant. Second, the overall temperature rise must exceed 0.25 °C in order for it to exceed the threshold below which it could be considered as merely a random fluctuation in the data. As I have pointed out previously, this threshold of 0.25°C may be on the low side as natural fluctuations in the long-term temperature trend may be much greater than this as the 5-year moving average in Fig. 62.1 appears to indicate.

Fig. 62.3: The (approximate) locations of long stations (large squares) and medium stations (small diamonds) in Chile. Those stations with a high warming trend are marked in red. Those with cooling or stable trends are marked in blue.

Clearly Fig. 62.3 shows that less than a third of stations in Chile have warmed over their history. It is also clear from Fig. 61.3 that there is a good spread of stations around Chile with little clustering, but a higher density of stations in the middle of the country south of Santiago. However, this does not appear to affect the simple averaging approach employed here to determine the regional temperature trend in Fig. 62.1 as the following graphs will show.

Fig. 62.4: Temperature trend in Chile since 1860 derived by aggregating and averaging the Berkeley Earth adjusted data for all medium stations. The best fit linear trend line (in red) is for the period 1891-2010 and has a gradient of +0.72 ± 0.02 °C/century.

The accuracy of the simple averaging process used in Fig. 62.1 can be tested by comparing two different trends that were each calculated using the same data but with different averaging techniques. The regional trend for Chile in Fig. 62.4 above was calculated using Berkeley Earth adjusted data and the simple averaging method. The trend shown in Fig. 62.5 below and published by Berkeley Earth was calculated using the same Berkeley Earth adjusted data, but with different weightings for each station based on station density and correlation with its neighbours. The data in the two graphs appear virtually identical, despite the fact that slightly fewer stations were used to generate the trends in Fig. 62.4.

This comparison of the trends in Fig. 62.4 and Fig. 62.5 clearly shows that a simple average of the adjusted data from the Berkeley Earth data files (Fig. 62.4) gives the same result for the regional trend in Chile as the Berkeley Earth version does (Fig. 62.5), even though Berkeley Earth appears to use weighted averages for its regional averaging. This in turn also suggests that weighted averaging is probably not necessary in Chile and simple averaging of stations is sufficient to generate a reliable trend.

Fig. 62.5: The temperature trend for Chile since 1840 according to Berkeley Earth.

However, what is also apparent is that the temperature trend produced by Berkeley Earth in Fig. 62.5 bares little or no resemblance to that which was derived from the original data in Fig. 62.1. A negative trend of -0.3°C per century in Fig. 62.1 has miraculously become a huge positive trend of +0.72°C per century in Fig. 62.4. The total difference between these two trends is illustrated in Fig. 62.6 below and amounts to an additional 1.02°C of warming over the last century. This is the result of adjustments made to the original data by Berkeley Earth. This explains the difference in trend gradient in Fig. 62.4 compared to that in Fig. 61.1. However, as I demonstrated in Post 57 previously, most of these temperature adjustments are unnecessary. Not only that, they are probably wholly unjustifiable from a statistical standpoint.

Fig. 62.6: The contribution of Berkeley Earth (BE) adjustments to the anomaly data in Fig. 62.4 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 1891-2010 has a positive gradient of +1.02 ± 0.04 °C per century. The orange curve shows the contribution just from breakpoint adjustments.

Conclusion

The results here indicate that there has been no global warming in Chile in the last 100 years. In fact the climate there has cooled substantially in that time.

 

 

Addendum

As I noted above, the use of 1961-1990 for the MRT interval results in the exclusion of 20 of the available datasets from the trend in Fig. 62.1. However, if the MRT interval is instead chosen to be 1931-1960, most of these stations will be captured while about 16 stations with data mainly after 1960 are ejected from the average. This is illustrated in Fig. 62.7 below where the number of stations in the 1930s and 1940s increases to about 32 (compared to 14 in Fig. 62.2) while in the 1980s the number falls to about 14 from about 28.

 

Fig. 62.7: The number of station records included each month in the mean temperature trend for Chile when the MRT interval is 1931-1960.

 

The impact of this on the trend is shown in Fig. 62.8 below. It can be seen that the temperature trend is now even more strongly negative with almost 0.5°C of cooling occurring before 2010. What is certainly clear is that there is no global warming occurring in Chile.

 

Fig. 62.8: The temperature trend for Chile since 1860. The best fit is applied to the interval 1901-2010 and has a negative gradient of -0.44 ± 0.06 °C per century. The monthly temperature changes are defined relative to the 1931-1960 monthly averages.