Fig1. First a rough illustration showing an area of Australia and Indonesia where a light cooling trend (blue area) can be found in unadjusted GHCN temperature data over the last century. Other areas of Australia show different warming trends.
Temperature data from Indonesia/Malaysia/Papa New Guinea is limited when t comes to the long series. So at least for now I will not try to give an estimate for these countries, but just show examples of temperature trends:
Fig2. Mostly cooling trends in the north Australian – Indonesian area. However especially in the 1930´ies some temperature series (red) has a significantly different trend. We will return to these. Especially the Christmas Island shows significant cooling in the last century.
Fig3 . Click to enlarge. Australia was divided into the above zones where areas of similar temperature profiles where grouped. Especially the coastal areas have quite different temperature trends than the non-coastal areas. This phenomenon has been observed, in for example NW Africa, NW Europe, East China and more.
Now that we have seen a row of temperature series from the Indonesian-Northern Australian area, let’s start out with a more detailed look at North Australia.
Temperature trends for mostly the western stations of Northern Australia:
Fig5. For these temperature series from Northern Australia we see some divergence mostly before 1910-20, otherwise good agreement.
Fig6. More stations added to the Northern Australian ensemble…
Fig 7. – And the last stations of the Northern Australia area..
This gives average temperature trend:
Fig8. Avg the Northern area Australian area.
As we will see, coastal stations normally have a significant different temperature trend than the non-coastal stations. However, this phenomenon appears less outspoken in areas of Northern Australia.
The near-coastal station Wyndham has been included in the Northern area although the trend appears partly like the coastal stations, partly like non coastal.
Fig9. Wyndham temperature series is cut in 2 parts from 2 stations almost at the same spot. However, sometimes this can still lead to different temperature level. However, when comparing with 2 near by stations, no need to adjust the 2 Wyndham series in any direction. The coastal series often has colder temperature series pre 1930 and warmer 1960-70 data. This is also the case for the near coastal station Wyndham, but still, the typical North Australia temperature trend where temperatures 1930-40 are comparable to 1980-90 data appears too, and thus, the Wyndham station does indicate that the Northern Australia area extends to the NW Australian area.
Fig 11: Wyndham station confirms the longterm trends of Northern Australia but shows typical coastal trend deviations.
Fig12. The Darwin station has been debated. Above a compare of the discussed temperature drop 1930-1950. Not only the Darwin station is in compliance with the general Indonesia-area trends, but a closer compare with North Australian stations shows, that Darwin unadjusted GHCN data is in full compliance with general North Australian trends.
- Or so to speak: North Australia temperature trends appears “homogenous”, and any data change is likely to induce “in-homogeneity”.
Above we also see, that the station “Winton Post office” shows the greatest temperature drop in the period, but sadly this station has no data public available before 1938.
Fig13. East Australia mostly has stations showing light cold trend but also the opposite here and there. For example, it seems that (warmer) trends from central Australia intrudes from west towards east just as colder trends from east Australia appears further east, so that the typically warmer coastal trends is gone for this part of the coast.
Fig 14. Click to enlarge. The elevation map appears to indicate the possibility of central Australian air to move further east where mountains do not prevent it? All in all the trends in East Australia appears a little mixed, here the average of the stations (30N-23N) :
Fig15. The mix of trends in the east Australian area results in a almost flat trend. The Barcaldine in red
shows an example of a trend apparently with more warming within the East Australian area.
Some readers might have noticed the extreme limited of data made public by GHCN after 1990. The data made not public is by far the largest source of uncertainty when estimating temperature trends world wide. For east Australia we see how basically 2 rather different temperature sets alone should justify an the estimate 1990-2010.
And worse: The datasets chosen to show 1990-2010 by GHCN are cut around 1950. This is a very important problem: Temperature stations can be located in places more or less constant in temperature trend. This means: If you for example chose a temperature station showing strong heat 1980-2010 it is very likely that the same station did so too in 1930-1950. Thus, by using only the 1950-2010 part is a massive source of Bias or at least the risk of twisting temperature trend significantly even without any data has been adjusted. By averaging wrongly with a chopped of 1950-2010 dataset, this can induce much more error to the resulting average trend than adjustments or anything else.
I see absolutely no reason for GHCN to make all these temperature data NOT public. How can it benefit anyone in any way that data are available is no one seeks to hide anything? This is a scandal.
Coastal stations Eastern Australia
Fig 16. Click to enlarge. The East Australian coastal stations are shown in purple. In general, coastal stations - especially with mountains on the land side - tend to have significantly different temperature trends than the non-coastal series. Temperature data shown so far has been non-coastal.
Fig 17. NE Australian Coastal temperature trends: A rather constant increase in temperatures. The strong difference between Coastal and non-coastal trends may be explained by shifts in dominating wind directions and the general temperature increase in the large oceans.
Fig18. Unlike the NE Australian coastal temperature series where data is cut in 1907, the SE Australian coastal temperature sets are often intact all the way from 1970. Thes data reveals a very strong oceanic cooling 1880-1907 – or at least coastal cooling. In fact, the ocean-affected coastal stations still today has not fully reached the warm level of 1880-1900. 20 years of data from 5-6 comparable stations agrees on this.
Another problem is obviously, that all stations has no data public available 1990-2008.
South East Australia
Fig19 SE Australian area shown in blue.
Fig 20. Again, the biggest problem when estimating temperature is the massive limitation of public data 1990-2010. For the 50 datasets I used for SE Australia non-coastal, only a few holds some 1990-2010. And again, datasets with data 1990-2010 are often cut off before 1950.
I must repeat: This is a serious problem, because stations with more temperature variability will obviously show more heat in both the recent warm peak 1980-2010 and the earlier warm peak 1930-1950. So to use a temperature station with more variability than average, and then only use the 1950-2010 part induces a trend with a serious warm bias.
In fig 20, we have this issue illustrated, it seems? Both the Adelaide station (blue) and the Cobar station (red) has some more temperature variation than the average curve. It seems that the variability in the Cobar and Adelaide stations are similar, and notice how the Adelaide early years are far warmer than average. This can be a coincidence, but illustrates the problem of not showing the pre-1950 temperatures of the few datasets you use for 1990-2010 data. This way any kind of datasets could be used for 1990-2010.
The problem is also seen for the Condobolin Hadcrut station where we see large variability both in recent ears and earlier years. For both Condobolin and Adelaide, it seems that the pre-1950 temperature where at least as warm as today’s temperatures.
Fig 21. With this problem caused by extremely poor GHCN data 1990-2010 public available, I believe that the best solution is to use the stations 1990-2010 that most resembles the average trend 1950-1990. These stations are East Sale and Launceston. Despite the obvious lack of data, this appears somewhat useful.
Here all datasets of SE Australia together:
Fig22. Estimate of SE Australia Non-coastal temperatures. Quite a resemblance to the Northern Australia temperature trends.
SE Australia, South Coast.
Fig23. SE Australia, South Coast. (See yellow area, fig 19).
Like the SE coastal trends, the 1880-1900 data indicates coastal (oceanic) heat at least as warm as today. The 1990-2010 data are only scarcely public available, but the trend 1990-2010 strongly resembles the south coast further west, “South Australia” and also the trend from SE Australia.
Fig24. Just a small area on the central Tasmania appears not so affected by coastal trends.
Fig25. Again, differences between coastal and non-coastal are significant. It surprises me though, that the 1880-1900 data - here represented by 1895-1900 – are quite cold. This does not resemble the continental Australian SE coasts. On the “positive” side, both coastal and non-coastal trends 1990-2010 resembles the SE Australian datasets, and thus, the choice of datasets for 1990-2010 for SE Australia once again appears to be useful.
Fig26. Stations of the South Australian area. This coastal area deviates from the other coastal areas since coastal temperature trends also matches stations further away from the coast. The area is rather flat, and thus we might expect temperature trends to extend further from the coast.
Fig27. Temperature profile for South Australia. Again a significant drop in temperatures after 1990 and quite warm early years.
Fig 28. Stations for SW Australia, coastal (Green) and non-coastal (blue).
Fig29. SW Australia Coastal trend.
Fig30 SW Australia: Now a warm trend appears in data. However, this warm peak is again partly dependant on scarce data 1990-2010. Only 2 stations have any data 1990-2010:
Fig 31. The 2 stations where the public can see 1990-2010 data are cut of in the 1940´ies as usual.
The Meekathara A (yellow) station appears to vary strongly in temperature, and we can see that several times before 2010, this yellow graph is warmer than 2010. By using this temperature sensitive graph as 50% of the 1990-2010 data we risk have temperatures dragged up resulting in a partly artificial warming trend.
Fig32. SW Australia coastal temperature trends. Data appears homogenous and solid.
One odd station I would like to mention is Albany:
Fig33. This stitched pair of Albany trends yields a strongly cooling trend. This is not supported that well by other stations near by and thus I have not used the older part, “Albany town”. An outlier?
Fig 34. stations of the NW Australia area.
Fig 35. The NW non-coastal area holds some trends similar to the Northern Australia’s warmer 1930-40´ies for example marble Bar and perhaps Gascoyne Junction. However, I did not chose to extend the “Northern Australia” further south since the stations Nullagine especially before 1934 shows a different trend, more similar to the SW Australia trends.
Fig36. NW non-coastal Australia area temperature trends. No data 1990-2008.
Fig37. NW Australia coastal stations. Again this truly odd “system” : The temperature series used to show us recent 1990-2010 temperatures are cut of typically in 1950 (and 1940). Would it not be interesting to see how the stations showing strong heat in 2010 behaved in the 1930´ies?
Another feature here is, that this massive 2010 peak in NW coastal stations is not confirmed by Hadcrut:
Fig38. I must say it does not feel satisfying to work with data like this 2010 peak when Hadcrut and GHCN completely disagree. However, this is a study of GHCN unadjusted stations, so I will proceed with the GHCN versions.
Fig 39. Central Australia. We see a significant warming trend 1940-2010 but also a cooling trend 1880-1940. As normally seen, all temperature series where data 1990-2010 is public available are cut of in 1940-50 and we have to guess how these series behaved in the 1930´ies, 1940´ies.
I have highlighted the Charleville (yellow) temperature series. Would it not be relevant to know how this series behaved before 1940?
Australia overview and totals
Fig 40. To estimate Australian average temperature, the above fractions have been used to weight the temperature data. Coastal areas holds 9% of the Australian area.
Fig41. Australia temperature trend. The regional temperature trend is now shown with base period 1961-1990. Before 1882 less then 50% of the Australian territory is represented, and thus the starting point is set to 1882.
For the years 1882-1896 around 64% of the Australian area are represented:
West Australia Coast
SE Australia, West Coast
West Australia Coast
After 1898 89% of Australia is represented. After 1906 all areas are represented. BOM often show the 1910-2010 interval for Australian temperature anomaly. For this period the RUTI results gives a +0,29 K/century temperature trend.
The “flat trend” 1882-2010 has a slope of +0,13K/century.
Warmest years in Australia:
1 1889 : +1,00 K
2 1988: +0,70 K
3 1915: +0,64 K
4 2009: +0,62 K
5 1994: +0,59 K
6 1914: +0,59 K
7 1882: +0,58 K
8 2005: +0,57 K
9 1980: +0,57 K
10 1990: +0,57 K
Fig42. The roughly observed area of a cooler “Eastern Part” of Australia as indicated in fig 1 is confirmed. The rest of Australia, here called “Western Part” has a significantly different and much warmer trend. I find this fascinating also because in some cases just a few 100 km seems to make all the difference.
Fig43. East Australia part, compare coastal vs. non-coastal. For East Australia the coastal trends shows around 0,7K more warm trend 1900-2010 than the non-coastal trends.
Fig44. West Australia Part, compare coastal vs. non-coastal. For West Australia the coastal trends shows around 0,6K more warm trend 1900-2010. In general, the Western trends are far warmer than the Eastern trends, but non the less, the coastal vs. Non-coastal difference are quite similar for the 2 regions.
Fig45. Pacific coast temperature trends vs. Indian Ocean coast temperature trends. Significant difference before 1940.
Fig46. Compare with BOM´s results for Australian temperature anomaly. For Australia both RUTI and BOM use base period 1961-1990.The 1940-1980 period is very similar RUTI vs BOM. This can hardly be a coincidence and such a resemblance confirms correctness of both RUTI and BOM data and methods 1940-1980. But outside the 1940-80 period, BOM suggests far warmer trend than this writing, RUTI. The BOM trend 1910-2010 is around 0,9 K/ century while RUTI suggests 0,22K.
Fig47. The BOM map of temperature trends, there is not any area showing a cooling trend. Compare with fig 42.
The strong warming trends of the Eastern and western coasts cannot be spotted at all, as if BOM simply has not noticed this important trend in data? In addition, the most central area of Australia has a warm trend above the rest of Australia. Alice spring is located here, but I find that Alice spring in the heart of Australia happens to show us the overall Australian trend:
Fig48. On this illustration I seek to compare trends between the central Australian station Alice Springs with the overall Australian trends. Both temperature sets shows almost no trend 1880-2010 and a dive around 1940-50. To do this compare of trends, I reduced the Alice Spring trends with a factor 1:2,5 .
The Alice Spring temperature series has a trend 1910-2010 of approx 0,85K / century. BOM suggests with their graphic, that the area should have twice this warming trend.
Final comment: This article does not claim that RUTI results are fully correct at all, but non the less, it is relevant to analyse what unadjusted GHCN data can tell us. There is little doubt that BOM has not identified the significant differences between coastal and non-coastal trends and may have used coastal data that represents roughly 9% of Australia to represent more than 9%. This is just a guess for my part, fact is that BOM somehow ends up with a warming trend 1910-2010 5 times what I get when using GHCN unadjusted data. Critic and suggestions is welcome.