Fig1 China (except for a part of the Himalayas, see RUTI Himalaya) was analysed in 5 areas
Fig2. . The overall estimate was calculated weighting area “China NW1”: 30%, “China NW2”: 13%, “East Central China”: 18%, “China NE”: 17% and “South China” 22%.
Fig 3 For compare, the Wang 2001 reconstruction of Chinese temperatures based on temperature stations and temperature proxies: WANG 2001
Fig 4. In 2008 (http://www.agu.org/pubs/crossref/2008/2008JD009916.shtml) Phil Jones estimated Chinese temperatures to have increased roughly 0,8K 1951-2004, after taking out impact of UHI. In this writing the same temperature rise is estimated to be just slightly lower, 0,7K. However, when examining Chinese temperatures, it appears relevant to examine trends beginning earlier than 1951 due to the significant warm peak 1935-50.
The 5 Chinese areas.
As always in “RUTI” I try to define areas with somewhat similar temperature trends so that the averaging do not create too large errors, and one will easy notice if one temperature series has some kind of issue.
I the case of central China, few datasets contains the data for 1990-2010, and those who do are often urban to some degree. But here Is what I found:
Fig5. Notice the massive difference of data in years 1990-2010. This difference is strongly dependant on geography, and thus a rough dividing of the southern (tropic) temperature trends versus northern could be made. The orange stars (and one light blue) indicates the series used with data 1990-2010:
Fig6. Rough division of south vs north China based on differences in 1990-2010 temperature trends.
Since the Eastern part of central China do not offer 1990-2010 temperature trends, except for strongly urban data, then I decided to also do a check on differences in 1930-50 temperature when available. In this area, these data are more often available than 1990-2010 and I could confirm the above “line” between southern and northern climate trends, except, the Eastern central China resembled the northern trends more and the definition of the South area needed to be changed:
Fig 7. The number of temperature stations containing 1930-50 data is more numerous, and as we can see, the availability of 1990-2010 data is scarce in this selection. To define the South China area, I used the above stations and now the South China area is defined as:
Fig8. This definition of South China was then used to define other areas of China.
Fig9 For South China, I now included more data series, even datasets with just a few years 1930-50, and got the above temperature profile. Some urban stations are not used, and this gives a rather homogenous temperature trend for South China.
One issue here is of course the absence of data 1990-2010 when deselecting some urban sites with trends that significantly deviates from the above trend. Therefore, I concidered to include some urban data I had previously deselected:
Fig10. Examples of urban data now added to the bulk of South China data. My impression is that using such urban data makes results become more random than useful. So I decided to use only the 1978-2010 part of these urban series, and doing so by stitching these using 1978-1988 baseline, and this now yields my estimate of Southern (tropic) China temperature trends:
Fig11. Adding urban temperature trends 1978-2010 (using 1978-1988 baseline) gives a larger data foundation to give an estimate of Southern (tropical) China temperature trend. Adding the urban series resulted in a rise in temperature trend 1990-2010 of roughly 0,2K.
Fig12. North west China was split up in two regions NW1 and NW2
Fig13. China NW 1 shows a rather significant warm peak in the late 1940´ies. We do not have many data series for this period, but the data we have originates from different parts of the region, and thus it seems that the warm peak is spread out over a larger area.
Fig 14. Further east, for China NW 2, the period around 1940 is still warm, but we now see a strong warm peak too 1945-1951
China NE and Eastern Central China.
Fig15. Click to enlarge.
The North East Chinese temperature series are not that complete, but on the boarder to the NE China area, fortunately we have Russian temperature stations that seems to complete the missing Chinese data, 2 examples:
Fig16. Chinese: Yanji / Russian Pogranicnyj
Fig17. Chinese Qiqihar / Russian Nercinskij Za and Borzja
Fig 18. Here the trends based only on Russian stations (And one from Mongolia).
Fig19. Chinese NE including Russian temperature trends. For the Chinese NE area, the warm peak 1940-50 last shorter than the warm peak 1990-2010.
East central China.
For this area we have the opportunity to study the coastal versus non-coastal temperatures. However, Especially in the coastal Chinese area many temperature station.
Fig 20. Rural temperature stations of Central east China divided on Coastal – non-coastal location.
As earlier concluded in for example RUTI NW Africa, coastal location is extremely important to take into account when trying to identify the general temperature trends.
Now lets take a look at the temperature trends for east central China, urban vs. rural:
Fig21. As concluded before, the difference is extremely important if you seek actual temperature trends for an area.
Are the Rural data from East China well represented? The warm temperature trends data before 1950 are only public available for short periods, typically 5-7 years– as is also the case in some other areas of China. And thus, to establish a rural dataset, it is necessary to collect numerous rural datasets.
Fig22. “Graffiti temperature series”. The fragmented nature of the warmer temperature sets before 1950 in Eastern central China. These mostly rural datasets are typically represented with 5-7 years for the 1920-1950 period.
Despite the fragmented nature of the warmer datasets, there are still so many mostly rural datasets that do agree on a strong warm period before 1950 for Eastern Central China.
Fig23. East Central China temperature trends. For the East Chinese central temperatures, the data after 1990 is almost only available from strongly urban sites, and thus, just like for South China, I use 1978-88 as base period and stitch the mostly urban trends to the more rural data.
Overall Chinese temperature trend.
Fig 24. The estimated average Chinese temperature trend does not conflict with other continental Eurasian temperature trends.
Further comments of Chinese temperature data.
(will be updated now and then)
Hohot Hadcrut city-station.
This is the only temperature station added by Hadcrut in China which is not present in GHCN. In the same town “Hothot”, GHCN have a temperature station “Huehaote” a few km from the Hadcrut station.
Fig 25 . In the same town “Hothot”, GHCN do have a temperature station “Huehaote” a few km from the Hadcrut station.
Here we see the data available:
Fig 26 The GHCN station shows a warmer past not used by Hadcrut. (In addition, the Hothot stations are strongly urban).
GHCN temperature station Erenhot:
Fig 27 On the boarder to Mongolia we find a temperature station surrounded by mountains. This location is extremely sensitive to changes in temperature because of ice and snow areas will be present in large part of the year from most directions, and thus the ice albedo will amplify any general change in temperature by shrinking and growing. As we can see on the graph, the Erenhot station shows cold much more than average and likewise it shows heat more strongly than average.
The Erenhot data for previous warm period before 1951 is not public available from GHCN. There fore it will induce a significant warm error to use only recent very warm data from Erenhot and not warm data before 1951. So without pre 1951 data we cant use this station. This is sad because Erenhot is the only non-coastal rural station in East central China with data available for 1990-2010. Otherwise its just urban stations that holds data 1990-2010.
No doubt that the Erenhot station amplifies cold and warm trends, but has this got anything to do with its proximity to the mountains? It so happens, that we have a Mongolian station Zamyn-Uud located just a little longer from the mountains, and slightly outside the “mountain bay” where Erenhot is located.
Fig 28. Zamyn-Uud shows considerably smaller temperature variations than Erenhot, which confirms that location relative to mountains may be of high importance. Zamyn-Uud is located just slightly longer from the mountain area, but the difference in temperature trends is significant.