This post refers to my analysis of recorded annual mean temperature trends at 50 stations worldwide. The stations were all with long records (many over 100 years) and located in rural areas or small towns. The analysis used raw data (as recorded, without any adjustments) as published by KNMI (the Dutch Meteorological Institute) or BOM (the Australian Bureau of Meteorology). Polynomial curves of best fit were added to the plots.
In all, a total of 249 stations worldwide were used in a larger analysis, but these have been summarised here by selecting 50 stations that were typical for each region. The selected stations were not “cherry-picked”. Links to the analyses at the 249 individual stations are given below.
No attempt was been made to derive “worldwide average” temperature trends. Instead analyses were made at individual stations as this was considered to be more physically meaningful and practical using thermometer observation data. Using this approach any inconsistencies due to data quality would be identified as outliers.
The analysis showed that:
About 60% of stations had a strong cyclical temperature trend with periods (the time between peaks in the trends) of about 80 years. Commonly trend peaks occurred around 1930 and 2010 with trend troughs around 1900 and 1960, although this does vary. The “peaks” around 2010 are often difficult to confirm (this will be known in about 2020!) but it is common for the trends to have been “flat” since around 2000.
About 85% of stations had peak or plateau trends during the period since about 2000.
About 15% of stations had temperature trends that have been increasing since about 2000, with no evidence of a peak in the trend (the presence of a peak in the trend cannot be confirmed or rejected until several more years of data become available).
About 35% of stations had an overall neutral (level) or decrease in temperature trend over the past century.
Plots are shown below for six stations. These are examples of the typical long-term temperature trends at the 50 stations. It should be noted that the polynomial fit line (in red) can sometimes be misleading at either end of the series as the line is very dependent on the temperature values in individual years (this sensitivity does not occur away from the start and end of the series).
Morris, USA 1886-2016
[ The Morris station is located on an agricultural Experimental Station in rural Minnesota, in the upper mid-west of USA. There is a well-defined cyclical trend with a peak around 1930 and a trough around 1980 (and possibly around 1890). There has been another peak (or at least a plateau) in the trend around 2010. There was decrease of about 1.0 degC between ~1930 and ~1980, but an increase of about 0.8 degC between ~1980 and ~2010. The overall long-term trend was about zero.]
Valentia Observatory, Ireland 1869-2014
[ The Valentia station is located at an observatory in a rural coastal location in the southwest of Ireland. There is a well-defined cyclical trend with peaks around 1940 and 2010 and troughs around 1900 and 1970. Between ~1930 and ~1980 the overall trend in the peaks has been about +0.5 degC/century. The overall average trend was about +0.3 degC/century]
Vardo, Norway 1829-2015
[ The Vardo station is located on the coast in the far north-east of Norway and has continuous temperature records since 1829. There is a cyclical trend with a peak around 1940 and troughs around 1870 and 1980. Since 1980 there has been an increase of about 2 degC, with no evidence of a peak or plateau in the trend. Between ~1870 and ~1940 there was an increase of about 0.7 degC, but this then plateaued until about 1980. Prior to 1980 the overall average trend was about +1.1 degC/century]
Salehard, Russia 1882-2016
[ The Salehard station is located inland in the north-west region of Siberia, Russia. There is a well-defined cyclical trend with peaks around 1930 and 2010(?) and troughs around 1880(?) and 1970. There was decrease of about 0.7 degC between ~1930 and ~1980. The overall average trend was about +1.4 degC/century]
Richmond, Australia 1893-2016
[ The Richmond station is located inland in the north-east of Australia. There is a well-defined cyclical trend with peaks around 1910 and 2010 and a trough around 1950. There was decrease of about 0.4 degC between ~1910 and ~1950. The overall average trend was about +1.0 degC/century]
Isla Juan Fernandez, Chile 1901-2015
[ The Isla Juan Fernandez station is located on an island in the Pacific Ocean about 650 km off the South American coast. There is no cyclical trend with peaks, but a fairly constant decrease at a rate of about -0.4 degC since records commenced in 1901. ]
For convenience of management the 50 selected stations have been grouped into four large regions (see map above) as follows:
Americas region, 14 stations
Europe and northern Africa region, 9 stations
Asia region, 13 stations
Southern Africa, Australia, Antarctica region, 14 stations
Links to the analyses of the long-term temperature trends at all 50 stations are presented within the four large regional blog posts:
Similar links to the analyses of the long-term temperature trends at all 249 stations analysed are presented within 27 smaller regional blog posts:
Please let me know what you think of my analyses. firstname.lastname@example.org
This article was written by Brian Gunter of Narooma, NSW, Australia. In his previous life Brian was an engineering hydrologist involved over many years in the analysis of rainfall and river flow data for the planning of water resources projects in Australia, Asia and Africa. In recent years he has been one of the Marine Rescue NSW (previously Royal Volunteer Coastal Patrol) volunteer weather observers who operate the Narooma station for the Bureau of Meteorology.