G'day all,
Because I have been harping on about the data sets and how they show, well, nothing, I thought it is worth looking at a few weather station sites. These were my random selection done quickly for a lecture I gave from data freely available on the NASA site
http://data.giss.nasa.gov/gistemp/station_data/ . Urban effect should be at work in many of them. Here is a list plus some comments:
Alice Springs (Central Australia)
No pattern ? A slight overall cooling mainly because of a huge drop in 1978.
Five different data sets. Not continuous.
Baltimore Wso City
Very good trend upwards. A poster child for global warming (except for urban effect)
One data set. Continuos from 1890 and with data back to 1880.
Bathurst (rural inland eastern Australia)
Trend downwards. Little urban effect because it is a rural station outside the city.
One data set. Continuous only from 1963. Ignoring the slight gaps goes back to 1910.
Chongqing (China)
Very slight cooling trend.
Three data sets. Match in trend with some significant differences but quite different temperatures. Continuous from 1921.
Christchurch (city, South Island, New Zeland)
Maybe a slight warming trend but very big yearly variations.
One data set. Continuous from 1902.
Curitiba
Significant warming trend.
Three data sets. Continuous from 1884 to 1996.
Dawson, Yukon Territories, Canada
No trend. Very big yearly variations.
Three data sets. Line up reasonably.
Elko Faa Ap
Cooling trend but very slight. Very very big drop in late 90s.
One data set. Continuous from 1880.
Farina
Cooling trend except for exceptionally hot year in 1917.
One data set. Continuous from 1889 to 1939. No more recent data.
Fredericksburg Nation Park (US rural)
Very slight cooling trend.
One data set. Continuous from early 1890s.
Las Vegas/Mcc
Very distinct warming trend. Exactly as would be expected for a massively expanding major city from a tiny outpost.
One data set. Continuous from 1938.
Lithgow (An attempt to determine the urban effect of Sydney being a township on the fringes ? actually helps to show the futility of such an exercise as it and all other peripheral areas do not match in pattern to Sydney at all).
Perhaps a slight warming trend.
One Data set. 1915 to 1958. 1963 to present.
Livingstone
Distinct cooling then distinct warming from 1980.
Four Data Sets. Match with a few exceptions. 1918 to 1987. Some data for 1999 to 2001.
Mcgill
Depends if you include the discontinuous considerably lower date from late 1880s to 1900 a slight warming trend but if not then a slight cooling trend. The early data suggests a different system of measurement since even the highest in that period does not come up to the coldest of the remaining 80 odd years.
One data set. Late 1880s to 1990. 1916 to present.
Mirnyj
No trend. Yearly variation too great to reasonably ascribe a trend to it.
Six data sets. Closely correspond. Late 1950s to present.
Mistassini Post, Quebec
No trend. Another locale that has a discontinuous early reading that is very much lower than the rest of the set. Enough of these would suggest that methods of measurement in the late 1800s resulted in much lower temperature readings. Would make a good study.
Three data sets. 1899. 1915 to 1982 using the overlap of the three sets. Otherwise not continous.
Moosonee, Ontario
Perhaps a cooling trend from the 1950s peak but one data set shows a very high rise in 1999.
Eight data sets. 1890 to 2003. Must use all data sets to obtain a set up to 1995. Where they overlap the sets do reasonably correspond.
Nantucket/Faa Airport
Cooling trend.
One data set. 1946 to 1975. 1977 to 1983.
Owings Ferry Landing
Slight cooling trend.
One data set. 1917 to 1981. 1983 to 1999.
Porto/Pedras
Distinct Cooling trend (for what it is worth, and that is not much)
Three data sets. 1880 to 1901. 1959 to 1982. Quite large differences between the three sets (all in the 1960 to 1980 range).
Rome Italy
Line ball or slight cooling (should be a distinct warming trend because of the urban effect but Rome has been paved for a very long time and large commercial buildings may not have had as large an effect because of the very large areas of vegetation in the heart of the city)
One data set. 1880 to mid 1930s. Then 1941 to 1995.
Rome (34.2 N, 85.2 W)
Cooling trend overall, short sharp warming trend from mid 1990s.
One data set. 1880 to present with one year break 1890.
Royal Oak 2ssw (38.7 N, 76.2 W)
Distinct warming trend.
One data set. Early 1890s to present.
Salisbury/Wicomico Co Ap
Cooling trend.
Two data sets. Around 1950s and from early 1960s to present.
Wells (41.1 N, 115.0 W)
Distinct cooling trend overall, slight warming trend in later part.
One data set. 1880 to 1918 then 1940 to present.
Total: 25 (not a very big sample but it?s better than nothing and to include the 200 I did for the lecture would run to pages). I just picked these at random from the random sample. But feel free to get your own random sample.
Warming Trend: 4
Slight Warming Trend: 2
No Trend: 5
Slight Cooling Trend: 9
Cooling Trend: 5
If you averaged the data, there might be a warming trend because the distinct warming trends are larger than the distinct cooling trends. But overall, there is no pattern. This is the problem with an average world temperature.
Should you really average each locale you have without regard for distribution or degree of variation? Not only does urban effect come into play, but so does the distribution that means there are more sites in the warming trend areas than in the cooling trend areas even though the cooling trend areas cover much larger chunks of the earth. And even with this bias, in number you still get more cooling trend stations than warming ones, just the amplitude is not as large.
And what of the majority of the planet that is under water. SAT temperatures do not include any temperatures over the sea. Very big area to ignore. The satellite data does include such areas and gives a distribution unconcerned with where the stations are or how accurate the equipment or the recording. It simply averages the 17,000 evenly distributed measurements daily.
This is real science. The science of direct observation. Anyone can repeat the experiment. Anyone can do the calculations. Anyone with enough time could go through the whole 7,000 and create a distribution both in regions and in trends. Hasn?t been done but it is not difficult to do. Until then, samples at least show up the difficulties and the degree of variation and the fact that the majority of stations show a decrease despite what should be an overall increase because of urban effect. The patterns, including gaps and multiple sets, also show up the likelihood of accuracy. The US tends to have only one set and tends to be continuous (although surprisingly the actual continuous data sets in number are still quite small) whereas other parts of the world show gaps and multiple sets, suggesting problems with the data and its method of collection, plus the likelihood of the station being moved (and when they get moved they tend to move towards large population centres because weather records are the most local use for population centres, worst for trends, but the people taking the measurements are not doing them so that someone could use them for global studies).
Food for thought, for those that wish to think.
Regards
Richard
