Guest post by Ed Thurstan
Abstract
It is apparent from the data that CRUTEM4 temperatures adjustments have, in part, been made with reference only to the earlier CRUTEM3 data, rather than raw temperature data. Further, the adjustments depend on the month for the data, and these adjustments are made for 20 or 30 consecutive years.
In the case of Adelaide (946720), for 30 years from 1857, CRUTEM4
- Lowers all January temps by 1.4oC
- Lowers all Feb temps by 0.9oC
- Lowers all March temps by 1.7oC
- With April to December all lowered by 0.5 to 1.1oC.
Thereafter, there are no adjustments until 2000, when a smattering of adjustments appear, mostly raising the temperature.
There are many examples of this practice. The total effect of all the differences between CRUTEM4 and CRUTEM3 where there is corresponding data is to accentuate warming trend by lowering pre 1995 temps by 0.1 to 0.2oC, and raising post 1995 temps by a similar amount.
This does not mean that the overall effect of all CRUTEM4 updates will induce the same magnitude change in HADCRUT4 anomaly, as no account has been taken of deleted and added stations, or relative numbers of stations in the database, vs the number which display differences as analysed here.
I believe that CRUTEM4 is seriously flawed, due to the apparent selective mechanical adjustments to blocks of station temperatures where the criteria for adjustment are a function of the month name. Until this is adequately explained, CRUTEM4 should be withdrawn.
Background
In March 2012 the Hadley Climate Research Unit released the land temperature dataset CRUTEM4, along with the station data from which it was constructed. A cursory inspection of the new dataset revealed some irregularities in Australian data. In particular, there were puzzling differences between CRUTEM3 and CRUTEM4 data.
A program was written to compare the two complete sets CRUTEM data, and highlight the differences. It compares the two sets to report:
· Stations in CRUTEM3 which do not appear in CRUTEM4. That is, they have been dropped in the construction of CRUTEM4.
· Stations in CRUTEM4 which do not appear in CRUTEM3. That is, new stations.
· Stations which appear in both sets and which have an arithmetical difference in any month of any year. The whole 12 months are reported for these. All missing data (reported as -99 in CRUTEM data) was converted to zero. This means that some data is lost if a valid temperature appears in one set with a matching -99 in the other dataset.
To exclude minor differences due to issues (like roundoff, precision) which could arise in comparing two datasets derived in different ways, a threshold of 0.22 was applied. That is, years are selected only if they contain differences whose absolute value exceeds 0.22oC.
Some simple statistics about the two databases include:
1. Station/Years of data: In CRUTEM3 set – 399,303; In CRUTEM4 set – 466,246.
2. Number of -99 months: In CRUTEM3 set – 674,993; In CRUTEM4 set – 568,606.
3. Number of CRUTEM3 stations dropped from CRUTEM4 286
4. Number of stations added to CRUTEM4 738
5. Matching station years where there is at least 1 month difference 104,296
6. Total stations in CRUTEM3 5,113
7. Total stations in CRUTEM4 5,565
Results
Old temperature data has been adjusted
In the following examples, only a snapshot of small parts of the data is presented to illustrate the point. Positive differences imply that CRUTEM4 is higher than CRUTEM3.
30050 is LERWICK UK; 42160 is ILULISSAT-JAKOBSHAVN Greenland;66450 is BASEL-BINNINGEN Switzerland; 67000 is GENEVE-COINTRIN Switzerland.
Adjustments of this magnitude can be seen through most of the CRUTEM4 database, but especially in Europe and adjacent areas. With data from such early years being so sparse, it is difficult to see both why such adjustments have been made, and the basis on which they were made. They were clearly made with reference solely to CRUTEM3 data, and not from original data. There seems little value in adjusting such early data, unless the purpose is to lower early temperatures.
There are strange repeating adjustments
Strange adjustments have been systematically applied to CRUTEM3 data to create CRUTEM4 data. For example, for a period of 22 years, from 1951 to 1972 Station 915540 (Vanuatu) has the following set of adjustments applied:
With the original CRUTEM3 data showing
1951 26.5 26.3 26.5 26.0 25.2 24.4 24.2 24.1 24.2 25.1 25.6 26.1
1952 26.8 26.6 26.3 25.6 24.8 23.9 23.0 22.6 22.8 23.8 24.5 24.8
1953 25.3 25.4 25.1 24.3 23.7 22.6 22.5 23.1 23.1 24.2 25.0 24.4
1954 25.2 25.1 24.8 24.2 23.6 23.3 23.3 22.2 23.5 23.7 23.9 24.8
1955 24.6 25.2 24.1 24.1 23.9 22.6 23.1 22.3 23.7 24.0 24.5 24.7
1956 24.9 25.6 24.8 24.3 23.9 23.7 22.4 22.5 24.1 24.4 24.4 25.7
1957 25.7 25.7 25.1 25.0 23.1 21.6 21.7 23.2 23.1 23.5 24.3 25.0
1958 25.6 26.3 25.7 24.9 23.9 23.6 21.5 22.0 23.9 23.8 24.1 25.4
1959 25.3 25.7 25.4 24.6 23.7 23.1 23.4 22.8 23.3 24.2 25.3 25.5
1960 24.9 24.8 24.9 24.4 24.1 23.6 22.6 23.3 24.0 23.8 24.7 24.4
1961 26.0 26.6 26.0 25.0 24.9 24.3 23.9 23.7 24.3 24.5 25.1 25.7
1962 26.1 25.8 25.2 24.9 24.7 23.8 23.3 24.1 23.9 24.4 24.7 25.3
1963 25.6 25.9 25.6 25.0 23.7 24.4 23.0 23.8 23.7 23.6 24.4 25.3
1964 25.8 26.5 26.2 25.3 24.5 24.4 23.2 24.0 23.9 24.5 25.5 25.4
1965 25.5 25.8 25.5 24.6 23.5 23.1 22.5 21.9 22.8 23.1 24.1 25.4
1966 25.8 25.9 25.8 25.1 23.5 23.4 22.4 23.0 23.3 23.8 24.3 24.5
1967 25.4 26.0 25.6 24.6 24.7 24.0 22.9 23.5 23.8 24.1 24.2 25.5
1968 25.5 26.0 25.4 24.5 24.0 23.5 22.9 22.6 23.2 24.1 24.5 25.3
1969 25.6 25.9 26.1 25.6 25.0 24.4 23.3 23.6 23.2 24.1 24.7 25.3
1970 25.9 26.0 26.5 25.1 24.4 23.8 23.6 24.2 24.0 24.5 24.5 25.5
1971 25.3 25.7 25.2 24.8 24.1 24.1 23.2 23.9 23.9 24.1 24.5 24.8
1972 25.3 25.6 25.2 24.8 25.1 23.7 22.4 21.7 23.1 24.0 25.2 25.5
And the original CRUTEM4 data showing
1951 26.9 26.7 26.9 27.0 26.0 25.0 25.0 25.3 25.2 25.9 26.4 26.7
1952 27.2 27.0 26.7 26.6 25.6 24.5 23.8 23.8 23.8 24.6 25.3 25.4
1953 25.7 25.8 25.5 25.3 24.5 23.2 23.3 24.3 24.1 25.0 25.8 25.0
1954 25.6 25.5 25.2 25.2 24.4 23.9 24.1 23.4 24.5 24.5 24.7 25.4
1955 25.0 25.6 24.5 25.1 24.7 23.2 23.9 23.5 24.7 24.8 25.3 25.3
1956 25.3 26.0 25.2 25.3 24.7 24.3 23.2 23.7 25.1 25.2 25.2 26.3
1957 26.1 26.1 25.5 26.0 23.9 22.2 22.5 24.4 24.1 24.3 25.1 25.6
1958 26.0 26.7 26.1 25.9 24.7 24.2 22.3 23.2 24.9 24.6 24.9 26.0
1959 25.7 26.1 25.8 25.6 24.5 23.7 24.2 24.0 24.3 25.0 26.1 26.1
1960 25.3 25.2 25.3 25.4 24.9 24.2 23.4 24.5 25.0 24.6 25.5 25.0
1961 26.4 27.0 26.4 26.0 25.7 24.9 24.7 24.9 25.3 25.3 25.9 26.3
1962 26.5 26.2 25.6 25.9 25.5 24.4 24.1 25.3 24.9 25.2 25.5 25.9
1963 26.0 26.3 26.0 26.0 24.5 25.0 23.8 25.0 24.7 24.4 25.2 25.9
1964 26.2 26.9 26.6 26.3 25.3 25.0 24.0 25.2 24.9 25.3 26.3 26.0
1965 25.9 26.2 25.9 25.6 24.3 23.7 23.3 23.1 23.8 23.9 24.9 26.0
1966 26.2 26.3 26.2 26.1 24.3 24.0 23.2 24.2 24.3 24.6 25.1 25.1
1967 25.8 26.4 26.0 25.6 25.5 24.6 23.7 24.7 24.8 24.9 25.0 26.1
1968 25.9 26.4 25.8 25.5 24.8 24.1 23.7 23.8 24.2 24.9 25.3 25.9
1969 26.0 26.3 26.5 26.6 25.8 25.0 24.1 24.8 24.2 24.9 25.5 25.9
1970 26.3 26.4 26.9 26.1 25.2 24.4 24.4 25.4 25.0 25.3 25.3 26.1
1971 25.7 26.1 25.6 25.8 24.9 24.7 24.0 25.1 24.9 24.9 25.3 25.4
1972 25.7 26.0 25.6 25.8 25.9 24.3 23.2 22.9 24.1 24.8 26.0 26.1
Similar adjustments can be seen in many other stations. The above Vanuatu data comes from a tropical area where temperature varies only 2 to 4oC over the year. In this situation, an adjustment of 0.4 to 1.2oC degree seems extreme. Furthermore, there are no adjustments after 1972. But the same type of adjustment appears in more temperate 946720 (Adelaide Australia). For the 30 year period 1857 to 1886 the following adjustment has been applied to CRUTEM3 to create CRUTEM4.
when the two Datasets show
CRUTEM3
1857 22.6 28.3 19.7 17.5 13.0 11.6 11.8 12.2 14.2 15.4 18.4 23.4
1858 26.4 24.0 22.3 18.3 13.5 12.0 10.3 11.7 12.6 15.8 21.1 22.5
1859 23.8 22.4 20.3 17.1 13.2 10.9 10.7 12.4 13.3 17.7 19.7 22.9
1860 26.1 23.8 22.4 17.0 14.4 12.5 11.7 13.4 15.5 17.0 19.9 22.5
1861 23.2 22.3 23.2 19.1 14.2 13.7 10.6 11.4 14.3 17.7 20.0 19.8
1862 25.4 22.4 23.0 17.0 14.9 11.9 13.2 13.1 15.3 18.6 21.4 23.6
1863 23.5 24.0 22.3 20.7 15.8 13.3 11.7 12.2 13.5 16.3 19.0 21.9
1864 23.0 21.9 21.1 18.1 15.3 11.7 11.5 11.7 15.4 16.1 20.6 21.0
1865 21.6 21.7 21.3 19.5 13.6 11.8 10.6 12.7 14.6 17.1 21.7 20.8
1866 23.9 25.2 21.2 19.0 15.8 12.6 11.6 12.6 13.8 16.6 17.8 22.1
1867 24.1 24.3 20.7 18.2 15.6 14.0 11.5 12.6 13.2 16.2 18.9 20.1
1868 20.6 22.9 23.4 17.8 15.8 12.1 10.4 12.1 14.6 17.9 20.2 22.0
1869 22.1 22.6 21.3 17.4 13.4 12.4 11.3 12.8 11.8 16.4 20.4 21.7
1870 23.4 25.3 21.7 18.7 13.9 12.7 10.8 11.5 12.9 17.2 18.0 21.9
1871 23.0 23.8 20.3 18.6 15.4 13.4 11.6 13.4 14.8 16.6 18.6 24.0
1872 25.9 23.3 22.4 17.1 13.5 12.6 11.1 9.9 14.1 16.2 21.0 20.0
1873 23.9 22.6 19.9 16.7 14.9 11.8 10.7 12.6 14.2 18.3 17.3 23.6
1874 24.0 21.7 19.8 19.4 14.0 11.7 9.8 11.5 12.1 17.6 17.9 21.9
1875 23.6 22.9 20.9 18.1 13.2 11.9 10.6 12.2 13.8 16.6 18.2 19.3
1876 22.8 22.1 23.8 16.5 13.3 10.8 10.1 11.5 13.5 16.1 18.6 23.4
1877 23.1 24.7 20.3 17.7 14.2 11.7 11.2 13.6 12.4 16.1 17.2 20.6
1878 25.6 23.0 21.6 18.3 14.1 9.9 11.7 12.9 14.3 17.5 20.1 21.4
1879 24.4 23.9 20.9 18.6 12.4 11.6 10.1 12.1 13.2 16.2 18.6 21.1
1880 25.6 26.4 21.5 17.5 13.7 12.1 10.4 12.9 13.7 15.4 17.8 21.9
1881 23.2 21.7 21.4 17.4 14.9 10.6 10.6 11.8 13.9 15.3 18.6 21.3
1882 22.9 24.0 22.6 17.4 15.1 10.6 9.6 11.2 14.0 16.9 20.8 21.6
1883 23.6 21.9 20.9 18.4 13.2 13.1 10.8 11.3 12.4 15.3 19.6 20.9
1884 21.2 23.6 22.1 17.1 13.9 12.2 9.8 13.3 14.1 15.5 18.9 19.5
1885 21.6 21.6 18.9 17.2 15.9 10.8 10.7 12.8 14.0 18.0 19.2 23.2
1886 24.4 20.6 20.4 17.3 14.0 11.6 11.6 12.4 16.1 14.7 19.9 21.9
CRUTEM4
1857 21.2 27.4 18.0 16.4 12.1 11.0 11.4 11.4 13.7 14.7 17.5 22.6
1858 25.0 23.1 20.6 17.2 12.6 11.4 9.9 10.9 12.1 15.1 20.2 21.7
1859 22.4 21.5 18.6 16.0 12.3 10.3 10.3 11.6 12.8 17.0 18.8 22.1
1860 24.7 22.9 20.7 15.9 13.5 11.9 11.3 12.6 15.0 16.3 19.0 21.7
1861 21.8 21.4 21.5 18.0 13.3 13.1 10.2 10.6 13.8 17.0 19.1 19.0
1862 24.0 21.5 21.3 15.9 14.0 11.3 12.8 12.3 14.8 17.9 20.5 22.8
1863 22.1 23.1 20.6 19.6 14.9 12.7 11.3 11.4 13.0 15.6 18.1 21.1
1864 21.6 21.0 19.4 17.0 14.4 11.1 11.1 10.9 14.9 15.4 19.7 20.2
1865 20.2 20.8 19.6 18.4 12.7 11.2 10.2 11.9 14.1 16.4 20.8 20.0
1866 22.5 24.3 19.5 17.9 14.9 12.0 11.2 11.8 13.3 15.9 16.9 21.3
1867 22.7 23.4 19.0 17.1 14.7 13.4 11.1 11.8 12.7 15.5 18.0 19.3
1868 19.2 22.0 21.7 16.7 14.9 11.5 10.0 11.3 14.1 17.2 19.3 21.2
1869 20.7 21.7 19.6 16.3 12.5 11.8 10.9 12.0 11.3 15.7 19.5 20.9
1870 22.0 24.4 20.0 17.6 13.0 12.1 10.4 10.7 12.4 16.5 17.1 21.1
1871 21.6 22.9 18.6 17.5 14.5 12.8 11.2 12.6 14.3 15.9 17.7 23.2
1872 24.5 22.4 20.7 16.0 12.6 12.0 10.7 9.1 13.6 15.5 20.1 19.2
1873 22.5 21.7 18.2 15.6 14.0 11.2 10.3 11.8 13.7 17.6 16.4 22.8
1874 22.6 20.8 18.1 18.3 13.1 11.1 9.4 10.7 11.6 16.9 17.0 21.1
1875 22.2 22.0 19.2 17.0 12.3 11.3 10.2 11.4 13.3 15.9 17.3 18.5
1876 21.4 21.2 22.1 15.4 12.4 10.2 9.7 10.7 13.0 15.4 17.7 22.6
1877 21.7 23.8 18.6 16.6 13.3 11.1 10.8 12.8 11.9 15.4 16.3 19.8
1878 24.2 22.1 19.9 17.2 13.2 9.3 11.3 12.1 13.8 16.8 19.2 20.6
1879 23.0 23.0 19.2 17.5 11.5 11.0 9.7 11.3 12.7 15.5 17.7 20.3
1880 24.2 25.5 19.8 16.4 12.8 11.5 10.0 12.1 13.2 14.7 16.9 21.1
1881 21.8 20.8 19.7 16.3 14.0 10.0 10.2 11.0 13.4 14.6 17.7 20.5
1882 21.5 23.1 20.9 16.3 14.2 10.0 9.2 10.4 13.5 16.2 19.9 20.8
1883 22.2 21.0 19.2 17.3 12.3 12.5 10.4 10.5 11.9 14.6 18.7 20.1
1884 19.8 22.7 20.4 16.0 13.0 11.6 9.4 12.5 13.6 14.8 18.0 18.7
1885 20.2 20.7 17.2 16.1 15.0 10.2 10.3 12.0 13.5 17.3 18.3 22.4
1886 23.0 19.7 18.7 16.2 13.1 11.0 11.2 11.6 15.6 14.0 19.0 21.1
The updates accentuate the global warming argument
There is pronounced tendency to flex the graph of global temperature vs time in a way which accentuates warming in recent years. However, the dominant effect is to lower temperatures prior to 1995.
The differences between CRUTEM3 and CRUTEM4, with zero tolerance, were consolidated to annual temperature differences, and plotted against time along with the count of the stations which contributed to the graph. This gives
The adjustments indicate that in CRUTEM4:
· A comparatively small number of stations have been cooled between about 1820 and 1900 anywhere up to 0.4oC.
· A substantial number of stations have been cooled about 0.1oC between about 1910 and 1995.
· Between about 100 and 400 stations have been warmed by up to about 0.2oC since about 1995.
This does not translate into the same change in the anomaly vs time graph, because it is not the whole dataset, but simply the changed data between CRUTEM3 and CRUTEM4. Nor does it take into account the effect of dropped and added stations. But it does indicate the direction of the change.
Nitpicking Differences
The nature of the CRUTEM database makes it obviously difficult to manage. Data comes from many sources. It is unreasonable to expect Hadley or UEA personnel to understand the geography of the data they receive. So errors will appear in the database. Some positional errors can be dismissed on the grounds that CRUTEM is directed at anomalies. This means that if a station’s position data is wrongly represented, then provided the error is contained within the same 5o v 5o gridcell, the anomaly is unaffected.
However, it is difficult to see how Station number 237070, name listed as “Unknown, Russia”, with a Lat/Long of -99.9/-999.9 (ie unknown), could escape being found by Quality Control, while still feeding temperatures to the gridding/anomaly calculation. Station 288020 is similarly identified, and also supplies data.
“Normals” are, by convention, calculated over 30 years from 1961 to 1990. While this appears to be observed, it is quite common to calculate Standard Deviations over a different period, commonly 1941 to 1990.
Extreme differences occur when the same station Number is used for two different locations. For example, In CRUTEM3, Station 840270 is the high altitude Tulcan El Rosal in Ecuador. But in CRUTEM4 station 840270 is Esmeraldas Tachina, also in Ecuador, but several hundred km distant from its CRUTEM3 namesake, and at almost sea level. This appears to be a CRUTEM3 error, probably difficult to find once the error is made.
Sydney Airport (947670) is still badly identified. The data runs from 1859 to the present, but only data after 1990 comes from Sydney Airport. The earlier data comes from Sydney Observatory, 10km away.
Effect of added and deleted stations
No attempt is made in this report to assess the effect of the 286 station deletions and the 738 additional stations, except to observe that most of the additions are high latitude NH stations.
Discussion
When updating a database such as CRUTEM, I would expect the steps to be roughly
· Tidy up the precursor database, which should be mostly stable data that has had years of scrutiny, and which would require limited correction or additions.
· Delete data which is considered poor quality.
· Add new data, edited and homogenised.
· Run a check as has been done in this report, looking at differences which might suggest irregularities.
But Hadley/CRU do not have appears to have done this. They have added and deleted stations, but it seems strange that very early data – 18th and 19th Century data, should be added, especially when much of it is sparse, of perhaps questionable quality and not germane to the current temperature/time/CO2 discussion.
But the biggest problem with the new HADCRUT4 database is the frequent practice of systematic, repetitive temperature modifications on blocks of station data. What reason could there be for making the following set of adjustments on Adelaide data for every year from 1857 to 1886 ?
The perpetrators of this change have clearly not gone back to original raw data and re-appraised their original homogenisation processes. They have simply taken every year of CRUTEM3 data (with it’s possible homogenisation adjustments), from 1857 to 1886, and
· Deducted 1.4oC if the month name was January,
· Deducted 0.9oC if February,
· Deducted 1.7oC if March
· And so on.
It is difficult to conceive what the justification could be for these changes. There are many instances of such changes in HADCRUT4, covering periods of 5 to 30 years, in both ancient and recent data. The adjustment vector differs in each, but the magnitudes of the individual elements are equally large. Their signs may vary.
The temperature differences, where there are corresponding data in each of the datasets works to promote the impression of a temperature surge in the 1990s.
It has done this by lowering temperatures before about 1995, and raising them thereafter.
The effect of this on the final gridded anomaly data has yet to be calculated. But it will be less than depicted above. The above graph shows the difference in the data induced by changes in data which appear in both CRUTEM3 and CRUTEM4. It does not include the effect of
· Dropped stations
· Added stations
· Stations which record a valid temperature in one set, but a Null (-99) in the other.
There are other relatively minor issues regarding data quality. I am surprised that neither CRU quality control nor users of the data have picked up these errors.
The CRUTEM4 database appears even more like one which has been assembled by amateurs with little concept of accuracy or integrity. If you were of a suspicious nature, you might get the impression that the changes are targeted at accentuating the Warmist message.
Until the matters raised in this report have been adequately explained, CRUTEM4, and therefore HADCRUT4 are seriously flawed.
Acknowledgements
Thank you Warwick Hughes (http://www.warwickhughes.com/blog/) for encouraging me to write this report, and for helpful suggestions on the construction of the report.
Data available
The original data is available at: http://www.cru.uea.ac.uk/cru/data/temperature/.
The data and processing programs used in preparation of this report are available from the author at thurstan@bigpond.net.au., or from Warwick Hughes blog http://www.warwickhughes.com/blog/ They are in two zipped packages.
Data
This comprises
· The CRUTEM3 and CRUTEM4 station data for all examples mentioned in this report.
· The output of a full CRUTEM run, with some analysis of the output. It will not run under XL97, which is limited to 64K rows, when this workbook holds over 400,000 rows.
This is about 26mb compressed.
Programs
· An Excel 2010 Workbook. The one that produces the output used in this report. It is a Visual Basic driven Excel workbook. It extracts stations missing from each dataset, and the differences. An untested XL97 version is included. It will not run the full CRUTEM database, as XL97 is restricted to 64K rows, and CRUTEM generates over 100,000. But it is possible to split the two sets of input into smaller parcels. Australia/New Zealand generated only about 650 rows of output.
· Instructions for use are on Worksheet “MAIN”.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
What is notable here is not just cooling of earlier temps and bumping up or recent ones. There is a general reduction in the long term variations, ie removal (reduction) of the natural cycles.
Note the pre-1940 warming gets reduced, the post war cooling get warmed etc. the only exception being the recent warming which gets boosted.
Similar play seems to be happening in the sea surface data:
http://judithcurry.com/2012/03/15/on-the-adjustments-to-the-hadsst3-data-set-2/
HadSST3 is the other half of HadCrut4
As noted in the article on HadSST3, these adjustments are entirely speculative and sometimes actually involve rewriting part of the temperature record that does not fit the preconceived ideas.
PS, also note how the significant late 19th century cooling gets warmed up.
This is a real trend that is seen in a number of different proxies such as 0-50m ocean temps and tide gauge readings, but does not fit the CO2 propaganda.
@John Whitman says:
Point 0. – Locate people in government, academia and the media who are ideologically gullible enough to believe in the concept of original sin; they will be necessary to voluntarily advocate your plans to control society.
Good point – It appears the Mayans, Aztecs, Druids, [SNIP: A swipe too far. We don’t really want to derail this thread, do we? -REP], Climatologists, Scientologists, etc. successively stumble on the fact that locating gullible, scared, vaguely guilty, compliant, ‘flocks’ is easier than one might suspect. Opportunistically inclined broadcasters, educators, and politicians needing gullible scared voters and audiences cannot help but participate in the frauds of the day.
“MattE says:
May 3, 2012 at 9:27 am (Edit)
It seems to me that they and GISS are trying to get rid of the 20th century bumps that don’t align with CO2. They cool the clearly non-CO2-induced rise in the 30s while quashing the cooling of the 70s.
##########################
As we collect more data the 30 WILL COOL.
The reason is pretty simple. in past datasets such as GHCN monthly the spatial distribution of stations is skewed to the northern hemisphere. You know that complaint people have about the amount of coverage ( too few stations ) well it does have an effect during certain times. the 30’s
is one of those times. As we add more data from the SH during the 30s you will see the global average come down. Why, because of polar amplification. The trend tends to be higher in the NH than the SH. during the 30s the sample is skewed to the NH. As we add stations ( remember the argument that we need more stations ) from other archives as we add more SH data the 30s will come down a bit.
AGW in action courtesy of the hadley center. (CRUTEM4)
http://www.woodfortrees.org/plot/hadcrut4gl/from:1995/plot/hadcrut4gl/from:1995/trend/plot/hadcrut3gl/from:1995/plot/hadcrut3gl/from:1995/trend
Must have been seriously hot over 2007 in the Arctic to increase the peak by 0.2c from HADCRUT3 to HADCRUT4.
Yet also the stronger El Nino in 1997/98 with the same change apparantly causes a decrease by 0.1c from HADCRUT3 to HADCRUT4.
The rate of warming is now similar between GISS and HADCRUT4.
http://www.woodfortrees.org/plot/hadcrut4gl/from:1995/plot/hadcrut4gl/from:1995/trend/plot/gistemp/from:1995/plot/gistemp/from:1995/trend/plot/rss/from:1995/plot/rss/from:1995/trend
One major problem and why the confidence in the first statement is due to the difference between these and RSS. RSS measures up to 82.5N and therefore only a very small region in the Arctic is missed by this satellite. The temperature difference if only explalined by changes in using extra Arctic data are nonsense because the amount of warming in the missing Arctic region would have to be over 50 to 100c warmer than normal most of the time to justify this change. This below shows the difference between GISS and RSS. The difference between HADCRUT4 is not yet shown, but has a similar rate compared with GISS to RSS.
http://img829.imageshack.us/img829/5412/gissvrssextarc.png
Was calculated using the data from graph below.
http://img854.imageshack.us/img854/7658/gissvrss19812010.png
The major adjustment that needs to be done throughout the data is for urban heat island temperature increase – which is not done.
One can argument how much one wants, the urban heat phenomenon is real. Think of the locations where we have thermometers – in cities which grew – and a growing city has a growing urban heat – it is logical if one puts 2 or more such UHI islands the resulting one will have a little higher urban heat island.
Now Steven Mosher and others will say that Berkeley proved that UHI is not creating any trend – they found even a bit cooling in the cities lately – but why?
Simply because all the database is contaminated with urban heat.
There is practically no location in the database that is not influenced in a way or another by the UHI. And why did they found a difference showing less growing in cities?
Because growing had stalled in many cities. Birmingham, Leeds Frankfurt, Vienna, Budapest – name those cities – are no longer growing as they did between 1800, 1900 and1950.
Exactly what Berkeley found – cities not warming as much as other locations since 1950s – is the fact that demographically human locations are not growing once the countries get at a certain level of energy comfort and civilization. UHI for American, Russian and European cities is indeed getting stationary or growing slowlier since growing of those cities stopped – some in 1950s some 1960s some later.
So what Berkeley found is UHI signal on trend in the database – exactly the opposite of what they say. And this UHI growing should be removed.
Once that is removed then we bring terrestrial measurement closer to satellite measurement.
Of course no warmista wants to do it as it will cut half or more of the temperature increase. This is why they all try to hide the UHI effect on temperature in all ways. And this is why they rely only on locally measured data which they can adjust how they please and not on satellite data which cannot be adjusted so easily.
And this is another reason why the Earth temperature is no longer growing “lately” as it did before as the population is no longer growing as it did before – UHI effect on trend is getting less significant.
Artificial adjustments to the temperature database will tend to reach a plateau. Temperatures beyond that will flatten and not increase. This is in effect what has happened to the temperature data over the past 12 years. CO2 goes up but the temperature remains the same. The only way to suggest an urgency to act is to lower the earlier numbers and show a continued upward slope of the temperature curve. Those in control of the temperature database cannot artificially adjust the current numbers higher, but, they can continue to adjust the earlier numbers lower. It seems to me the newer the database, the less credible the accuracy of the values. It looks like a new FOI needs to be filled right away while the gun is still smoking.
Mosher: “As we add more data from the SH during the 30s you will see the global average come down. Why, because of polar amplification. The trend tends to be higher in the NH than the SH. during the 30s the sample is skewed to the NH. As we add stations ( remember the argument that we need more stations ) from other archives as we add more SH data the 30s will come down a bit.”
So, we will not expect hemispherical trends to change in a specific direction with the addition of new stations?
Niels A Nielsen says:
May 3, 2012 at 1:18 pm (Edit)
Mosher: “As we add more data from the SH during the 30s you will see the global average come down. Why, because of polar amplification. The trend tends to be higher in the NH than the SH. during the 30s the sample is skewed to the NH. As we add stations ( remember the argument that we need more stations ) from other archives as we add more SH data the 30s will come down a bit.”
So, we will not expect hemispherical trends to change in a specific direction with the addition of new stations?
#############
how did you you get that from what I wrote? you can get changes in the hemispheres as you
add stations. It depends on where the stations are added and when they are added.
In general if you look at the distribution of places that are missing measurements, the expectation would be: as you add stations the past will generally cool and the present will
warm. Generally. Of course the devil is in the details. But notions that there is some kind
of “plot” to cool the past and warm the present do not hold up. My experience is that
whenever I add new data ( ghcn daily, gsod, colonial records, ) the general result is the
same. if the curves move at all they tend to move cooler in the past and warmer in the present.
Generally.
Of course none of these minor changes means anything. There was a LIA, the planet is warming. changes in the global temperature index are EXPECTED when you add more data.
We clamored to add more data. Now we have more data. Guess what? the answer changes.
slightly. The idea that you can look at changes in series and DEDUCE a conspiracy, isnt “skeptical” thinking at all. The deep irony here is that many of us asked for code and data.
we asked for more stations. Why? because we were skeptical. Now that we have more data some people run off and make all sorts of hyperbolic alarmist claims. “theres a plot!”
not very skeptical. kinda funny. kinda depressing.
Lesson: be careful what you wish for.
Lars P
“Now Steven Mosher and others will say that Berkeley proved that UHI is not creating any trend – they found even a bit cooling in the cities lately – but why?
Simply because all the database is contaminated with urban heat.
######################
1. Berkeley Proved no such thing.
2. The last study Zeke, nick stokes and I did, suggested a UHI trend from 1979-2010
That trend, about .04C per decade, is consistent with the handful of regional
studies of UHI which all show trend bias of .03C to .125C per decade over the same
period. It is REGIONALLY variable. One UHI does not fit all. In the SH, UHI is much
smaller. In china and japan and Korean building practice drives it higher.
The argument that the whole database is contaminated is lacking one thing: evidence.
Here is a good question
Are you familiar with CRN? the network that Anthony approves?
would you call them rural? or do you disagree with Anthony?
think hard, it might be a trick question
Mosher: “As we add more data from the SH during the 30s you will see the global average come down. Why, because of polar amplification. The trend tends to be higher in the NH than the SH. during the 30s the sample is skewed to the NH. As we add stations ( remember the argument that we need more stations ) from other archives as we add more SH data the 30s will come down a bit.”
Not according to CRU. The 30’s do not seem to be down in the new CRU data set. Not globally and not in the southern hemisphere either:
http://www.woodfortrees.org/plot/hadcrut4sh/from:1920/plot/hadcrut3vsh/from:1920/plot/hadcrut4sh/from:1920/trend/plot/hadcrut3vsh/from:1920/trend
Niels,
As I said, the 30s. that does not mean you calculate a trend from 1920 on. That would be
“the trend from the 20s”
You look at the 1930 to 1940
Then you calculate the trend during that time
http://www.woodfortrees.org/plot/hadcrut4sh/from:1930/to:1940/plot/hadcrut3vsh/from:1930/to:1940/plot/hadcrut4sh/from:1930/to:1940/trend/plot/hadcrut3vsh/from:1930/to:1940/trend
And you see what I am saying GENERALLY. GENERALLY, on average, over a number of different datasets, when you add data the general effect will be to lower the trend in the 30s.
and you see the same thing in the NH, albiet smaller in scale.
http://www.woodfortrees.org/plot/hadcrut4nh/from:1930/to:1940/plot/hadcrut3vnh/from:1930/to:1940/plot/hadcrut4nh/from:1930/to:1940/trend/plot/hadcrut3vnh/from:1930/to:1940/trend
Just to show that I am not biased, DMI shows the year 1998 cooler than 2007 (80N+), for those that didn’t know.
http://ocean.dmi.dk/arctic/meant80n.uk.php
(just click on 1998 and 2007 to compare)
Although the difference being a tiny fraction of what this graph shows would be required.
http://img829.imageshack.us/img829/5412/gissvrssextarc.png
Steven Mosher, you yourself, as a professional scientist and a logical person, already believe in conspiracy – conspiracy demands only ONE man or woman who, in a weak moment, decides that “the story” is more important than the whole truth… What are the odds that it is only one person? You would have to say, “Unlikely. People are imperfect.”
Either all men and women in science have no weak moments (HA!), or conspiracy theory is simply human nature and happens in ALL sciences, always to the detriment of knowledge and advancement…
If all science was under the white hot spotlight like climatology, a great many scientists would be disgraced, especially considering most sciences lack the vast and mostly unquestioning support of most major medias…
Stephen Rasey says:
May 3, 2012 at 3:17 pm (Edit)
Re: Steven Mosher at 1:51 pm: think hard, it might be a trick question
Yes, indeed. It is a trick question. For Steven Mosher’s idea of “very rural” is hardly what I think of as very rural: See CA “Berkley ‘Very Rural’ Data Dec 20, 2011, 12:01 pm to Jan 1, 2012 12:36pm
#############
Typical response.
My question is do you accept Anthony’s endorsement of CRN as good stations?
For me. I do the following.
1. People suggest (pielke for example) that land use changes are important
So I look at land use using standard metrics a) nighlights, b)percentage of land
that is concrete. c) percentage of land that is “built”, is the land inside the admistrative
boundaries? outside? distance from the urban center, airport no airport. you name it
2. Try to combine all metrics
3. Use a pattern matching approach: you say CRN is rural, I’ll go find sites that match
the rural characteristics.
4. Use stations that others point to as rural. hey we know there is UHI because somebody compared urban to rural, so I use those rural as a “rule” for what counts as rural.
5. use population.
When I have those metrics I do sensitivity analysis.
Example: Hansen says if nighlights is less than 30, its rural. I dont do that
I ask?
Whats the difference between rural and urban IF we define rural as nighlights = 0,
nightlights = 1,2,3,4,5,6,7,8,9,10….
I do the same thing for any and every metric that people can throw at me.
BUT, it has to be an objective metric. measurable. not your sisters opinion
Re: Steven Mosher at 1:51 pm: think hard, it might be a trick question
Yes, indeed. It is a trick question. For Steven Mosher’s idea of “very rural” is hardly what I think of as very rural: See CA “Berkley ‘Very Rural’ Data Dec 20, 2011, 12:01 pm to Jan 1, 2012 12:36pm
Mosher: “And you see what I am saying GENERALLY. GENERALLY, on average, over a number of different datasets, when you add data the general effect will be to lower the trend in the 30s.”
That’s not what you said above: “As we add more data from the SH during the 30s you will see the global average come down. Why, because of polar amplification. The trend tends to be higher in the NH than the SH. during the 30s the sample is skewed to the NH. As we add stations ( remember the argument that we need more stations ) from other archives as we add more SH data the 30s will come down a bit.” You are talking about the _global average_ – not the trend. You say that the effect of polar amplification from adding stations should be to lower the the global _average_ in the 30’s compared to the present. Which makes sense to me. But that is not what the hadcrut4 shows.
You also said above “As we collect more data the 30 WILL COOL.” and “In general if you look at the distribution of places that are missing measurements, the expectation would be: as you add stations the past will generally cool and the present will warm.”
Surely, you are not going to tell me that you by “the past” meant the 30’s and “the present” the 40’s 🙂
My point is that you don’t see the 30’s cooling in hadcru4 compared to hadcru3 as you claimed above.
Correction: …Surely, you are not going to tell me that you by “the past” meant the 1930 and “the present” 1940 🙂
cartoonasaur says:
May 3, 2012 at 2:52 pm (Edit)
Steven Mosher, you yourself, as a professional scientist and a logical person, already believe in conspiracy – conspiracy demands only ONE man or woman who, in a weak moment, decides that “the story” is more important than the whole truth…
##############
if you want to redefine the word conspiracy to mean purple, then I suppose I believe in purple.
I will put it simply. There is no evidence whatsoever that any one person consciously and with intent changed temperature data to give the result they wanted.
If you have the evidence, and understand what would count as evidence of this conscious intent then let me know.
The closest case would be the blip in SSTs
I have found a similar hijacking of the historical record: http://endisnighnot.blogspot.co.uk/2012/03/giss-strange-anomalies.html
These cheeky monkeys at GISS and CRU are lowering older temperatures in order to create a spurious warming trend. Is this legal?
Mosher,
Can you show the trend of the dropped stations versus the newly added ones?
I haven’t seen this analysis done yet.
Best is supposed to have included all of them for Land temperatures. Is this true?
NHills says:
May 3, 2012 at 6:21 am
Regarding “Climate and weather of Australia by H.A.hunt and Griffith Taylor and E.T.Quayle” , looks like you can get it here
http://openlibrary.org/books/OL7221939M/The_climate_and_weather_of_Australia
pdf, epub, read online.
Thanks, yes that is the same perforated copy (University of California) I had printed off many of the pages while doing research on historical data/writings relating to floods and droughts in Australia, and within that report is a great deal of information on early documented reports before the Australian Bureau of Meteorology was formally set up. The historical hydrological information is also “interesting” in view of the ‘modern” C.S.I.R.O./B.O.M. propaganda that came out supporting Flannery’s claim that we would see decreasing rainfall patterns in Australia (now flooding of course) instead of confirming the weather variability that we know applies to Australia.
In that more complete copy that you linked, It is interesting when you flip through to the end, the library lending record shows it to be then held at the UCL Berkley and appears to have been last accessed in the 1960″s. I wonder if the copy is still in that Library. (Mosher and Nick Stokes should be able to confirm!)
There is a wealth of information within that booklet for more detailed examination by those studying the historical Australian climate record. I commend the work being done by some dedicated Australians to try and undo the vandalism to our Australian Climate History, especially those who have been successful in digging out old photographs of Stevenson screens in situ in the 1800’s and , confirming their use.
Also useful is the actual B.O.M (Australia) 100 year official History “The Weather Watchers” by David Day first published in 2007 ISBN 9780522852752 and its neat graphical depictions of each years weather variability maps for the years 1900 to 2005. I commend the title as it fits neatly to this subject – who is now watching over our precious weather watching record. Thanks Anthony
for your dedication and to those Australians who really care about integrity in climate science.
I also studied differences between CRUTEM4 and 3 and agree with most of what Ed writes. The most obvious difference is that many new stations have been added , while many others have been dropped. There are now 5549 stations in the set compared to 5097 in CRUTEM3. 628 new stations have been added while 176 stations have been discarded. All the new stations are in far northern latitudes around the arctic. Many stations in The US have been dropped.
Anyone who looks in to this data should also be aware that many stations have changed numbering ! This can cause huge confusion when comparing station by station. A list of all the renumbering can be found at http://clivebest.com/data/changed.txt. Details of where the stations are located are at http://clivebest.com/blog/?p=3493 and the next post.
Clivebest
Thank you for that note. I missed your original post, and I did not think of the possibility you raise.
I have just compared the two databases for exact matches on Lat/Long. I get 122. You got 124.
This means that 122/4 stations should get aliases so they will be treated as matching stations, and be treated accordingly. I don’t think it changes the tenor of my argument much, if at all when I include the effect of deletions/replacements. But I will nut out how to do it over the next few days, and figure what other implications it has before I introduce aliases..
Thank you.
http://woodfortrees.org/plot/hadcrut3vgl/last:180/trend/plot/hadcrut3vgl/last:180/plot/hadcrut4gl/last:180/plot/hadcrut4gl/last:180/trend
@Mosher: The argument that the whole database is contaminated is lacking one thing: evidence.
…
My question is do you accept Anthony’s endorsement of CRN as good stations?
Here are some of the things Anthony said about CRN wrote about CRN on April 7, 2012. (You provided no link, so I attempted to find someting…)
I note that there are only 4 instances of the word ‘rural’ in the thread, none by Anthony. I also note that there a comparative handfull, at pressent only 114, of these CRN stations.
I also note the following from NCDC on the need for CRN:
I accept Anthony’s endorcement of the CRN project as an improvement of USHCN modernization. I even accept that the few stations that he visited are absent of UHI — today. How well that quality is maintained over time is anyone’s guess. That Anthony is undertaking a privately financed, but sorely needed, project to disseminate the data to the public is not one of CRN strong points.
Ok. I accept Anthony’s “endorcement”. CRN is good. It is an improvement. What does that say about the crisis of non-CRN data that comprises the bulk of the whole database use in analytical work?
Lars P. says:
May 3, 2012 at 12:55 pm
/////////////////
I find your comments to be nn the right track.
The only important data is ocean temperature but this is defficient due in no small part to woefully inadequate coverage and short time scale.
All land based temperature measurements should be ditched. The various data series are now too contaiminated to shed any useful light on anything, and due to loss of original raw data it would appear that these are now beyond repair. BEST should have gone back to raw data and only used data that they were 100% certain was raw and had not already been the subject matter of adjustments. Whenever there has been equipment change, siting change etc, the record should stop. There should be no attempt to make it a continuous record by making some adjustment to supposedly account for the change that took place. The state of the various data sets is such that we are today essentially merely reviewing trends induced by adjustments rather than trends which arise truly from the data.
We should only be looking at satellite data. This shows no warming these past 30 or so years, just a step change around the super El Nino of 1998. This data strongly suggests that there has been no CO2 induced warming during the satellite period.
This raises a number of interesting questions, such as why has the temperature that was released around 1998 not dissipated? Is this because of so called GHGs which have ‘trapped’ or delayed the dissipation of this temperature, or is there some other explanation as to why temperatures remain high? What conditions need to be met for this temperature to dissipate, and over what length of time can we expect the dissipation to take place, and to what base level will temperatures return? Unfortunately, these questions are difficult to answer due to lack of knowledge and understanding of the many natural processes involved.
O/T, the Daily Mail is running an article which suggests that a new report suggests that Greenland Glacier melt is taking place at a far slower pace than had previously been thought and as a consequence sea level rise due to Greenland Glacier melt has been vastly over-estimated.