date: 02 Dec 1996 17:16:33 +0000 from: Trausti Jonsson ( Trausti Jonsson ) subject: NAO paper to: p.jones@uea.ac.uk Dear Phil, here at last there is a first draft of the "Icelandic chapters" in the NAO paper. I will probably make some changes during the next few days, so don't bother with my hesitant English at this point, we can clear that up at a later stage. But if you think something is missing? I should be finished with the "splicing" soon and the other tables (i.e. the tables of SW-Iceland pressure series derivation and the table of early Icelandic instrumental series) should be ready later this week. Best wishes, Trausti Extension to the NAO using.... Some comments and additions: I have been busy with the "cleaning" of the data series. I am now fairly confident in the series from 1857 to the present. I will suggest that we call it the SW-Iceland pressure series. Its construction is rather simple. From 1921 onwards it is simply the Reykjavík pressure. I also use the Reykjavik measurements during the period 1893-1905 directly. During the period prior to 1893 I use the Stykkishólmur series with a seasonal (i.e. monthly) constant subtracted as you suggested. During the 1906 - 1920 period I used the formula (Akureyri pp + Vestmannaeyjar pp)/2 - constant. I also made a minor adjustment to the Akureyri series during this period. By using this procedure we can circumvent the period 1894 - 1920 that is problematic in Stykkishólmur due to ill-documented changes in the station location. This "instability" can easily been seen in difference series and is greater than corresponding problems for Akureyri. I compared all of this series and some combinations with the independent Eastern Iceland series. This last series extends from 1874-1960 and I have extended it to the present. How much of this cleaning procedure should be included in the paper I don't know. I tend to think that it should be mentioned but not in extenso. But on the other hand there is certainly an interest in a corrected pressure series like this. I am not completely finished with the around 1850 splicing of the two series. As I told you before there are some problems. I still think that the "final" version of the Reykjavík 1820-54 series is slightly too low. There are problems in this early part of the Stykkishólmur series. During the first two years there are no separate barometer temperature measurements. In 1852 something happens that I have not found out what is. In late 1856 there is a new barometer, but the old is also read for comparison for 10 years. There is a substantial difference between the two barometers (almost 7mb). There are many published and independently compiled tables of the pressure during this period, WWR, Scottish Met. Soc., myself, as well as many old calculations of different sorts. These rarely agree on exact values. I will have to go back to the original journals. It is very lucky that I now have access to the Akureyri measurements from 1847-1854 to assist in my fog. I will sort this out in a few days and then send you the complete table. 1. Introduction: comment: The Akureyri pressure data has actually been published back to 1874, why Rogers (1984) stopped in 1882 I don't know. Monthly data from the years 1847-1854 from Akureyri are also available and last week I found daily pressure observations that were probably made with the same instrument from 1854 to the 1880's. The Akureyri series could thus conceivably be extended all the way back to 1847. The part covering 1854 - 1860 will be digitized within the ADVICE project. 2.2. Icelandic data, a comment on sources. My 1992 lectures are now completely outdated and a mention of them can be omitted. A slightly newer (and better) version is to be found in a volume in the series "European Paleoclimate and Man", that has been in print for two years, but is really going to be published now around Christmas, I am told. We can include the reference when it has arrived. 2.2. Icelandic data The first Icelandic record in major climatological databases (e.g. Bradley et al., 1985 and WWR) was taken at Stykkisholmur in northwest Iceand from November 1845. Earlier instrumental data have been known to have been taken and summary of possible sources/sites is given by Jonsson (1994). Since this publication additional sources have been found. The location of the early records/diaries are listed in Table 2 with a brief assessment of the variables measured and the periods. A number of records are being digitized with more information given in Jonsson (1994). Then unchanged... 3.2. SW-Iceland pressure series. 3.2.1. The series after 1853. The pressure series presented here is a composite one with a main emphasis on the quality of the subseries used. As mentioned before the Stykkisholmur series extends backwards to November 1845, excluding a gap of 5 months in 1919. The Akureyri pressure series in northern Iceland extends back to 1874 with a gap of almost two years in 1918 - 1919. The East-Iceland Teigarhorn series covers the period of 1874 - 1960 and the Vestmannaeyjar in the south back to 1881. Monthly values for all of these stations have been published in the year books of the Icelandic Met. Office and (prior to 1920) in the year books of the Danish Met. Institute. These published values were all compared and the appropriate corrections to sea level and 45°N made and a conversion to mb where necessary. The comparison revealed a number of inhomogeneities. In the SW-Iceland presented here these have been adjusted for. The exact derivation of values is summarized in Table XX. In almost all cases (after 1853) these differ only slightly from the WWR publication except for the systematic difference of Reykjavik and Stykkisholmur. 3.2.2. The Reykjavik/Nes series. During the period 1 August 1820 to 28 February Jon Thorsteinsson, at the time a medical doctor with the title of Country Physicus, made regular meteorological observations in the Reykjavik area in Iceland. These observations were made at least daily and the complete observational lists have survived, except for a 6 month period September 1821 - February 1822 which have perished. There are pressure measurements during the whole period. The barometer was placed by a window in an unheated room. After 1841 there were up to 9 barometric readings per day. The daily observations have been digitized. The observations of Thorsteinsson were made under the auspices of the Danish Scientific Society (Videnskabernes Selskab) and started on 1 August at Thorsteinsson home in Reykjavik. In July 1821 Thorsteinsson moved to the location Nes a few kilometers to the west of the Reykjavik side. The house at Nes is still standing. He then moved back to Reykjavik on 18 October 1833, but the exact location of the station within the old town is not known at the present time. During this period of observation there were radical changes in the exposure and reading practice of the temperature measurements, but the pressure measurements seem to be relatively homogenous. In the beginning there was a barometer with a french inch scale in use and an attached thermometer in °R. The temperature of the barometer is always noted until 29 February 1852. After the relocation to Nes the barometer was located in an unheated brick house near a window facing east. The window was always open, except during storms. At Nes the height of the barometer was about 14m a.s.l., but 10m in Reykjavik after the 1833 relocation. The height during the first Reykjavik period is uncertain, but probably not far from 10m. According to the notes of Thorsteinsson there are three barometers in use during the whole period. These were in good agreement according to the notes. During a period after 1835 there is a french mobile barometer (Syphon) in use instead of the older Repsold one, which at that time was sent to Paris for calibration and cleaning. According to Thorsteinsson the Syphon barometer was consistently about 1mm higher than the other, but this was corrected for prior to the journal entry. The older barometer returned after an unknown time. Later (probably in 1841 or 1842) the barometer in use seems to have been graded in mmHg, but converted to french inches in the journal. This scale was in use until 29 February 1852, when there was suddenly a change to mmHg in the journal. During most of the period observations were only noted once daily, about 9 in the morning. After 1841 there are multiple observations pr. day noted in a special diary. This additional diary is not used here and the monthly averages presented are made by the use of only one observation per day. At the coasts of Iceland any regular diurnal variation is negliable. The most critical aspect for using the observation is the lack of information on corrections. Except for the correction of the Syphon instrument cited above there is no information if any corrections were employed before the observation was written down. It is very likely, however, that no pre-entry temperature correction was made at least until the change of observation practice in the 1840s. Information on "index" errors are nonexistent. Corrections for gravity and height were usually not practiced and it can be fairly safely assumed that these are not included in the original tables. The daily values were corrected individually for instrument temperature. A table for these corrections was generated by the help of the daily observations published in Holmsted (1859). A constant height correction was used at each site as well as a constant gravity correction. The errors generated by these assumptions of constancy are negliable compared to other uncertainties. After 1 March 1852 there is no note of the instrument temperature. During this period the monthly average correction during the 1846 to 1851 period was used for the temperature reduction. Each observation was checked against the neigbouring values in time and a number (about 50) of errors corrected. Only overwhelmingly "unlikely" values were corrected. Usually these were clearly due to "unnatural" jumps of whole inches in otherwise seemingly stable weather conditions. A comparison was made of the day-to-day variability of the series compared to the "normal" variability of recent years in Reykjavik. The values are similar. (A table??). The average annual value is on the low side. ----- ((I am trying to cope with this and will hopefully reach a definite conclusion in a few days)).