Überprüfung der Datenreihe der Jahresmitteltemperaturen für Halle (Saale) von 1851 bis 2015 auf Homogenität

Abstract

Schadwill, H., Döring, J., Wolter, M., Borg, H.: Examination of the annual mean air temperaturesfor Halle (Saale) from 1851 to 2015 for homogeneity. - Hercynia N. F (2017): 115 –133.

Sound statements about past and future developments of our climate require long data records. In longrecords it is not uncommon that the data are distorted by factors such as relocation of the measurementsite, changes in the instrumentation, or urbanisation. This must be examined to preclude wrong interpretationsof the data. To recognize inhomogeneities in a record the data of interest must be compared todata from other sites which are not too far away and free of the aforementioned distortions.In this study the record of the annual mean temperatures for Halle (Saale) which dates back to 1851 wasexamined. Since then the measurement site was relocated twice (at the end of 1900 and 1953) and theinstrumentation was changed several times. Two weather stations were selected for comparison: Theone in Potsdam which was never moved since data collection began in 1893, and the one in Lindenbergwhich was moved twice, but only by a few metres, and whose record commenced in 1907.The record for these two stations is much shorter than for Halle. To be able to go back until 1851 the dataseries for Berlin which starts 1851, too, was utilized as well. However, the recording site in Berlin wasmoved in March 1908 and thrice thereafter. Berlin is therefore not a good choice as a reference for theyears from 1908 onwards and is only fully considered here for the period from 1851 to 1907.To examine the data for Halle the differences in the annual mean temperatures between Halle and thereference stations Berlin, Potsdam und Lindenberg were computed. There was no hint that the relocationof the station in Halle at the end of 1900 had any effect on the data. This would have been visible in thecomparison of the Halle data to those from Berlin or Potsdam. In contrast, with respect to all three referencestations there was a sudden drop in the temperature differences from 1953 to 1954. This occurreddue to the relocation of the station in Halle from the city to Kröllwitz, a moderately built-up suburbwith lots of green areas, and the continued registration of the measured temperatures without taking thedifferent surroundings of the new location into account. Before and after this drop the data record forHalle is alright.To get a consistent data record the annual mean air temperatures for Halle prior until 1954 were adjustedbased on the data for Potsdam. First, a representative mean of the temperature differences betweenHalle and Potsdam for the periods 1893 to 1953 and 1954 to 2015 was calculated. Then, the differencebetween these two means was determined (0.4 K) and subtracted from the data measured in Halle from1851 to 1953. This eliminated the aforementioned inhomogeneity. The resulting annual mean air temperaturesfor Halle prior to 1954 are now 0.4 K lower than before. Due to this adjustment a consistentdata record is now available for Halle from 1851 to 2015.According to the original data four of the ten warmest years occurred since 1990, and six before 1954with 1934 as the warmest year overall. In the corrected data eight of the ten warmest years appear since1990, and 1934 is now only the second-warmest year, together with the year 2000. This demonstratesthat it is absolutely necessary to examine data records for homogeneity and correct them if necessary to avoid a misinterpretation of the climatic development.avoid a misinterpretation of the climatic development.

Schadwill, H., Döring, J., Wolter, M., Borg, H.: Examination of the annual mean air temperaturesfor Halle (Saale) from 1851 to 2015 for homogeneity. - Hercynia N. F (2017): 115 –133.

Sound statements about past and future developments of our climate require long data records. In longrecords it is not uncommon that the data are distorted by factors such as relocation of the measurementsite, changes in the instrumentation, or urbanisation. This must be examined to preclude wrong interpretationsof the data. To recognize inhomogeneities in a record the data of interest must be compared todata from other sites which are not too far away and free of the aforementioned distortions.In this study the record of the annual mean temperatures for Halle (Saale) which dates back to 1851 wasexamined. Since then the measurement site was relocated twice (at the end of 1900 and 1953) and theinstrumentation was changed several times. Two weather stations were selected for comparison: Theone in Potsdam which was never moved since data collection began in 1893, and the one in Lindenbergwhich was moved twice, but only by a few metres, and whose record commenced in 1907.The record for these two stations is much shorter than for Halle. To be able to go back until 1851 the dataseries for Berlin which starts 1851, too, was utilized as well. However, the recording site in Berlin wasmoved in March 1908 and thrice thereafter. Berlin is therefore not a good choice as a reference for theyears from 1908 onwards and is only fully considered here for the period from 1851 to 1907.To examine the data for Halle the differences in the annual mean temperatures between Halle and thereference stations Berlin, Potsdam und Lindenberg were computed. There was no hint that the relocationof the station in Halle at the end of 1900 had any effect on the data. This would have been visible in thecomparison of the Halle data to those from Berlin or Potsdam. In contrast, with respect to all three referencestations there was a sudden drop in the temperature differences from 1953 to 1954. This occurreddue to the relocation of the station in Halle from the city to Kröllwitz, a moderately built-up suburbwith lots of green areas, and the continued registration of the measured temperatures without taking thedifferent surroundings of the new location into account. Before and after this drop the data record forHalle is alright.To get a consistent data record the annual mean air temperatures for Halle prior until 1954 were adjustedbased on the data for Potsdam. First, a representative mean of the temperature differences betweenHalle and Potsdam for the periods 1893 to 1953 and 1954 to 2015 was calculated. Then, the differencebetween these two means was determined (0.4 K) and subtracted from the data measured in Halle from1851 to 1953. This eliminated the aforementioned inhomogeneity. The resulting annual mean air temperaturesfor Halle prior to 1954 are now 0.4 K lower than before. Due to this adjustment a consistentdata record is now available for Halle from 1851 to 2015.According to the original data four of the ten warmest years occurred since 1990, and six before 1954with 1934 as the warmest year overall. In the corrected data eight of the ten warmest years appear since1990, and 1934 is now only the second-warmest year, together with the year 2000. This demonstratesthat it is absolutely necessary to examine data records for homogeneity and correct them if necessary to avoid a misinterpretation of the climatic development.