Stratospheric temperatures over the Arctic: Comparison of three data sets

Three sets of stratospheric temperatures (at the 50- and 30-hPa level) are compared for middle and particularly high northern latitudes. Two of the data sets are re-analyses (NCEP/NCAR and ERA40) while the FU-Berlin data are historical hand-analyses. The time period covered is September 1957 till June 2001, because the FU-Berlin data span this period. As the variability and any trends over the Arctic are of profound interest in connection with Global Change and the ozone problem, it is very important to see if and how these data sets agree and to what extend the early data, before satellites became available, are reliable. Therefore, 30-hPa temperatures over the North Pole and 50-hPa temperatures at 80 degrees N are compared, i.e. long-term monthly means, standard deviations and trends. In such a comparison one has to remember that the natural variability is very high during the arctic winter and that radiosonde stations are not available directly over the North Pole nor over the Arctic Sea. Further, the North Pole point is just one point and one shouldn't expect absolute agreement, particularly as we are dealing with three completely different analysis schemes. Considering these caveats, the agreement of the long-term mean temperatures and of the trends is in general surprisingly good. Of special interest is the agreement in the change of the sign of the trends between the two sub-periods during winter: During the first sub-period cooling is observed in December and warming in March and April. But during the second sub-period warming is observed in December and strong cooling from February till April. Therefore, the overall trend is practically zero. Time series of the original monthly mean 30-hPa temperatures are compared directly for December and March and it is of great interest to see the generally excellent agreement of most of the extremes in the analyses of FUB and NCEP throughout the whole period. This is valid also for ERA40 in the first sub-period (before satellite data became available). This gives confidence in the analyses of the early period.