Identification of atmospheric and oceanic teleconnection patterns in a 20-year global data set of the atmospheric water vapour column measured from satellites in the visible spectral range

Abstract

Abstract. We used a global long-term (1995–2015) data set of total column\nwater vapour (TCWV) derived from satellite observations to quantify to which\nextent the temporal patterns of various teleconnections can be identified in\nthis data set. To our knowledge, such a comprehensive global TCWV data set\nwas rarely used for teleconnection studies. One important property of the\nTCWV data set is that it is purely based on observational data. We developed\na new empirical method to decide whether a teleconnection index is\nsignificantly detected in the global data set. We compared our new method to\nwell-established hypothesis tests and found good agreement with the results\nof our approach. Based on our empirical method more than 40 teleconnection\nindices were significantly detected in the global TCWV data set derived from\nsatellite observations. In addition to the satellite data we also applied\nour method to other global data sets derived from ERA-Interim. One important\nfinding is that the spatial patterns obtained for the ERA TCWV data are very\nsimilar to the observational TCWV data set indicating a high consistency\nbetween the satellite and ERA data. Moreover, similar results are also found\nfor two selections of ERA data (either all data or mainly clear-sky data).\nThis finding indicates that the clear-sky bias of the satellite data set is\nnegligible for the results of this study. However, for some indices, also\nsystematic differences in the spatial patterns between the satellite and\nmodel data set were found probably indicating possible shortcomings in the\nmodel data. For most “traditional” teleconnection data sets (surface\ntemperature, surface pressure, geopotential heights and meridional winds at\ndifferent altitudes) a smaller number of significant teleconnection indices\nwas found than for the TCWV data sets, while for zonal winds at different\naltitudes, the number of significant teleconnection indices (up to\n > \u200950) was higher. The strongest teleconnection signals were found\nin the data sets of tropospheric geopotential heights and surface pressure.\nIn all global data sets, no “other indices” (solar variability,\nstratospheric AOD or hurricane frequency) were significantly detected. Since\nmany teleconnection indices are strongly correlated, we also applied our\nmethod to a set of orthogonalised indices, which represent the dominant\nindependent temporal teleconnection patterns. The number of significantly\ndetected orthogonalised indices (20) was found to be much smaller than for\nthe original indices (42). Based on the orthogonalised indices we derived\nthe global spatial distribution of the cumulative effect of teleconnections.\nThe strongest effect on the TCWV is found in the tropics and high latitudes.