Climatology of free tropospheric humidity: extension into the SEVIRI era, evaluation and exemplary analysis

Abstract

Abstract. A new free tropospheric humidity (FTH) data record is presented. It is based on observations of Meteosat-2–5 and Meteosat-7 Meteosat Visible and Infrared Imager (MVIRI) and Meteosat-8 and -9 Spinning Enhanced Visible and Infrared Imager (SEVIRI) at the water absorption band at 6.3 μm. With the extension to SEVIRI observations the data record now covers the period 1983–2009 with a spatial and temporal resolution of 0.625° and 3 h, respectively. The data record is referenced under digital object identifier (doi): 10.5676/EUM_SAF_CM/FTH_METEOSAT/V001 and is freely available from http://www.cmsaf.eu/wui . The relation between the observed brightness temperature (BT) and FTH is well established: the observed BT is proportional to the logarithm of the mean relative humidity (RH). Under the given assumptions, constant lapse rate and random strong line theory, it means that the observed BT is mainly a function of RH alone and not of temperature and specific humidity separately. Here, existing retrievals have been refined mainly through the consideration of relative humidity Jacobians in the training process of the statistical retrieval. The temporal coverage has been extended into the SEVIRI era, the homogenisation of the BT record has been improved and the full archive has been reprocessed using updated regression coefficients. The FTH product is compared against FTH computed on the basis of the Analysed RadioSoundings Archive (ARSA) observations. An average relative bias and root mean square difference (RMSD) of −3.2 and 16.8%, respectively, are observed. The RMSD confirms the expectation from an analysis of the total uncertainty of the FTH product. The decadal stability is 0.5 ± 0.45% per decade. As exemplary applications the inter-annual standard deviation, differences on decadal scales and the linear trend in the FTH data record and the frequency of occurrence of FTH <10% (FTHp10) are analysed per season. Maxima in inter-annual standard deviations as well as maxima in absolute differences occur in gradient areas between dry and wet regions and areas with minima in FTH and maxima in FTHp10. An analysis of the linear trends and associated uncertainty estimates has been attempted to identify possible problems with the data record. Positive trends in FTHp10 coincide with gradient areas and regions of minimum FTH, maximum FTHp10 as well as with negative differences between decadal FTHp10 averages of the 1990s and 2000s. However, they are accompanied by maximum standard deviation and are therefore hardly significant which is also valid for FTH trend estimates. These activities plus inter-comparisons to other humidity data records are part of the Global Energy and Water Exchanges Project (GEWEX) water vapor assessment (G-VAP) and will be extended to other FTH data records in the near future.