In the present paper the authors analyse the drought occurrence over the European region by using the NCEP/NCAR reanalysis precipitation rates covering the period from 1948 to 2000. The drought assessment is based on the Standardized Precipitation Index (SPI), which has been proposed as an indicator of drought condition. At variance with other fields derived from precipitation, the SPI is, by construction, a Gaussian field. Thus, the understanding of its covariance structure exhausts the study of the associated density distribution. A method allowing a factorisation of a multivariate Gaussian distribution is the one known as Principal Component Analysis (PCA) or Kauman-Loeve decomposition. Therefore, a PCA is used to study the main spatial patterns and the time variability of drought first over Europe and then over the Northern Hemisphere. The analysis reveals a downward trend for the index over most of central Europe and the Mediterranean basin, implying an overall decrease of precipitation in the above mentioned regions. Moreover, the scores associated with the PCA covariance decomposition, besides the aforementioned trend, show few long-term periodicities. Similar drought analyses have been performed by considering the Palmer Drought Severity Index (PDSI). A preliminary comparison between the SPI and PDSI obtained by using the previously discussed data set is presented. It is shown that the indices compare favourably in assessing drought variability. Finally, when the SPI analysis is extended to the Northern Hemisphere some interesting spatially remote teleconnnections linking the Tropical Pacific with the European area are shown.
