8.11
European drought climatology and prediction using the standardised precipitation index
Benjamin Lloyd-Hughes, University College London, Dorking, United Kingdom; and M. A. Saunders
Monthly Standardised Precipitation Indices (SPI) have been calculated for the period 1901-1999 over a 0.5o latitude by 0.5o longitude grid covering Europe from 35oN-70oN and 35oE-10oW at timescales of 3,6,9,12,18, and 24 months. Computation of the SPI requires assumptions about the particular statistical distribution used to model the raw precipitation data at each site, month and timescale of interest. Three models were tested for goodness of fit using the Lilliefors test, namely, the normal distribution, the log-normal, and the gamma. The gamma distribution is found to provide the best fit for most of Europe at all timescales. None of the distributions tested could adequately model rainfall across eastern Turkey or north west Spain.
Area averages of SPI values for the European region are compared with the Palmer Drought Severity Index (PDSI). The PDSI correlates with the SPI on timescales of 6 to 12 months, with a peak of r=0.73 at 9 months. Cross-correlations between the indices reveal a large degree of redundancy. SPIs standardised over 3 and 12 months are shown to be adequate for describing moisture availability across Europe. Significant positive trend is seen in SPI12 throughout the period 1901-1999. A spatial analysis reveals this to be concentrated over Scandinavia in the north of the study region.
The level and origin of seasonal forecast skill in SPI values for northern Europe is assessed through canonical correlation analysis (CCA) with four predictor fields, namely, (a) European precipitation (i.e. persistence), (b) quasi-global sea surface temperature (SST), (c) local SST, and (d) northern hemispheric mean sea level pressure (MSLP). For each month predictions are made of SPI3 and SPI12 at leads from 0 to 15 months. Skill is measured by the mean improvement over climatology provided by the forecasts for the period 1970-1999. Maximum skill is found during the spring at a lead of 3 months and is located in a 10o band centred on 50oN. The majority of this skill arises from an ENSO like pattern in the global SST field, with a smaller contribution from local SST. Warm wintertime conditions in the NINO3 region are shown to precede a springtime regime which favours cyclonic activity over the North Sea with high pressure over north Africa. Warm moist air is advected northward into central Europe where it meets cold polar air giving rise to increased precipitation.
Session 8, Drought Assessment and Prediction
Wednesday, 15 May 2002, 8:00 AM-12:00 PM
Previous paper Next paper