Palmer drought severity index as soil moisture indicator: physical interpretation, statistical behaviour and relation to global climate

Abstract

Palmer Drought Severity Index (PDSI) series, based on monthly homogenised temperature and precipitation data, are analysed for the 1901–1999 period at three stations in West-Hungary, i.e. in an objectively separating region of the country, concerning spatial variations of the monthly PDSI fluctuations. All displayed results represent computations by the Thorthwaite-type potential evapotranspiration. Some comparison with those index-series, computed by the Blaney–Criddle method is given in the Discussion. Series of PDSI exhibit strong correlation with series of two independent soil moisture estimations. Having the regression coefficient standardised by standard deviation of the soil-moisture, we obtain similar coefficients during the year (allowing for only 10–30% difference). This means, PDSI can be considered as a soil-moisture indicator. It is shown that the monthly standard deviation of PDSI exhibit small variation, ca. 10%, with a minimum in the summer period. Distribution of monthly PDSI can be considered as Gaussian, according to the Kolmogorov–Smirnoff test, whereas according to the χ 2-test this is true for more than 2/3 of the cases. Exceptions all fall in the second half of the year. Finally, multi-annual relation of PDSI to the global temperature trends are analysed using the method of “slices” ( Mika, 1988), dividing the local and global values into uniform time sequences, the so called time-slices and calculating regression coefficients between the local PDSI and two hemispherical temperature variables. One of the latter is the hemispherical mean, the other is the continent–ocean air temperature contrast. This correlation is always negative and frequently significant, which means that in the 20th century local soil moisture conditions became drier parallel to the hemispherical changes.