Long-term variation of PDSI and SPI computed with reanalysis products

Martins DS, Paulo A, Pires C, Pereira LS
European Water 60: 271-278, 2017,

Download PDF


Long-term temporal variation of droughts in Portugal was assessed using the Palmer Drought Severity Index (PDSI) and the Standardized Drought Index (SPI) with a 9 month time scale. Weather data from the NOAA-CIRES Twentieth Century Reanalysis Project version 2c, which spans from 1851 to 2014 with a spatial coverage of 2.0 latitude x 2.0 longitude were used. For PDSI, monthly evapotranspiration was computed using the FAO PM-ETo method. Monthly data from reanalysis products consisted of maximum and minimum air temperature, net radiation, wind speed and relative humidity. Precipitation data was obtained from the same source. To assess how the long-term climate variability influences the identification of dry events in the drought indices, five different calibration periods were selected to estimate, respectively the potential values of evapotranspiration, runoff, soil moisture recharge and percolation loss of the PDSI water balance and the distribution function parameters of the SPI. Results show that the adopted calibration period have a significant impact of the frequency of extreme events detected in both drought indices. A Fourier analysis was applied to PDSI and SPI time series to search for significant cycles that could relate to return periods of droughts, with North Atlantic Oscillation cycles previously identified for Portugal.