Parametric vs. Non-Parametric Approach for the Estimation of the SPI Drought Index

Accurate drought identification is important for both scientists and decision-makers to be able to make informative decisions. In this study, parametric and non-parametric approaches for analyzing meteorological drought are compared, aiming at simplifying the calculation of the Standardized Precipitation Index (SPI). The comparison is performed across various meteorological stations covering the entire territory of Greece, using monthly rainfall data spanning from 1961 to 2021. Meteorological drought is assessed through the SPI for the 12-month reference period. A two-parameter gamma distribution, with parameters estimated using the maximum likelihood estimation method, is employed for the estimation of the SPI drought index as the parametric classic approach. For the non-parametric approach, the SPI drought index is estimated using six empirical probability plotting positions: Beard, Blom, Cunnane, Gringorten, Hazen, and Weibull. Results indicate that the empirical approach can effectively identify drought events in comparison to the classic approach. However, caution is advised, particularly when different drought classes are identified, as the non-parametric approaches may underestimate drought severity. In addition, for the Greek meteorological conditions, the results revealed that in the case of extreme drought events, the estimation of SPI employing the classic approach is to be preferred.