Factors Influencing Markov Chains Predictability Characteristics, Utilizing SPI, RDI, EDI and SPEI Drought Indices in Different Climatic Zones

Comparability analyses were carried out to investigate behavioural aspects of effective drought index (EDI), standardized precipitation index (SPI), reconnaissance drought index (RDI) and standardized precipitation evapotranspiration index (SPEI), considering 3-month, 6-month and annual time periods. Investigations included parametric/non-parametric correlation analysis among indices, climatic zone influence, record length impacts and evapotranspiration role (RDI and SPEI) on Markov chains predictability characteristics. Except for the EDI, all indices/cases (all climatic zones) showed significant correlation. In arid/semi-arid climates, the 3-month and 6-month maximum drought severities were detected by the RDI and annual maximum drought severities were detected by the SPEI, emphasizing the evapotranspiration influence. In all climatic zones, the EDI values for wet (dry) periods were higher (lower), compared to other indices. First order dependency was detected for the EDI (all cases) and the SPI (most cases), over entire period (1951–2011) and sub-periods [(1951–1981), (1982–2011)]. The largest number of second order dependency was detected by the SPEI, followed by a relatively large number of such cases by the RDI (3-month time period), for the 61-year data period. This research showed that several factors influence Markov chains predictability characteristics in drought studies, particularly the impact of record length and evapotranspiration (RDI and SPEI) were confirmed. Copyright Springer Science+Business Media Dordrecht 2013