Wet and dry period identification method through serial correlation decomposition

Abstract

ABSTRACT The serial correlation coefficient is the linear dependence procedure used most frequently to identify the short-memory dependence in a given hydrometeorology time series. The serial correlation formulation can be decomposed into four hydrometeorologically important parts: the wet (surplus) spell and dry (deficit) spell (drought) durations in addition to up-crossing and down-crossing components. The only negative values are available for up-crossing and down-crossing cases in the two successive standard variate multiplications, which implies that the first-order serial correlation coefficient has a lower dependence value with more contributions from the up- and down-crossing components. The difference between this methodology and those available in the literature is in the explicit use of the correlation coefficient for the identification of the abovementioned drought features. The validity of this new methodology is revealed through extensive Monte Carlo simulation and actual application to lengthy (1840–2003) Danube River monthly and annual discharges.