“…Among the different proposed techniques, exhaustively reviewed by Sharma and Mehrotra (2010), the most commonly adopted approach to the problem since the 1960s is the Markov-chain (MC) simulation: in its classical form, it is a linear model which cannot simulate the variability and persistence at different scales. Solutions to deal with this limitation consist of introducing exogenous climatic variables and large-scale circulation indexes (Hay et al, 1991;Bardossy and Plate, 1992;Katz and Parlange, 1993;Woolhiser et al, 1993;Hughes and Guttorp, 1994;Wallis and Griffiths, 1997;Wilby, 1998;Kiely et al, 1998;Hughes et al, 1999), lower-frequency daily rainfall covariates (Wilks, 1989;Briggs and Wilks, 1996;Jones and Thornton, 1997;Katz and Zheng, 1999) or an index based on the short-term daily historical or previously generated record (Harrold et al, 2003a, b;Mehrotra and Sharma, 2007a;Mehrotra and Sharma, 2007b) as conditioning variables for the estimation of the MC parameters. By doing this, nonlinearity is introduced in the prior model, and the MC parameters change with time as a function of some specific lowfrequency fluctuations.…”