A simplified semiempirical approach to probabilities of extreme hydrologic droughts

On the basis of an available theory approximately valid for stationary first‐order two‐state Markovian processes and an assumption that the distribution of drought severity is approximately two‐parameter gamma, a semiempirical method is proposed for the estimation of the probabilities of extreme drought durations and severities which may occur in the annual flow sequence of a stream during a period of finite length. The method is applied to the annual flows of eight streams in various parts of the world, and the semiempirical probability distributions obtained using this method are compared with experimental distributions obtained using synthetic streamflows generated with a lag one Markov model as well as a long‐memory broken line model. The comparisons suggest that considerable insight into the extreme drought probabilities can be gained by the use of the simple semiempirical method introduced here.