The historically adjusted range and the historically rescaled adjusted range

Abstract

It was remarked by Hurst in 1951 that the adjusted range gives the size of the smallest reservoir capable of providing a constant discharge equal to the mean inflow. Since that time this range and its rescaled modification, the Hurst range, have been widely discussed, not however primarily with a view to applying them to reservoir design problems, but rather on account of their possible relevance to the simulation of geophysical time series.

Acknowledging the well-known conceptual weaknesses of adjusted ranges and the theoretical difficulties that inhibit their direct utilisation in the design and operation of real reservoirs, the authors argue that the interest displayed on ranges during the past few decades justifies the effort of eliminating one in particular of these weakness, namely their non-implementability as operating policies, a consequence of the fact that they can only be retrospectively evaluated. The paper proposes modifications in which the unknowable mean and standard deviation of future samples are replaced by the known mean and sample standard deviation of historical data, leading to the historically adjusted range and the historically rescaled and adjusted range. The latter is produced as an implementable approximation to Hurst's (1951) solution to the optimal reservoir problem.

The expected values of the new ranges are evaluated and numerically tabulated.