The density difference and weighted RVA approaches for assessing hydrologic regime alteration

Abstract

Assessing the natural or man-made alterations in the hydrological characteristics of watershed systems is a prerequisite to develop watershed management, restoration, and conservation strategies and plans. The Range of Variability Approach (RVA), based on a suite of Indicators of Hydrologic Alteration (IHA), is most widely used to assess hydrologic alterations. However, the traditional RVA and its revised versions and alternatives all suffer from major challenges such as binning the data and using only the target range to compare the probability distributions of the hydrologic variables in the pre- and post-impact periods through applying simplified statistical analyses. This paper proposes the Density Difference Approach (DDA) and the weighted RVA to overcome these challenges. Despite the previous approaches, which divide the time series of the continuous variables into discrete groups, DDA uses the continuous probability density function (pdf) estimated by the Gaussian Kernel Density Estimation (KDE) technique. It considers the magnitude, frequency, and order position of all individual data in the alteration analysis. Furthermore, to account for the variations in the values and frequencies of the variables beyond the target range, which is neglected in the traditional RVA, the weighted RVA approach has been proposed. These suggested approaches, along the other widely used alteration assessment approaches, have been applied to quantify the hydrologic alteration in the Gharnaveh watershed in the northeast of Iran using 50 years of daily discharge data recorded at the Tamar hydrometry station at the outlet of the watershed. The performance of the alteration analysis approaches was further evaluated using the randomly generated time series with the arbitrary changes in the probability distribution parameters. The results indicate a significant difference in the alteration values and ranks calculated by different alteration analysis approaches. Compared to the traditional RVA, the weighted RVA resulted in higher alteration degrees for most of the IHA indicators. DDA outperformed the other alteration analysis approaches for both the observed and randomly generated data. It better accounted for the alteration in the key characteristics of the probability distribution (including the location, dispersion, and skewness parameters) of the randomly generated data series. We recommend DDA as an appropriate hydrologic alteration assessment approach that assists water resource managers and decision-makers through accounting for multi-dimensional attributes of hydrologic alterations.