Redundancy and the choice of hydrologic indices for characterizing streamflow regimes

TLDR

Hydrologic indices are widely used to describe streamflow regimes, but the growing number of available indices forces researchers to select a subset that balances computational efficiency and representation of flow characteristics. This study reviews 171 hydrologic indices from 420 long‑term flow records and offers recommendations for selecting a reduced, non‑redundant set. By analyzing redundancy patterns, the authors develop statistically and ecologically based guidelines and assess index transferability across diverse stream types. The resulting framework enables researchers to identify non‑redundant indices that capture key flow attributes and to choose them parsimoniously for hydroecological studies. © 2003 John Wiley & Sons, Ltd.

Abstract

Abstract The utility of hydrologic indices for describing various aspects of streamflow regimes has resulted in their increased application in riverine research. Consequently, researchers are now confronted with the task of having to choose among a large number of competing hydrologic indices to reduce computational effort and variable redundancy prior to statistical analyses, while still adequately representing the major facets of the flow regime. The present study addresses this concern by providing a comprehensive review of 171 currently available hydrologic indices (including the commonly used Indicators of Hydrologic Alteration) using long‐term flow records from 420 sites from across the continental USA. We highlight patterns of redundancy among these hydrologic indices and provide a number of statistically and ecologically based recommendations for the selection of a reduced set of indices that can simultaneously (1) explain a dominant proportion of statistical variation in the complete set of hydrologic indices and (2) minimize multicollinearity while still adequately representing recognized, critical attributes of the flow regime. In addition, we examine the transferability of hydrologic indices across ‘stream types’ by identifying indices that consistently explain dominant patterns of variance across streams in varying climatic and geologic environments. Together, our results provide a framework from which researchers can identify hydrologic indices that adequately characterize flow regimes in a non‐redundant manner. In combination with ecological knowledge, this framework can guide researchers in the parsimonious selection of hydrologic indices for future hydroecological studies. Copyright © 2003 John Wiley & Sons, Ltd.