A comparison of algorithms for stream flow recession and baseflow separation

TLDR

Simple hydraulic models for outflow of groundwater into a stream suggest that the form of the storage–discharge relationship for groundwater changes from linear, for a confined aquifer, to quadratic, for an unconfined flow. The study compares algorithms for stream flow recession and baseflow separation during surface runoff events. The authors extended the linear recession model to leaky catchments, tested it on two datasets, and compared one‑, two‑, and three‑parameter baseflow algorithms on data from 11 streams. Tests show that most catchments exhibit strongly nonlinear recession relationships, yet linear models approximate short recessions; baseflow algorithms differ markedly, with the two‑parameter algorithm providing the most consistent results, and parameter variability reflects spatial recharge differences. © 1999 John Wiley & Sons, Ltd.

Abstract

Simple hydraulic models for outflow of groundwater into a stream suggest that the form of the storage–discharge relationship for groundwater changes from linear, for a confined aquifer, to quadratic, for an unconfined flow. Tests of the form of stream flow recessions in 11 streams, during periods of no recharge, show that for most catchments the storage–discharge relationship is more strongly non-linear than the quadratic form. However, for the commonly occurring case of recessions of duration up to about 10 days, the linear model remains a very good approximation, using a biased value of the groundwater turnover time. In contrast, estimates, from the stream hydrograph, of recharge during a storm event are very sensitive to the form of the storage–discharge relationship. The results of this study also show great variability in the parameters of the recession algorithm from one recession to another, attributable to spatial variability in groundwater recharge. An extension of the linear model to 'leaky' catchments, where the recession reaches zero flow, has been tested on two data sets. The second part of the paper deals with algorithms for baseflow during surface runoff events—the problem of hydrograph separation. Algorithms with one, two and three parameters have been compared, using data for the same 11 streams, and the results show significant differences in the baseflow index (BFI) predicted for some catchments. The two-parameter algorithm, which is fitted subjectively, is more consistent in providing plausible results than either the one- or three-parameter algorithms, both of which can be fitted objectively. Copyright © 1999 John Wiley & Sons, Ltd.