This paper presents an evaluation of several automated techniques concerned with base flow' separation and recession analyses. Two base flow techniques were considered, one based on a digital filter and the other on simple smoothing and separation rules. A comparison between two commonly used techniques of recession analyses, the correlation method and the matching strip method, was also undertaken. The relative performances of the techniques were evaluated using the results obtained from the daily streamflow records of 186 catchments in southeastern Australia. The work described in this paper was undertaken within the general framework of defining the low-flow characteristics of small rural catchments, the overall objective being the development of a regional model for use on ungauged catchments. k, which is called the recession constant. There appears to be some contradiction in the literature as to who first derived this equation, though it appears that J. Boussinesq, E. Maillet, and R. E. Horton all independently derived the same function around the year 1904 [see Horton, 1933; Werner and Sundquist, 1951; Hall, 1968; Appleby, 1970]. Subsequently, Werner and Sundquist [1951] showed that (1) is the linear solution of the one-dimensional general differential equation governing transient flow in artesian aquifers (the diffusion equation).Barnes [1939] suggested that the three individual components of runoff, overland flow, interflow and groundwater flow, may be distinguished by plotting the logarithms of the flows against time. Recessions that obey (1) plot as a straight line on semilogarithmic graph paper, the gradient of which is equal to the recession constant, and thus the different components may be distinguished by the different straightline segments. Bank storage has also been considered as a fourth store that contributes to streamflow; for large values of time and an infinite aquifer, Cooper and Rorabaugh [ 1963] showed that the recession due to bank storage approaches the simple exponential form of (1). The range of daily recession constants has been found typically to be [e.g.,
Klaassen and Pilgrim, 1975]: 0.2-0.8 for surface runoff, 0.7-0.94 for interflow, and 0.93--0.995 for base flow. The overlapping ranges reflect the difficulties inherent in identifying a particular recession as being either surface runoff, interflow, or base flow. Alternatively, more complex functions than (1) have been used to describe recession flow. However, after detailed investigation of recession behavior, many authors [e.g., Ineson and Downing, 1964; Nutbrown and Downing, 1976; Anderson and Butt, 1980; Petras, 1986] have concluded that no single linear plot can be constructed for base flow recession. The nonlinearity is a function of factors such as carry-over storage from a prior period of recharge, variations in areal pattern of recharge, channel, bank and flood 1465