Improved streamflow recession parameter estimation with attention to calculation of − dQ/dt

The rate of streamflow recession can be used to assess storage-outflow properties of source aquifers. A common method of analyzing streamflow recession is to plot the time rate of change in streamflow Q as a function of Q in a log-log space. Theory predicts, for diagnostic recession regimes, a power law relationship − dQ/dt = aQb, where recession coefficients a and b are functions of the hydraulic and geometric properties of the aquifer and of boundary and initial conditions. Observational error reduces the accuracy of estimates of a and b with errors in estimating the time derivative of the late-time recession (−dQ/dt) being particularly sensitive to observational error. Here we propose a method to improve estimation of a and b with particular focus on the estimation of −dQ/dt. Compared to previously published methods we find greater robustness in estimates of −dQ/dt and recession parameters and less sensitivity to the methodological parameters employed. Previous methods result in up to 50 to 100% error when estimating the recession parameter b, while the proposed methodology produces errors below 5% in the cases analyzed.