Saline intrusion in partially mixed estuaries

Abstract Restricting interest to partially mixed estuaries, earlier studies of tidally averaged linearised theories relating to the vertical structure of salinity and velocities (accompanying saline intrusion) are extended to take account of tidal straining and associated convective overturning. The applicability of these theories is evaluated by reference to a ‘single-point’ numerical model in which the time-varying cycle of depth-averaged tidal current amplitude, U , and a (temporally and vertically) constant saline gradient, S x , are specified. This model highlights the importance of convective overturning in counteracting unstable density structures introduced by tidal straining. By omitting overturning in the model, results agree closely with linearised theoretical derivations. However, incorporating overturning substantially increases tidally averaged surface-to-bed differences for both residual currents, δ u , and salinity, δ s . The vertical structure of tidal currents is a maximum, and hence the effect of tidal straining, in shallow macro-tidal estuaries. The propagation of tidal elevations and currents remains insensitive to saline intrusion in partially mixed estuaries. The applicability of the model was evaluated by simulation of recent measurements by Rippeth et al. (J. Phys. Oceanogr. 31 (2001) 2458). To explore the generality of estuarine responses, the model was run for a wide range of values of saline intrusion lengths, L , and water depths, D . Additional sensitivity analyses were made for changes in U and bed stress coefficient, k . Response frameworks are shown for: δ u , δ s , potential energy anomaly φ , work done by bed friction and internal shear, rates and efficiency of saline mixing and ratios of relative mixing by diffusion to overturning. By equating the rate of mixing associated with vertical diffusion with river flow, Q , an expression for saline intrusion length L∝D 2 /k U U o ( U o river flow velocity) was derived. This formulation agrees with an earlier derivation based on flume tests and showed reasonable agreement with observed values in six estuaries (eight cases). However, in funnel-shaped estuaries, axial migration of the intrusion introduces major variations in D , U and U o , thereby complicating the application of the above expression for L . Moreover, the time lag involved in the adjustment of L to changes in U and Q may explain much of the complexity encountered in the observations.