Estimating freshwater flows from tidally affected hydrographic data

Abstract

Detiding end-of-catchment flow data are an important step in determining the total volumes of freshwater (and associated pollutant loads) entering the ocean. We examine three approaches for separating freshwater and tidal flows from tidally affected data: (i) a simple low-pass Butterworth filter (BWF); (ii) a robust, harmonic analysis with Kalman smoothing (RoHAKS) which is a novel approach introduced in this paper; and (iii) dynamic harmonic regression (DHR). Using hydrographic data collected in the Logan River, Australia, over a period of 452 days, we judge the accuracy of the three methods based on three criteria: consistency of freshwater flows with upstream gauges; consistency of total discharge volumes with the raw data over the event; and minimal upstream flow. A simulation experiment shows that RoHAKS outperforms both BWF and DHR on a number of criteria. In addition, RoHAKS enjoys a computational advantage over DHR in speed and use of freely available software. Key Points: RoHAKS gives improved estimates of freshwater flow Detiding methods can exhibit periods of upstream flow Detiding based on robust statistical methods shows improved performance