A Statistical Evaluation of Models for Extrapolating Current Velocities from Boat-Mounted ADCP Profiles

Abstract

This paper presents a statistical evaluation of two theoretical fitting methods — logarithmic and power law — commonly used to extrapolate velocities to unmeasured areas of a vertical profile, such as near-bottom velocities. Analysis of covariance (ANCOVA) was used to test the models' performance when applied to velocity data collected with a boat-mounted acoustic Doppler current profiler (ADCP) at the six tidal inlets of the Ria Formosa (southern Portugal) under spring-and neap-tide conditions. Results show that both models' goodness-of-fit is negatively correlated with velocity. Correlation also diminishes under ebb conditions, with no relation to velocity, but rather to transect depth. Both models consistently produce better simulations in inlets with a larger cross-sectional area, where turbulence scales are more attenuated due to spatial averaging. The logarithmic law model proved to be more robust across different velocities and channel morphologies and should be adopted as the preferred method for extrapolating velocity profiles when using boat-mounted ADCPs at confined channels and inlet entrances.