A Practical Assessment of the Errors Associated with Full-Depth LADCP Profiles Obtained Using Teledyne RDI Workhorse Acoustic Doppler Current Profilers

Abstract

Lowered acoustic Doppler current profilers (LADCPs) are commonly used to measure full-depth velocity profiles in the ocean. Because LADCPs are lowered on hydrographic wires, elaborate data processing is required to remove the effects of instrument motion from the velocity measurements and to transform the resulting relative velocity profiles into a nonmoving reference frame. Two fundamentally different methods are used for this purpose: in the velocity inversion method, a set of linear equations is solved to separate the ocean and instrument velocities while simultaneously applying a combination of velocity-referencing constraints from navigational data, shipboard ADCP measurements, and bottom tracking. In the shear method, a gridded profile of velocity shear, which is not affected by instrument motion, is vertically integrated and referenced using a single constraint. The main goals of the present study consist in estimating the accuracy of LADCP-derived velocity profiles and determining which processing method performs better. To this purpose, 21 LADCP profiles collected during four surveys are compared to velocities measured simultaneously by nearby moored instruments at depths between 2000 and 3000 m. The LADCP data were processed with two slightly different publicly available implementations of the velocity inversion method, as well as with an implementation of the shear method that was extended to support multiple simultaneous velocity-referencing constraints. Regardless of the processing method, the overall rms LADCP velocity errors are <3 cm s?1 as long as multiple velocity-referencing constraints are imposed simultaneously. On the other hand, solutions referenced with a single constraint are associated with significantly greater errors. The two primary instrument characteristics that influence data quality are range and sampling rate. Dependence of the LADCP velocity errors on those two parameters was determined by reprocessing range-limited subsets and temporal subsamples of the LADCP data. Results indicate an approximately linear increase of the velocity errors with decreasing sampling rate. The relationship between velocity errors and instrument range is much less linear and characterized by a steep increase in velocity errors below a limiting range of ≈60 m. To improve the quality of the velocity data by increasing the instrument range, modern LADCP systems often include both upward- and downward-looking ADCPs. The data analyzed here indicate that the addition of a second ADCP can be as effective as doubling the range of a single-head LADCP system. However, in one of the datasets the errors associated with the profiles calculated from combined up- and down-looker data are significantly larger than the corresponding errors associated with the profiles calculated from the down-looker alone. The analyses carried out here indicate that the velocity errors associated with LADCP profiles can be significantly smaller than expected from previously published results and from the uncertainty estimates calculated by the velocity inversion method.