How to Fly an Autonomous Underwater Glider to Measure an Internal Wave

Abstract

Internal waves are ubiquitous features of lakes and oceans, significantly contributing to vertical mixing across large areas, driving gas, nutrient, sediment, and heat exchange between deep and surface waters. Traditionally moorings have been used to sample internal wave fields; providing good temporal data but limited spatial information. More recently, buoyancy-driven autonomous underwater vehicles (aka gliders) have enabled additional characterization of internal waves. Gliders are capable of travelling large distances over periods ranging from days to months, some yo-yoing to depths in excess of 1000 m, making them a promising tool for internal wave observation that better resolve spatial variability of these phenomena. They have been used to characterize mixing from internal waves in multiple ocean locations (e.g. the South China Sea, the Pacific Ocean off the California Coast, the Tasman Sea, and the Faroe Bank Channel) and several deep lakes (e.g. Lake Tahoe, Lake Geneva and Lake Superior). To date no optimal method to implement gliders to measure internal wave characteristics has been determined, although different research groups have experimented with several approaches. Previously employed approaches include using a glider as a virtual mooring, along-shore transects, across-shore transects, and flying a zig-zag pattern across the path of wavefronts. In the work presented here, data were collected in Lake Tahoe (USA), where local bathymetry (down to a maximum depth of 501 m) is known to produce trapped internal waves across a wide range of depths. At various locations, internal waves were concurrently measured using a G2 Slocum glider, moored thermistors and a moored Acoustic Doppler Current Profiler (ADCP). Using these complementary datasets, we can compare the effectiveness of existing and new sampling approaches by flying multiple missions where the vehicle's path varies in orientation and length relative to the wavelength and return period of known internal waves. Internal waves can be identified using spectral analysis of temperature time-series measurements, producing a power spectrum similar to the Garrett-Munk spectrum. Peaks in the power spectrum are expected at known (from previous observational and modelling work) and calculated frequencies corresponding to specific internal wave modes. The success of the sampling method can be assessed by the extent to which the expected spectral peaks are evident in the power spectrum. Additionally, through the application of a dynamic flight model, the vertical water velocities experienced by the glider can be estimated and compared to those expected from internal waves and measured by a nearby moored ADCP. Other criteria to be considered are the width of the confidence interval, the possibility of double counting waves in closed basins, and minimization of spatio-temporal smearing. Identifying the best way to employ a glider to measure an internal wave will lead to improved data collection from future glider deployments.