Abstract
This project evaluated how high-resolution, spatially distributed field data can be used to refine and validate site-scale hydrodynamic simulations of tidal channels. Use of such spatially-distributed field observations or site-scale hydrodynamic simulations will be needed for producing accurate predictions of tidal energy production over larger arrays of tidal turbines.
Here, we compared field observations from a series of short surveys in Agate Pass taken on August 20th, 2020 against a high-resolution hydrodynamic model of the site. The field observations included water velocities measured by station-keeping ADCP measurements, drifting downward-looking ADCPs, and microFloats, gathered by researchers at APL-UW under a previous project.
Supported by this TEAMER funding, PNNL refined an existing hydrodynamic model of Puget Sound, WA, to produce a high-resolution (10-20 m horizontal, 1 m vertical) subdomain model of Agate Pass, and conducted model simulations over the field survey periods at 30 second intervals. Using these high-resolution maps of tidal currents, APL-UW and PNNL compared the model against the field observations.
Results indicated agreement between the model and the field observations, with a few notable trends:
- PNNL performed subdomain model refinement to improve model accuracy. While there was overall improvement, the improvement in error statistics is not significant when comparing to NOAA fixed-station ADCP data.
- Comparison of model and microFloat data showed moderate overall agreement, though larger than comparisons between the model and NOAA fixed-station ADCP data.
- There is better agreement between model and field observations during ebb surveys than during flood. This is likely due to the location of the field observations relative to spatially constrained vs. unconstrained flow. During ebb surveys, observations were primarily restricted to the channel. During flood, the observations extended from within the channel into the body of Port Orchard, primarily carried by a jet. While the model captures the presence of the jet, the exact location is unlikely to match reality, resulting in differences between model and observations.
- Comparisons between simulated microFloat trajectories and the observed ones also confirmed the model bias in simulating flow fields in Agate Pass. The model-data difference is consistent with previous comparisons between NOAA fixed-station ADCP measurements and the model.
The microFloat has the potential to provide cost-effective, spatially-distributed resource assessments in tidal, river, and ocean energy applications. Through the model-data analysis pursued here, we sought to determine how microFloat data products may be best utilized in model validation, improving their TRL and aiding in current-energy site-characterization. The answer was, while somewhat unsatisfying but not surprisingly, “It’s complicated.” The methodology implemented here is straightforward and can be applied to future model-observation comparisons, but the field data set was too sparse (i.e., too few and too short deployments) to draw broadly applicable conclusions. To further the investigation, a bigger field observation data set would be needed, preferably multiple sequential tidal cycles (a few days) gathered over a minimum of two different periods of the lunar cycle. The microFloat data can be used to support model validation for tidal energy resource characterization and assessment.