The industry's recognized authority for design, engineering and application of equipment and services in the global ocean community
Issue link: http://sea-technology.epubxp.com/i/477967
36 st / March 2015 www.sea-technology.com dictions. However, HF radar data presented a number of potential issues that required investigation to ensure conf- dence in the tidal current predictions. The Infuence of Missing Data The current measurements collected by CO-OPS from self-contained acoustic Doppler current proflers (ADCPs) typically have a high degree of accuracy with very little bad or missing data (less than 10 percent). HF radar data can have much greater amounts of data missing or removed, especially at points near the perimeter of the grid where 80 percent or more of data can be missing over the course of one year. Although the least-squares harmonic analysis method can handle data gaps, studies were conducted to determine what percentage of data is needed to provide a reliable harmonic analysis result. Utilizing one year of HF radar data from Chesapeake Bay (at a grid point with only approximately 5 percent data missing) an algorithm was developed to randomly remove data points—iteratively from 5 percent to 99 percent of the data removed—and then rerun the harmonic analysis to domain (2-kilometer resolu- tion for Chesapeake Bay and 1-kilometer resolution for San Francisco Bay). The tidal cur- rent predictions are calculated based on a harmonic analysis of one year of observations at each grid point in the domain. The predictions are calculated each hour and are provided to users for the previous and fol- lowing 48 hours. The surface current obser- vations must traverse multiple stages of review and processing each hour prior to reaching po- tential users. The process begins at the instrument level, as HF radar data are collected, pro- cessed and quality controlled (QC) locally prior to being transmitted to the HF Radar National Network hosted by the Scripps Institution of Oceanography. At the National Net- work, data undergo additional QC and processing prior to being made available for CO-OPS to access and ingest into the back-end application of the Web product. At this point, additional QC is performed at CO-OPS, where data points that show poor data return values (data points with less than 40 percent data return over the past year) are removed. The surface current speed and direction for the most recent 48 hours is then presented via the map Web interface. Harmonic Analysis of HF Radar Observations The current predictions shown at each grid point are of the speed and direction of the tidal component of the sur- face current at that location. These predictions are generated from 37 harmonic constituents derived from a least-squares harmonic analysis of one year of HF radar surface current data. CO-OPS has been performing harmonic analyses on current data for decades, and publishes the nation's current predictions annually in the NOAA Tidal Current Tables and through a new online interface called NOAA Current Pre- (Top) The time series plot of the observed and predicted surface current speed and direction, which opens when a user clicks on a map vector. (Bottom) The percent residual (i.e., nontidal) variance for the current speed in the major axis direction at San Francisco Bay. This is one example of the diagnostics used to assess performance of the har- monic analysis results used to generate the tidal current predic- tions .