This is a proposal to conduct a simulation-based analysis of errors in high frequency (HF) radar derived ocean surface current measurements and develop means of assessing these errors in observational systems in use today. The systematic identification and description of these errors is needed for assimilation of HF radar-derived current measurements into near-shore ocean circulation models. It is also needed for incorporation into the many new operational products that are being created based on continuous maps of ocean surface currents from HF radar networks. Conclusions will be applicable to HF radar-derived current measurements in general, but analyses will focus on systems that use compact, collocated antenna geometry and direction finding techniques, such as the CODAR/SeaSonde systems. Other commercially available systems, such as the Welan Radar or WERA, can also be operated in a direction-finding configuration. The interest in HF radar systems for near-shore mapping of ocean surface currents is presently undergoing a period of rapid acceleration. Systems are already in place at numerous locations around the coastal United States as well as many other locations around the world. These systems provide near real-time measurements of the ocean circulation within the top one or two meters of the ocean surface, on a 1-3 km resolution grid, with a range over the ocean surface of about 70 km with up to 200 km for the long range units but reliable estimates of the point-by-point uncertainties in the data remain unavailable even though such error values, together with their statistical descriptions, are required for proper assimilation into numerical circulation models. While estimates of overall uncertainty levels have been provided through comparisons with in situ measurements, those comparisons on their own are not sufficient for a number of reasons. Firstly, the in situ and remote-sensing measurements are inherently different. Current meters and, to a lesser extent, drifters measure velocity at a fixed point while the radar is sensitive to the current averaged over several square kilometers. Secondly, the radar measures the current very close to the surface where measurement with conventional current meters is problematic. Thirdly, the continuous, two-dimensional coverage of the radar cannot be duplicated and, therefore systematic errors, such as pointing biases in the direction finding algorithms, cannot be assessed based solely on comparisons with in situ observations.
The broad impact of effective assimilation of HF current measurements in coastal areas would be felt by the scientific, coastal engineering, public safety and recreational communities through more detailed and accurate nowcasts and forecasts for ocean circulation and wave conditions with greatly improved accuracy and resolution. For example, prediction of surface flows with high spatial and temporal resolution would enhance air-sea rescue and oil and toxic spill response by allowing accurate drift forecasts. Similarly, more accurate predictions of coastal circulation could lead to forecasting shifts in marine populations from phytoplankton to whales. This project seeks to address the problem of error analysis and assessment through the use of computer simulations backed up, where appropriate, by comparisons with existing in situ data sets. The effects of a wide variety of error sources, including ship echoes, ionospheric echoes, random noise, random variations in the ocean currents within a measurement cell, imperfect knowledge of system properties, such as antenna pattern distortion, and limitations of the direction finding algorithms as a function of antenna design will be examined. A systematic look at error propagation for the commonly used HF radar configurations is overdue and that, once documented, these error models will accelerate the exploitation of these remarkable oceanographic instruments.
