Temporal error analysis for compact cross-loop direction-finding HF radar

Surface currents and ocean swell are measured using an HF radar with a compact cross-loop antenna design. This unique direction-finding radar has been widely used for long-term monitoring and for coastal oceanography applications over the last ten years. Numerous studies have evaluated errors associated with these devices under a host of environmental and seasonal conditions, and reliable comparisons between HF radar surface current measurements and those from other technologies indicate a range of 0.08 to 0.15 ms^-1, while internal error estimates that are made using the spread of data values in a time series or spectrum from the radar systems generally lead to errors of around 0.2 ms^-1. These error estimates are typically based on manufacturer-recommended operating parameters that can include up to 25 specific parameters or assumptions made about the data. In particular, one of the dominant parameters in computing ocean surface currents from SeaSonde devices is the averaging time of both the power spectra and the computed radial components of the surface current vectors. This paper explores the relative effect of temporal averaging for two SeaSonde stations inWestern Australia. By analysing 31 days of radial data, we evaluate the statistical characteristics of noise in the spectra and determine the limits to the assumption that it represents a stationary ensemble of independent samples. The improvement in accuracy as we integrate over different time intervals from 10 to 180 minutes is determined by these characteristics. These results are applied to error analysis for the radial components of surface currents and compared with the errors in the gridded values of the radial components. Finally, the effects of calibrated and uncalibrated antenna patterns on the radial components is evaluated in the context of the temporal and spatial errors.