SciPi; A. Kirincich
    PROJECT; Exploring Submesoscale Features in the Kuroshio East of Taiwan using HF Radar
    EXPERIMENT; 2017 HF Radar observations of the East Taiwan Coast.
    Abstract;  High-frequency radar-based observations of surface currents along the east coast of Taiwan, obtained over a 50-day period in early 2017, are used to examine the occurrence, generation, and downstream advection of submesoscale eddies in the Kuroshio. Measured at an effective depth of 2 m and radial resolution of 3 km from four land-based HF radar systems spanning an 250-km along-stream distance, the surface current observations reveal the instantaneous position of the Kuroshio on hourly time scales as well as the occurrence of numerous high relative vorticity features. Vortex features with spatial scales of 5-20 km were concentrated in the first 30 km offshore, with many created at the southern tip of Taiwan on tidal timescales.  Most features, with relative vorticities approaching zeta/f=1, translated northward along the coast at the speed of the Kuroshio itself and were coherent over the 250-km length of the Taiwanese coastline. Both tides and regional winds appear to influence when long-lived features form, and the offshore advection of surface waters by the vortices are observable in intermittent Satellite images of surface chlorophyll.  While most features are advected northward with the current, a submarine ridge acts to impede the flow, scattering northward moving features and causing occasional southward-migrating vortices.
    	DESCRIPTION; The surface current observations used here were obtained from four long-range (4 MHz transmit frequency) land-based coastal radar systems, operated by the Taiwan Ocean Research Institute (TORI) and the National Taiwan University (NTU).  All systems were Codar Ocean Sensors SeaSondes, with the three southern stations operated by TORI, and the northern-most station by NTU.  Collected over the time period spanning February 1st to March 26th, 2017, the hourly observations of Doppler cross-spectra had a radial resolution of 3 km. Horizontal resolution was dependent on both the resolution of the measured antenna patterns (1 degree in azimuth) as well as the inherent azimuthal resolution of the radar returns themselves.  
		SENSOR_INFORMATION; Radio frequency interference from the ionosphere is a particular problem for the TORI and NTU radars, due to a combination of latitude and transmit frequency, causing elevated background noise during local nighttime.  Returns at ranges of 90 km, the distance to the primary scattering layer within the ionosphere, are especially affected. SNR was used as an effective screening tool to isolate and eliminate data contaminated by ionospheric radio noise common in the region, adding further improvements to the radial velocity results.  However, data from a 50x50 km region directly offshore of the radar site near 23deg 30' N 121deg 30' E (Fig. \ref{fig:timestack}) was excised during the hours of 11 to 17 UTC each day during the observational period due to poor data returns during times of high ionospheric reflections and radio noise that resulted in poorly resolved and inaccurate vector current estimates.  Using synthetic radials from adjacent HFR sites \cite{Emery2019}, surface current uncertainties are estimated to be 5-10 cm/s.
	DATA_PREPARATION_DESCRIPTION; Observed Doppler cross-spectra were reprocessed following Kirincich et al. (2012) using adjusted measured antenna patterns and advanced quality control metrics to estimate the radial surface currents observed at each site.  Measured antenna response patterns were adjusted iteratively to reduce radar-to-radar inconsistencies defined using synthetic radials estimated from adjacent radars as well as systematic biases in mean vorticity and divergence patterns.  Vector combinations of the radial surface currents, representative of the average currents over the top 2 m of the water column (StewartJoy, 1974)  were estimated using power-weighed least-squares methods (Kirincich et al. 2012, Kaplan et al 2005) with a fixed horizontal averaging length-scale of 3 km, and masked for errors due to the geometric dilution of precision (GDOP) greater than 2  (Barrack, 2002).  	
	DATA_ORIGIN;  Taiwan Ocean Radar Observing System (TOROS) ocean current radar team and the National Taiwan University (NTU)
    	FUNDING_SOURCE;  HF Radar observations used here were funded by Taiwan's National Applied Research Laboratories as well as the National Taiwan University. A. Kirincich was funded by the U.S. Office of Naval Research under contract #N000141712761.
    COORD_SYSTEM; GEOGRAPHIC