Assessment of CODAR SeaSonde and WERA HF Radars in Mapping Surface Currents on the West Florida Shelf

Yonggang Liu College of Marine Science, University of South Florida, St. Petersburg, Florida

Search for other papers by Yonggang Liu in
Current site
Google Scholar
PubMed
Close
,
Robert H. Weisberg College of Marine Science, University of South Florida, St. Petersburg, Florida

Search for other papers by Robert H. Weisberg in
Current site
Google Scholar
PubMed
Close
, and
Clifford R. Merz College of Marine Science, University of South Florida, St. Petersburg, Florida

Search for other papers by Clifford R. Merz in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

Concurrently operated on the West Florida shelf for the purpose of observing surface currents are three long-range (4.9 MHz) Coastal Ocean Dynamics Applications Radar (CODAR) SeaSonde and two median-range (12.7 MHz) Wellen Radar (WERA) high-frequency (HF) radar systems. These HF radars overlook an array of moored acoustic Doppler current profilers (ADCPs), three of which are presently within the radar footprint. Analyzed herein are 3 months of simultaneous observations. Both the SeaSonde and WERA systems generally agree with the ADCPs to within root-mean-square differences (rmsd) for hourly radial velocity components of 5.1–9.2 and 3.8–6.5 cm s−1 for SeaSonde and WERA, respectively, and within rmsd for 36-h low-pass filtered radial velocity components of 2.8–6.0 and 2.2–4.3 cm s−1 for SeaSonde and WERA, respectively. The bearing offset and tidal and subtidal currents of total velocities are also assessed using the ADCP data. Despite differences in a variety of aspects between the direction-finding CODAR SeaSonde (long range, effective depth of 2.4 m, integration time of 4 h, and idealized antenna patterns) and the beam-forming WERA (median range, effective depth of 0.9 m, and integration time of 1 h), both HF radar systems demonstrated good surface current mapping capability. The differences between the velocities measured with the HF radar and the ADCP are sufficiently small in this low-energy shelf that much of these rmsd values may be accounted for by the expected measurement differences due to the horizontal, vertical, and temporal sampling differences of the ocean current observing systems used.

Collaboration for Prediction of Red Tides Contribution Number 31.

Corresponding author address: Yonggang Liu, College of Marine Science, University of South Florida, 140 7th Ave. S, MSL 119, St. Petersburg, FL 33701. E-mail: yliu18@gmail.com

Abstract

Concurrently operated on the West Florida shelf for the purpose of observing surface currents are three long-range (4.9 MHz) Coastal Ocean Dynamics Applications Radar (CODAR) SeaSonde and two median-range (12.7 MHz) Wellen Radar (WERA) high-frequency (HF) radar systems. These HF radars overlook an array of moored acoustic Doppler current profilers (ADCPs), three of which are presently within the radar footprint. Analyzed herein are 3 months of simultaneous observations. Both the SeaSonde and WERA systems generally agree with the ADCPs to within root-mean-square differences (rmsd) for hourly radial velocity components of 5.1–9.2 and 3.8–6.5 cm s−1 for SeaSonde and WERA, respectively, and within rmsd for 36-h low-pass filtered radial velocity components of 2.8–6.0 and 2.2–4.3 cm s−1 for SeaSonde and WERA, respectively. The bearing offset and tidal and subtidal currents of total velocities are also assessed using the ADCP data. Despite differences in a variety of aspects between the direction-finding CODAR SeaSonde (long range, effective depth of 2.4 m, integration time of 4 h, and idealized antenna patterns) and the beam-forming WERA (median range, effective depth of 0.9 m, and integration time of 1 h), both HF radar systems demonstrated good surface current mapping capability. The differences between the velocities measured with the HF radar and the ADCP are sufficiently small in this low-energy shelf that much of these rmsd values may be accounted for by the expected measurement differences due to the horizontal, vertical, and temporal sampling differences of the ocean current observing systems used.

Collaboration for Prediction of Red Tides Contribution Number 31.

Corresponding author address: Yonggang Liu, College of Marine Science, University of South Florida, 140 7th Ave. S, MSL 119, St. Petersburg, FL 33701. E-mail: yliu18@gmail.com

1. Introduction

Shore-based high-frequency (HF) radars with operating frequencies of 3–50 MHz are routinely used for remotely sensing coastal ocean surface currents (e.g., Paduan and Washburn 2013). Presently there are around 100 such HF radars in operation as part of the U.S. Integrated Ocean Observing System (e.g., Paduan et al. 2004; Seim et al. 2009; Harlan et al. 2010). Among these applications, there are primarily two types of commercially available HF radars, the Coastal Ocean Dynamics Application Radar (CODAR; Barrick et al. 1985) and the Wellen Radar (WERA; Gurgel et al. 1999).

CODAR long-range SeaSonde radars, designed for seawater applications by CODAR Ocean Sensors, Ltd., utilize a compact directional antenna system consisting of three antenna elements (two orthogonally mounted loops and a monopole whip) in a single housing for receiving and a separate omnidirectional antenna for transmitting frequency-modulated interrupted continuous wave (FMICW) pulses. The cross spectra, computed from the time series of voltages from the different receive antennas, are used to infer the range and speed of radial velocity components within sectors of a polar map from the Bragg scattering principle and Doppler frequency shifts (e.g., Crombie 1955; Barrick et al. 1977; Lipa and Barrick 1986; Paduan and Graber 1997). The bearing of the radial velocity component is determined through a direction-finding algorithm known as multiple signal classification (MUSIC; e.g., Schmidt 1986; Barrick and Lipa 1997). Two or more SeaSonde radars looking at the same area, but from different angles, are used to generate the total velocity vector (current) field.

WERAs, developed by the University of Hamburg (Gurgel et al. 1999), are manufactured by Helzel Messtechnik GmbH (Helzel). A WERA system is composed of two separate antenna arrays, one for receiving and the other for transmitting frequency-modulated, continuous wave (FMCW) chirps. The transmitting array often has 1–4 antennas and the receiving array has 12–16 antennas. The bearing of the radial velocity is determined through a beam-forming process that uses the phase differences between signals received on the antenna array elements that depend on the direction of arrival (e.g., Helzel et al. 2006). When the Doppler spectra of the individual array elements are summed with the proper phase differences applied for a given bearing, a digital narrow beam is formed to a selected direction and a peak-picking algorithm is applied to the resultant spectrum to determine the surface current speed and direction. WERA systems can also be configured to be in direction-finding mode, using a compact receiving array of four antennas set in a square (e.g., Helzel et al. 2010).

CODAR SeaSonde direction-finding HF radars are widely used throughout the U.S. coastline, for example, in California (e.g., Paduan and Rosenfeld 1996; Halle and Largier 2011; Kim et al. 2011), Oregon, and Washington (e.g., Kosro et al. 1997); the Mid-Atlantic Bight (e.g., Kohut et al. 2004; Roarty et al. 2010; Ullman et al. 2012); and the West Florida shelf (WFS, e.g., Merz et al. 2012). In contrast, WERA phased-array beam-forming HF radars are operated only at a few selected sites along the U.S. coasts, such as the Miami coast (e.g., Shay et al. 1995, 2002), the Hawaiian coast (e.g., Zaron et al. 2009; Chavanne et al. 2010), the continental shelf of Georgia and South Carolina (Savidge et al. 2011), and the WFS (e.g., Merz et al. 2012). Simultaneous operations of the CODAR SeaSonde direction-finding and WERA phased-array HF radars in the same region are rare, an exception being Merz et al. (2012) for the WFS.

Early validation studies of HF radar–derived currents were mainly based on drifter (e.g., Stewart and Joy 1974; Barrick et al. 1977) and moored single-point current meter observations (e.g., Holbrook and Frisch 1981). With the advancement of ocean-current-measuring technologies, acoustic Doppler current profilers (ADCPs) were then also used to validate HF radar–derived currents (e.g., Graber et al. 1997). An early review of HF radar validation is provided by Chapman and Graber (1997). With the increasing number of deployments of HF radars for measuring currents in coastal oceans, validations against a variety of in situ observations have expanded considerably in the last two decades, both for CODAR (e.g., Emery et al. 2004; Paduan et al. 2006; Ohlmann et al. 2007; Cosoli et al. 2010; Kohut et al. 2012) and WERA (e.g., Shay et al. 2007; Savidge et al. 2011; Robinson et al. 2011).

Only a few direct comparisons exist for different types of HF radars in measuring surface currents. Fernandez and Paduan (1996) performed simultaneous current measurements using both directional-finding CODAR SeaSondes and beam-forming ocean surface current radars (OSCRs) in Monterey Bay, and compared them against 2 days of ADCP observations at 9-m depth. In a 6-week experiment, Teague et al. (1998) brought together three HF radar systems (multifrequency phased-array radars, OSCARs, and CODAR SeaSondes) to measure the Chesapeake Bay outflow plume. Teague et al. (2001) further compared the multifrequency HF radar data with the ADCP measurements at 2-m depth, and found that the HF radar measurement of surface currents were accurate to about ±10 cm s−1. Gurgel et al. (1997) reported a comparison of current velocities derived from direction-finding and beam-forming techniques. In a 6-day experiment, Essen et al. (2000) compared both CODAR (although not a SeaSonde) and WERA surface current measurements with an S4 electromagnetic current meter, and obtained root-mean-square difference (rmsd) values of 14–15 and 11–13 cm s−1 for CODAR and WERA radial velocity components, respectively. Note that these experiments were performed over short durations (several days to weeks) in 1995–96, some 18 years ago, and these rmsd values were generally larger than those found in recent years. The quality and reliability of the in situ current measurement devices along with the CODAR and WERA systems and the data processing algorithms have been improved in recent years (e.g., Helzel et al. 2010). To our knowledge, comparisons of CODAR SeaSonde and WERA systems for simultaneously measuring currents against moored ADCPs have not previously been reported.

For the past 20 years, and as part of a coordinated coastal ocean observing and modeling system, the Ocean Circulation Group at the University of South Florida has deployed various observing components on the WFS (e.g., Weisberg et al. 2005, 2009a). Ocean current measurements with moored ADCPs began in 1993, first at a single location and then at multiple locations across the shelf (e.g., Weisberg et al. 1996, 2009b; Liu and Weisberg 2012). A CODAR long-range SeaSonde HF radar was installed on the west Florida coast in 2003, and two others were added over the next two years to form an HF radar network (Liu et al. 2010; Merz et al. 2012). These combined HF radar and moored ADCP measurements on the WFS were used in describing the three-dimensional coastal ocean circulation on different time scales (Liu et al. 2007), gauging the coastal-altimetry-derived surface geostrophic velocity (Liu et al. 2012) and improving model simulations through data assimilation (Barth et al. 2008). Two WERA HF radar systems were then added to the region in summer 2010, increasing the total number of simultaneously operating HF radars within this WFS network to five. Thus, we established a test bed with HF radars overlooking an array of moored ADCPs. Especially relevant is the collocation of CODAR and WERA HF radars at the Venice, Florida, site with a moored ADCP within their collective footprint (Fig. 1). These coastal observing assets provided a unique opportunity to evaluate the performance of the two types of HF radars for measuring surface currents.

Fig. 1.
Fig. 1.

Map of CODAR and WERA HF radar footprints in the WFS. The long-range CODAR sites include Redington Shores, Venice, and Naples; and the medium-range WERA sites Ft. De Soto and Venice. Locations of moored ADCPs are shown as small red triangles (C10, C12, and C13).

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Based on multiyear concurrent HF radar–ADCP measurements on the WFS, Liu et al. (2010) assessed CODAR SeaSonde performance in measuring the surface currents. Despite the challenge of achieving a long-term record of continuous backscatter from a region of high lightning occurrence, frequent storms, and offshore low-energy environment (Weisberg et al. 2012), the CODAR data compared well with the ADCP measurements. The rmsd values ranged from 6 to 10 cm s−1 for hourly and 3–6 cm s−1 for low-pass filtered radial velocity components. Bearing offsets were in the range of −15° to +9°, and coherent variations of the CODAR and ADCP radial currents were seen across both the tidal and subtidal frequency bands.

The present study is a continuation of Liu et al. (2010), extending the performance evaluation to include the WERA system. The main purpose of this paper is to assess the performance of the two types of commercially available HF radars, CODAR SeaSonde and WERA, in measuring surface current velocity on the WFS gauged against moored ADCP records. The configurations of the HF radar systems are described in section 2, followed by the data processing in section 3. Comparisons between the velocity components and vectors measured with HF radar and ADCP are presented in section 4, followed by a discussion in section 5 and a summary in section 6.

2. Ocean current observing systems

To improve our understanding and predictive capability for coastal ocean processes in a variety of environmental applications (e.g., Weisberg 2011; Liu et al. 2011; Weisberg et al. 2014), the University of South Florida’s College of Marine Science initiated a real-time WFS Coastal Ocean Monitoring and Prediction System (COMPS) in 1998 (e.g., Merz 2001). The COMPS program’s observing assets presently consist of offshore buoys, shore-based HF radars, and coastal tidal stations. The HF radar network consists of three direction-finding long-range CODAR SeaSonde HF radars and two phased-array WERA systems, with one site (Venice) containing collocated CODAR SeaSonde and WERA systems. Evolution of the HF radar network on the WFS is reported in Merz et al. (2012).

The three long-range CODAR SeaSonde HF radars, operating at a nominal frequency of 4.90 MHz, are deployed along the west Florida coast (Fig. 1) at Redington Shores (27°49.937′N, 82°50.032′W), Venice (27°04.655′N, 82°27.096′W), and Naples (26°09.729′N, 81°48.632′W). Such 5-MHz radars estimate radial velocity components (radials) at an effective depth of 2.4 m (e.g., Stewart and Joy 1974; Paduan and Graber 1997), with nominal range and bearing resolutions of 5.8 km and 5°, respectively. Table 1 provides the specific SeaSonde settings. The radials are processed using manufacturer-supplied software. Because of a lack of field measurements, idealized antenna patterns are used. To improve data returns in this low-energy environment, five successive raw cross spectra (with 2048 points FFT, output every 34 min) are averaged over a total of 4 h, with the averaging interval advanced each hour. These (4 h) averaged radials are archived as hourly time series.

Table 1.

Configuration settings of CODAR SeaSonde and WERA HF radars.

Table 1.

The two medium-range WERA HF radars are installed in Pinellas County’s Fort De Soto Park (27°38.16′N, 82°44.30′W) and at U.S. Coast Guard Auxiliary Station 86 in Venice. The Fort De Soto site has one transmit antenna, and the Venice site has four transmit antennas arranged in a rectangular array. Both sites have 12 receiving antennas configured alongshore in linear arrays. These radars operate between frequencies of 12.275 and 13.20 MHz. Specific configurations of the WERA are listed in Table 1. Different from CODAR SeaSonde systems, the actual WERA transmit frequency and bandwidth is determined by first scanning the entire Federal Communications Commission (FCC)-licensed 1-MHz operational band prior to each acquisition via WERA’s frequency-adaptive “Listen before Talk Mode” software to determine the region of lowest noise level and corresponding bandwidth (Helzel 2007). These WERA HF radars have a higher resolution coverage area within the larger CODAR footprint (Fig. 1). The WERA systems also use 2048 spectral Doppler bins in the FFT. The radials and vector totals are processed using standard WERA-supplied software with a nominal range and bearing of 1.5 km and 8°, respectively, at broadside (90°), changing to 1.5 km and 20° at ±60°, respectively [an example of the 16-element system is provided by Heron and Atwater (2013)]. The WERA data are archived every 20 min.

Three surface moorings with Teledyne RD Instruments ADCPs, measuring vertical profiles of velocity throughout most of the water column, are located within the CODAR footprint (C10, C12, and C13), and two of them (C10 and C12) are located within the WERA footprint (Fig. 1). Specific settings of the ADCPs are listed in Table 2. The ADCPs are claimed by the manufacturer to be accurate to 0.3% of water velocity ±0.3–0.5 cm s−1, which is small compared to the current variability on the WFS, where the range of two standard deviations for the along-shelf currents is about ±16 cm s−1 (e.g., Liu and Weisberg 2005a,b, 2012). The relative distances ρ between the moorings and the HF radar site origins, and the moorings’ bearing angles θ to the HF radar site origins, are listed in Table 3.

Table 2.

ADCP mooring information.

Table 2.
Table 3.

Comparison statistics of HF radar–ADCP radial currents. Here, ρ denotes the relative distances between a mooring (C10, C12, and C13) and an HF radar site origin, and θ measures the mooring’s bearing angle to the HF radar site origin, clockwise from north (i.e., 0° = north, 90° = east).

Table 3.

3. Data processing

The hourly CODAR SeaSonde radials are interpolated onto uniform radial grids (bearing and range intervals of 5° and 5.8 km, respectively) using a community toolbox HFR_Progs (https://cencalarchive.org/~cocmpmb/COCMP-wiki/index.php/HFR_Progs_download_page), which is the implementation of the least squares method as described in Lipa and Barrick (1983). Outliers (unusually large speeds >50 cm s−1) are removed prior to the interpolation. Data gaps are filled by interpolation over one missing bin in range and two missing bins in bearing. Total velocity vectors are generated from the interpolated radials using the HFR_Progs software.

Both the radials and vector totals processed by the WERA software suite are on rectangular grids. Also, discrepancies resulting from ionospheric contamination and very short bursts of interference/lightning are removed within the CODAR software suite, so that they do not bias the FFT analysis of Doppler shift. Unlike for CODAR, these discrepancies are not automatically removed in the current version of the WERA processing software, so outliers are sometimes seen in the WERA radial and total velocity data. For each WERA radial velocity data map, the “accuracy” (e.g., Parks et al. 2009) of the radial component of the surface currents is provided as a separate column in the radial data file. We examine the remaining radial currents by setting different accuracy values. Fewer outliers are seen if a smaller accuracy value is chosen. However, this is at the expense of removing more potentially valid data points from the map (Fig. 2). By empirically setting a threshold of 3 cm s−1 in velocity accuracy, these outliers can be effectively removed, while a good number of the radial currents are retained. Outlier removal is important because such points may result in large differences in the comparison statistics (e.g., Kirincich et al. 2012). The WERA radials are interpolated onto uniform radial grids (bearing and range intervals of 2° and 2 km, respectively) and combined into total velocities using the HFR_Progs software. The 20-min velocity time series are further smoothed (by averaging in time among the adjacent three data points) to produce hourly time series.

Fig. 2.
Fig. 2.

A snapshot of WERA radial velocity vectors with different accuracy (acc) thresholds: (a) all the available data, (b) acc < 5 cm s−1, (c) acc < 4 cm s−1, (d) acc < 3 cm s−1, (e) acc < 2 cm s−1, and (f) acc < 1.5 cm s−1.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

We choose the time period of October–December 2011 for comparison, because the long-range CODAR SeaSonde HF radars generally have higher data returns in winter months on the WFS (Liu et al. 2010). Data returns generally exceed 80% in time over majority of the CODAR footprints (Fig. 3). Among the three CODAR sites, Venice has the highest and Naples has the lowest data return, which is consistent with previous findings (Liu et al. 2010). The coverage areas of two WERA HF radars are smaller than those of CODAR SeaSonde radars; however, their data returns are generally higher than CODAR’s within the 60–70 km ranges from the site origin (Fig. 3). The coverage is related to the presence of Bragg waves, which are different for the HF radar transmit frequencies (Table 1). Mooring C10 is located within the high data return areas of Redington Shores and Venice radars, but on the edge of the high data return area of the Fort De Soto WERA radar. Mooring C12 is located within the high data return areas of Redington Shores and Venice CODAR radars but not within the high data return area of the WERA site, while mooring C13 is located within the footprints of Venice and Naples CODAR radars.

Fig. 3.
Fig. 3.

Percent temporal coverage of valid radial currents for the three CODAR HF radars: (a) Redington Shores (RdSr), (b) Venice (Veni), and (c) Naples (Napl); and two WERA HF radars: (d) Ft. De Soto (FtDS) and (e) Venice (Veni). The ADCP stations are also shown in the map as triangles.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

4. Velocity comparison

We assess the two HF radar systems in terms of both radials (including radial speed and bearing offset) and totals. The radial speed comparison is mainly based on mooring C10 data. The spatial variability of the CODAR current mapping is further gauged using moorings C12 and C13.

a. Radial speed

The radials are bilinearly interpolated onto the mooring sites to get VHF, where V represents velocity. The ADCP velocity vectors are projected into the same direction as the mooring’s bearing angle θ (see Table 3) to the HF radar site origin to get VADCP. We quantify the comparison of VHF and VADCP scalars using linear correlation and regression analyses. Commonly used statistical measures include 1) rmsd; 2) coefficient of determination r2, which is the squared correlation coefficient; and 3) regression coefficient b and mean bias c, which are defined as VADCP = b × VHF + c. Hourly and 36-h low-pass filtered time series of radial pairs VHF and VADCP for CODAR SeaSonde are shown in Figs. 46, and those for WERA are shown in Fig. 7. The statistical measures are summarized in Table 3.

Fig. 4.
Fig. 4.

Comparison of (a),(c) hourly and (b),(d) 36-h low-pass filtered CODAR and ADCP radial current time series at mooring C10 for (top) RdSr, and (bottom) Veni sites. Statistics are shown in the top-right corner of each panel: rmsd, r2, b, and c, with their 95% confidence levels in the parentheses estimated following Emery and Thompson (2001, p. 253).

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Fig. 5.
Fig. 5.

As in Fig. 4, but for mooring C12 and the CODAR sites RdSr and Veni.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Fig. 6.
Fig. 6.

As in Fig. 4, but for mooring C13 and the CODAR sites Veni and Napl. Note that the ADCP data are available only during 1–25 Oct 2011, and the time axis is different from that of Fig. 5.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Fig. 7.
Fig. 7.

As in Fig. 4, but for the WERA sites FtDS and Veni.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

1) CODAR SeaSonde versus ADCP

The two CODAR SeaSonde HF radars at Redington Shores and Venice compare well with mooring C10 for radials (Fig. 4). The rmsd values are 5.1–5.5 cm s−1 for hourly data, and they are both reduced to 2.8 cm s−1 for 36-h low-pass filtered data. These rmsd values are even slightly lower than those previous comparisons for the same station pairs, but for a different (multiple years, 2003–08) time period (Liu et al. 2010). The r2 values are 0.44–0.66 and 0.76–0.86 for hourly and 36-h low-pass filtered time series, respectively. The regression coefficients are 0.75–0.93 and 0.86–1.00 for hourly and 36-h low-pass filtered time series, respectively.

The Redington Shores site is more distant from mooring C10 than the Venice site (ρ = 75 vs 48 km), but the comparison is slightly better for Redington Shores than for Venice—for example, rmsd = 5.1 vs 5.6 cm s−1, r2 = 0.66 vs 0.44, and b = 0.93 vs 0.75, respectively, for hourly radials (Fig. 4). The bearing direction of mooring C10 to Redington Shores is more aligned in the along-shelf direction, while that to Venice is roughly in the across-shelf direction (Fig. 1). The currents on the inner shelf are mostly wind driven, and the dominant subtidal currents are generally polarized, with a stronger along-shelf component and a weaker across-shelf component (e.g., Liu and Weisberg 2005b, 2007).

At the offshore location C12, the comparisons between CODAR SeaSonde and ADCP find increased rmsd values of 9.0–9.2 and 6.0–6.4 cm s−1 for hourly and 36-h low-pass filtered time series, respectively (Fig. 5). The r2 and b values are also generally reduced. Larger rmsd at C12 relative to those at C10 may be partially due to the HF radar sampling area changes as a function of range (larger radar sector areas in longer ranges). Note that mooring C12 is located 98 and 133 km from Redington Shores and Venice, respectively, while C10 is much closer to the two radar sites (ρ = 75 and 48 km, respectively). Ohlmann et al. (2007) attributed about 5 cm s−1 rmsd to spatial sampling differences in their study case. On the other hand, currents may have larger variability on the midshelf than the inner shelf (e.g., Liu and Weisberg 2005b); the spatial inhomogeneity in currents also contributes to the rmsd between HF radar and ADCP data (e.g., Chapman and Graber 1997).

Examining another offshore mooring, C13, the CODAR SeaSonde–ADCP radial comparisons (Fig. 6) also find larger velocity differences with respect to mooring C10 because of the longer ranges from the HF radar sites. The rmsd values are 8.5–8.5 and 4.3–5.3 cm s−1 for the hourly and low-pass filtered time series, respectively.

2) WERA versus ADCP

The two WERA HF radars at Fort De Soto and Venice also compare well with mooring C10 in measuring radials (Fig. 7). The rmsd values are 3.8–6.5 cm s−1 for hourly time series, and they are reduced to 2.2–4.3 cm s−1 for the 36-h low-pass filtered data. The rmsd of 3.8 cm s−1 for the Venice site is close to the previous comparison (4.1 cm s−1) for the same station pair, but for a different (1 month in 2003) time period (Shay et al. 2007).

The WERA–ADCP radial current difference for the Fort De Soto site is generally larger than that of the Venice site, with rmsd = 6.5 versus 3.8 cm s−1, r2 = 0.38 versus 0.77, and b = 0.63 versus 1.07, respectively. This is mainly because the C10 ADCP is located on the outer edge of the WERA Fort De Soto site footprint (Fig. 1); it is also on the outer edge of that site’s high data return area (Fig. 3d). The differences over the outer edge are expected to be larger than those in the inner part of the radar footprint due to a large steering angle that leads to a wider beam.

3) CODAR SeaSonde versus WERA

Compared to the CODAR SeaSonde radar at the same site (Venice), the WERA HF radar performs slightly better as gauged against the data at C10: lower rmsd values (3.8 vs 5.6 cm s−1 for hourly, and 2.2 vs 2.8 cm s−1 for 36-h low-pass filtered data), higher r2 (0.77 vs 0.44 for hourly data, and 0.88 vs 0.73 for 36-h low-pass filtered data), and b values closer to 1 (1.07 vs 0.75 for hourly data, and 1.10 vs 0.86 for 36-h low-pass filtered data). Here, the rmsd values are much smaller than those gauged against the S4 current meter records of 19 days, which are 11 and 14 cm s−1 for WERA and CODAR, respectively (Essen et al. 2000).

Note that the WERA comparison results are based on hourly data, which are obtained from 20-min acquisitions using a three-point moving average (with 1-h window). However, the CODAR SeaSonde hourly data are averaged (through the cross-spectra processing using the CODAR SeaSonde standard software) over a 4-h window. If the moving average window is changed to 4 h (averaging over 13 points of the 20-min data) for WERA data, then the WERA–ADCP comparisons become even better: the rmsd for the Venice site is reduced to 3.3 cm s−1 for hourly data (same 2.2 cm s−1 for 36-h low-pass filtered data); the rmsd values for the Fort De Soto site are also reduced to 5.7 and 4.3 cm s−1 for hourly and 36-h low-pass filtered data, respectively.

b. Bearing offset

The resolution in azimuth (bearing direction) of radial currents is another important measure of HF radar performance. For direction-finding systems such as the CODAR SeaSonde radars in this study, bearing offsets may be attributed to many factors, including the resolution capabilities of the combined receiving antenna and direction-finding algorithms (e.g., Barrick and Lipa 1986, 1997; Laws et al. 2000; Emery et al. 2004; Paduan et al. 2006; De Paolo and Terrill 2007). Since the direction of a received signal is derived from the processed signals in the frequency domain by means of the MUSIC algorithm (Schmidt 1986), the azimuthal resolution is related with the integration time (FFT length) and the signal-to-noise ratio (SNR) (e.g., Barrick and Snider 1977; Lipa and Barrick 1983). For beam-forming systems such as the WERA radars in this study, the azimuthal resolution depends on the number of antenna elements and their spacing (Gurgel et al. 1997; Helzel et al. 2010). Incorrectly calibrated phases among antenna elements, cables, and receivers can also cause errors in the beam-steering direction and distortions of the beam pattern (Flores-Vidal et al. 2013).

We estimate the bearing offset following the procedures described in Emery et al. (2004) and Liu et al. (2010). For a given range to a radar site, corresponding to a given mooring location, all the VHF are sampled for all of the bearing angle sectors and then correlated with the VADCP at the given mooring location. Thus, with VADCP fixed and VHF varying over all radial sector bins for a given radial range, the bearing angle sector exhibiting the highest correlation is thought to be the one most indicative of the correct angle sector. With no offset, the radial sector containing the mooring would be expected to exhibit the highest correlation, but this is not always the case due to the presence of the sources of the errors explained above. Bearing offset is therefore defined as Δθ = θr − θm, where θm is the bearing to the mooring, and θr is the bearing to the center of the sector with the maximum VADCP and VHF correlation (r2), and with positive Δθ indicating that the sector is displaced clockwise from the mooring.

Bearing offset analysis results are provided for all the CODAR SeaSonde–ADCP pairs in Fig. 8. The r2 values are shown for each pair as a function of bearing along with vertical lines indicating the bearing to the particular radar and the bearing corresponding to the maximum r2, the offsets being defined as the differences between these two vertical lines. Among the six comparisons, two show that the peak r2 is correctly positioned within the 5° sector where VADCP is measured, that is, |Δθ| 5°. The other bearing offsets fall in the range of 11° θ| ≤ 16°. These bearing offsets are slightly larger than previous values obtained from the multiyear records for CODAR SeaSonde (Liu et al. 2010), mainly because only idealized antenna patterns are used in this study, while a mixed idealized and measured antenna patterns were used in Liu et al. (2010). It is known that directional calibration using measured antenna patterns (Barrick and Lipa 1986) can effectively reduce the distortion by the local environment (Kohut and Glenn 2003). However, these bearing offsets found on the WFS are still slightly smaller than those found for the 13-MHz CODAR SeaSonde on the U.S. West Coast (Emery et al. 2004). It is expected that bearing offsets could be as large as 30° if idealized antenna patterns are used (Paduan et al. 2006).

Fig. 8.
Fig. 8.

Bearing offset (θr–θm) for three CODAR sites evaluated in terms of the ADCP measurements at moorings C10, C12, and C13, where θr is the bearing to the center of the sector with maximum r2 between the HF radar and ADCP radial currents, and θm is the bearing to the mooring. Solid lines indicate r2 values along a certain ranges ρ of HF radar corresponding to the ADCP location.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

The bearing offset analysis is similarly performed for the two WERA–ADCP station pairs (Fig. 9). The Fort De Soto site has a large bearing offset |Δθ| of 32°, mainly because C10 is located near the edge of the WERA beam-steering range (Fig. 1). Both large rmsd values and the low valid data return rate are found in this area of the WERA footprint. The Venice site has a bearing offset |Δθ| of 8°, which is reasonably good for a radar system without any antenna field calibration. The WERA beam-forming methods use a “self-calibration” algorithm (Helzel and Kniephoff 2010), and the azimuthal accuracy is expected to be excellent, up to values better than ±1° (Helzel et al. 2010). However, our analysis result does not show such a high WERA system accuracy.

Fig. 9.
Fig. 9.

As in Fig. 8, but for two WERA sites evaluated in terms of the ADCP measurements at mooring C10.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

The bearing offsets of the CODAR SeaSonde and WERA at the Venice site evaluated with mooring C10 are −14° and −8°, respectively. So, without any measured antenna pattern calibrations, the WERA system with a linear receiving array of 12 antenna elements has a slightly smaller azimuthal error than the CODAR SeaSonde system.

The method of bearing offset estimation is not perfect. Reliable results may be obtained if the currents significantly change with bearing angle. In the case of low current velocities, the calculated correlation coefficients as a function of bearing may have a flat peak (Fig. 9a) or multiple peaks (Figs. 8c,e), and the estimated bearing offsets may have relatively large errors. A potentially more reliable method to measure bearing offsets would be to make use of ship echoes received by HF radar (e.g., Fernandez et al. 2003; Gurgel et al. 2010; Flores-Vidal et al. 2013), but this is beyond the scope of the present study.

c. Total velocity

Two or more radial current maps with overlapping areas may be combined in the least squares sense (Lipa and Barrick 1983) to get velocity vectors using a prescribed geometrical dilution of precision (GDOP; e.g., Chapman et al. 1997). During this combining process, the radial errors may be transformed into the velocity errors (e.g., Barth et al. 2010), depending on the angle at which the radial currents intersect. Thus, the current field as given by the velocity vectors at any time may have spatially variable errors.

1) Vector correlation

A correlation between two vector time series, as proposed by Kundu (1976), provides a quantitative comparison between the HF radar–estimated and the ADCP-measured velocity vectors. The vector correlation consists of the complex correlation coefficient,
e1
and the complex phase angle,
e2
where represents an average; and and are the Cartesian velocity components for the HF radar and ADCP vectors, respectively. The phase angle α represents the average veering angle between the two vector time series. Examples of such vector correlation usage are found for oceanographic data and model analyses, including a comparison of HF radar and ADCP currents (e.g., Essen et al. 2003; Kaplan et al. 2005; Shay et al. 2007; Parks et al. 2009; Weisberg and He 2003; Zheng and Weisberg 2012).

CODAR SeaSonde and ADCP velocity vectors have high correlation (Fig. 10), with a γ of 0.87 and 0.94, and an average α of −3.9° and −5.2°, for hourly and 36-h low-pass filtered data, respectively (Table 4). The CODAR SeaSonde current vectors generally appear to rotate clockwise from the ADCP currents at the 4-m level. High correlation is also seen between the WERA and the ADCP vectors, with a slightly lower complex correlation coefficient (γ = 0.85 and 0.92 for hourly and 36-h low-pass filtered data, respectively) and a slightly larger veering angle (α = 6.2° and 8.3° for hourly and 36-h low-pass filtered data, respectively). The rotation, however, is in the other direction (counterclockwise from the 4-m ADCP vectors). The slight decrease in WERA–ADCP total current comparison may be attributed to the C10 mooring being located on the edge of the Fort De Soto WERA site footprint. The larger error in the radial currents (rmsd = 6.5 cm s−1 for the Fort De Soto–C10 pair of hourly radials) increases the error of the WERA velocity vectors. We note that for such shallow water locations, the angular rotation from surface to bottom should be anticlockwise by virtue of the bottom Ekman layer, consistent with the CODAR SeaSonde finding but not with the WERA finding, which again may be the result of increased WERA error due to the C10 mooring position.

Fig. 10.
Fig. 10.

Stick plots of 36-h low-pass filtered (a) CODAR, (b) WERA, and (c) ADCP total current time series at site C10. Vector correlation statistics between HF radar and ADCP data are shown in the top-right corner of each panel: γ and α for the hourly and 36-h low-pass filtered time series, respectively. A negative angle indicates the HF radar velocity vectors rotate clockwise from the ADCP velocity vector.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Table 4.

Comparison statistics of HF radar–ADCP total currents.

Table 4.

The correlation between CODAR SeaSonde and ADCP total currents decreases at the farther offshore location of C12 (Fig. 11). The γ value is 0.63 and 0.73 for hourly and 36-h low-pass filtered data, respectively. The α values are also slightly larger than those at C10. There are only 23 days of CODAR SeaSonde–ADCP overlaps at C13 (stick plots not shown), and the γ values are 0.61 and 0.77 for hourly and 36-h low-pass filtered data, respectively. So, the quantitative vector comparisons for the CODAR SeaSonde decreases with increasing HF radar range.

Fig. 11.
Fig. 11.

As in Fig. 10, but for 36-h low-pass filtered (a) CODAR and (b) ADCP total current time series at mooring C12.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

2) Cartesian components

We evaluate the Cartesian components of the velocity vectors at location C10 using the same statistical measures as for the radial current components. The comparison statistics are listed in Table 4. For hourly values of the CODAR SeaSonde and the ADCP, the rmsd of u and υ at location C10 are 3.9 and 4.0 cm s−1, respectively. These rmsd values increase to 8–11 cm s−1 at the offshore locations C12 and C13. For 36-h low-pass filtered data, the rmsd values of u and υ at location C10 are 2.4 and 2.7 cm s−1, respectively, and these values increase to 4.7–7.6 cm s−1 at the offshore locations C12 and C13.

The Cartesian components of WERA and ADCP are compared at location C10 only (Table 4). The rmsd values of u is 3.7 and 2.4 cm s−1 for hourly and 36-h low-pass filtered time series, respectively, which is very close to those of CODAR SeaSonde. The rmsd values of υ is 5.1 and 3.5 cm s−1 for hourly and 36-h low-pass filtered time series, respectively, which is slightly larger than those of CODAR SeaSonde.

3) Tidal currents

The C10 location is used to examine the HF radars’ capability for observing the tidal currents. Limited data gaps exist in the WERA observations during early October 2011 and in the CODAR SeaSonde observations during late December 2011 (Fig. 10), so these portions of data are omitted from the tidal analyses. A common period, 16 October through 24 December 2011, is selected for analysis because the CODAR SeaSonde, WERA, and ADCP hourly data are almost continuous during this period. The major tidal constituents (M2, S2, K1, and O1) are estimated using the T-TIDE toolbox (Pawlowicz et al. 2002), and the four tidal ellipses are shown in Fig. 12. The CODAR SeaSonde–derived M2 tidal ellipse is smaller than that of the ADCP observations, while the WERA-derived M2 tidal ellipse is more polarized than that of the ADCP data (Fig. 12). The semimajor axis of the M2 tidal ellipses is 3.2, 5.0, and 4.0 cm s−1 for the CODAR SeaSonde, WERA, and ADCP data, respectively; and the semiminor axis is 1.7, 1.6, and 2.5 cm s−1 for the CODAR SeaSonde, WERA, and ADCP data, respectively. The ADCP results are consistent with previous tidal analysis on the shelf (e.g., He and Weisberg 2002). The smaller tidal amplitudes observed in the CODAR SeaSonde data may be caused by the averaging of the raw cross spectra in the data processing using the manufacturer standard software (to increase data return). Averaging over a 4-h sliding window may result in a reduction of amplitude at these relatively high frequencies. The increased eccentricity of the M2 ellipse observed in the WERA may be due to the larger errors of radial currents from the Fort De Soto radar at site C10 as previously mentioned. Also note that the effective depth of measurement for WERA is about 1 m below the surface (Table 1), whereas the topmost ADCP bin is at 4-m depth (Table 2). The 95% confidence intervals for M2 and S2 major axes are all about 0.4–0.5 cm s−1 for the CODAR SeaSonde, WERA, and ADCP data, and those for K1 and O1 are even larger (0.9–1.1 cm s−1). Considering these uncertainties in the tidal analysis results, the HF radars are capable of capturing the major tidal currents to the extent permitted by the temporal and spatial sampling differences of the instruments.

Fig. 12.
Fig. 12.

Comparison of tidal ellipses of the four major tidal constituents—(a) M2, (b) S2, (c) K1, and (d) O1—calculated with CODAR, WERA, and ADCP velocity data at site C10.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

4) Rotary power spectrum

To compare the HF radar–ADCP data across the entire frequency domain, we perform rotary power spectral analyses on the same velocity data that are used in the tidal analyses. The rotary power spectral analysis method is described in Mooers (1973) and Emery and Thompson (2001). The CODAR SeaSonde, WERA, and ADCP data all show a higher clockwise rotation of the velocity hodographs; that is, the clockwise rotary power spectra are much larger than the counterclockwise power spectra (Fig. 13). The diurnal and semidiurnal spectral peaks are well resolved in the clockwise rotary spectra. The WERA data exhibit slightly higher energy and the CODAR SeaSonde data slightly lower energy than the ADCP data in the semidiurnal band. In the diurnal band, the CODAR SeaSonde data compare better with the ADCP than the WERA data. These results are consistent with the tidal analyses. Also note that the noise level in the high-frequency band (periods < 0.25 day) is lower for the CODAR SeaSonde data relative to the other two datasets. This is also consistent with the explanation of smoothing in CODAR raw data processing.

Fig. 13.
Fig. 13.

Comparison of the rotary power spectra calculated with CODAR, WERA, and ADCP velocity data at site C10. Terms S++ and S−− are average spectral energy density associated with the (a) clockwise and (b) counterclockwise components of the velocity hodograph ellipse. Also shown are the 90% confidence intervals.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

5. Discussion

The comparison of velocity components measured with HF radar and ADCP is subject to numerous sources of errors beyond those of the instruments (e.g., Chapman and Graber 1997; Graber et al. 1997; Ohlmann et al. 2007). These include, but are not limited to, the horizontal, vertical, and temporal sampling differences between the different measurements.

To examine the horizontal variability of the HF radar currents, we use a method proposed by Graber et al. (1997), as suggested by a reviewer. The expected rmsd values Eυ of the radial velocities between two radial currents as a function of separation distance x between successive observations can be estimated from
e3
where συ is the standard deviation of the current speed and ρυ is the autocorrelation between two radial velocity values separated by the distance x. For each CODAR SeaSonde site, the expected rmsd values are calculated from hourly radial velocities along each radial direction, and then averaged across the bearing angle for each radial map and over time. For each WERA site, the expected rmsd values are similarly calculated but along the meridional and zonal directions instead (based on the data in rectangular coordinates). In general, the expected rmsd values of the WERA radial velocities are smaller than those of the CODAR SeaSonde data (Fig. 14). Based on these curves, we can estimate an upper bound of the expected rmsd due to spatial separation for a given radar radial sector. The half-arc lengths of radar sectors for moorings C10, C12, and C13 are 2.0, 5.8, and 5.4 km, respectively, on the Venice CODAR SeaSonde radial grid. These lengths correspond to maximum possible spatial separations of a mooring within a radial sector of 4.2, 11.6, and 11.2 km, respectively, given a range cell resolution of 5.8 km. From Fig. 14, these spatial separations correspond to the expected differences of 3.5, 5.8, and 5.7 cm s−1, respectively, in the ADCP–CODAR SeaSonde comparisons at the Venice site; that is, the expected rmsd values are larger for the offshore sites. The expected spatial differences for all the HF radar–ADCP pairs are listed in Table 5. The expected spatial differences of WERA–C10 at Fort De Soto and Venice are 3.3 and 2.3 cm s−1, respectively, with a larger value at the Fort De Soto site as expected.
Fig. 14.
Fig. 14.

Expected rmsd between radial velocities as a function of spatial separation.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Table 5.

Expected differences between HF radar and ADCP radial current measurements. Here, d denotes the maximum spatial separation of a mooring (C10, C12, and C13) within an HF radar (RdSr, Veni, Napl, FtDS) radial sector. Also listed are expected horizontal velocity differences (Δυx), current vertical shears (Δυz), total expected velocity differences (Δυ = Δυx + Δυz), the actual rmsd values (from Table 3), and their ratios (Δυ/rmsd).

Table 5.

In the present study, the 4.9-MHz CODAR SeaSondes effectively measure currents at 2.4 m below the surface, while the 12.7-MHz WERAs measure currents at 0.9 m (Table 1). Note that the topmost bins of the ADCP velocity records are at 4 m (Table 2). Recently, Kohut et al. (2012) studied the vertical differences of the velocity from the ADCP records in the Mid-Atlantic Bight, and found that the contribution of current vertical shear to the measured difference between HF radar and the moored ADCPs was 2–3 cm s−1. To know how large the expected velocity vertical differences are on the WFS, we similarly examine the vertical shears of the near-surface current velocity using the ADCP data as an example. We calculate the rmsd of velocity at the topmost ADCP bin with the bins 1–4 m deeper (Fig. 15). Vertical shears are seen for all three moorings, with the rmsd value of 2.1, 2.8, and 3.6 cm s−1, respectively, for moorings C10, C12, and C13 within the top 3 m. These values are close to those found by Kohut et al. (2012). We note that these vertical shears may not represent those between the surface and the 4-m level or those between the HF radar effective measurement depths (2.4 and 0.9 m for the CODAR and WERA, respectively) and the ADCP topmost bins (4 m).

Fig. 15.
Fig. 15.

Vertical shears of the currents in the upper water column as shown in rmsd of the current velocities between the topmost bin and the bins 1–4 m deeper: (a) mooring C10, (b) mooring C12, and (c) mooring C13.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

Compared to HF radar measurements (with radial sector range resolutions of 5.8 and 0.6–1.5 km for CODAR SeaSonde and WERA, respectively), ADCP moorings can be considered as point observations. However, might the motion of the buoy within its watch circle induce an error in the ADCP velocity measurement? To examine this drift question, we analyze the GPS records of the three ADCP moorings. During the 3-month study period, the surface buoys stayed within their targeted watch circle determined by the mooring chain length of about twice the local water depths; that is, for mooring C12 at the 50-m isobath, the length of the mooring chain is about 100 m. We calculate the drift speeds of the surface buoys from the GPS records of ~4-h intervals, and interpolate them into hourly time series. The standard deviations of the drift speeds are less than 0.3 cm s−1. The histograms of the drift speeds show that over 90% of the drift speed values are lower than 0.2 cm s−1 (Fig. 16). These values are generally much smaller than the current velocities on the shelf. Thus, the corrections of the velocity with GPS records of the surface buoys are ignored in this study.

Fig. 16.
Fig. 16.

Histograms of the drift speeds of the surface buoys: (a) C10, (b) C12, and (c) C13. The mean speeds and their standard deviations (STD) are also shown in each panel.

Citation: Journal of Atmospheric and Oceanic Technology 31, 6; 10.1175/JTECH-D-13-00107.1

In summary, while the surface mooring drift speeds are relatively small, the expected spatial differences are large in comparison with the rmsd of HF radar–ADCP current measurements. A rough estimate of the total expected spatial differences range from 3.5 to 9.0 cm s−1, comparable with those of rmsd of the eight HF radar–ADCP station pairs (3.8–9.2 cm s−1). These results suggest that a substantial portion (80%–100%) of the observed differences between near-surface HF radar and ADCP velocity measurements may be explained in terms of the horizontal and vertical separations between sensors (Table 5).

6. Summary

Currently operated on the WFS and overlooking an array of moored ADCPs are two types of HF radars (three CODAR SeaSonde and two WERA systems). These coastal ocean observing system elements facilitated a performance evaluation of HF radar efficacy in mapping surface currents. The collocation of CODAR SeaSonde and WERA HF radars offered a unique opportunity to compare the performance of both types of HF radars against simultaneous velocity observations by ADCPs.

Based on data collected during the 3-month interval, October–December 2011, both CODAR SeaSonde and WERA demonstrated very good surface current mapping capabilities, both for radial velocity components and total velocity vectors. The rmsd values of CODAR SeaSonde–ADCP radials were in the range of 5.1–9.2 and 2.8–6.0 cm s−1 for hourly and 36-h low-pass filtered data, respectively, consistent with previous findings of CODAR SeaSonde performance on the WFS (Liu et al. 2010). These rmsd values generally fell in the lower end of the rmsd range for CODAR SeaSonde–ADCP comparisons (e.g., Emery et al. 2004; Paduan et al. 2006; Ohlmann et al. 2007; Cosoli et al. 2010). The rmsd values of WERA–ADCP radials were 3.8–6.5 and 2.2–4.3 cm s−1 for hourly and 36-h low-pass filtered time series, respectively. These values were comparable to those in the previous WERA–ADCP comparison on the WFS (Shay et al. 2007). They were also generally smaller than the rmsd values of WERA–ADCP velocities in other coastal regions (e.g., Shay et al. 1998; Parks et al. 2009; Robinson et al. 2011). Note that our experiment was performed on a low-energy shelf where the currents are generally weaker than most of those referenced herein. This might be one of the reasons why the rmsd values were generally smaller than those observed in earlier HF radar–ADCP comparisons.

In terms of radial currents obtained from CODAR SeaSonde and WERA at the same site origin (Venice, Florida), WERA showed slightly better comparison with the moored ADCP at location C10 than CODAR SeaSonde. The rmsd values of CODAR SeaSonde–ADCP versus WERA–ADCP were 5.6 versus 3.8 cm s−1 and 2.8 versus 2.2 cm s−1 for hourly and 36-h low-pass filtered radial current data, respectively. The bearing offset of CODAR SeaSonde was also larger than that of WERA (−14° vs −8°). Note that idealized antenna patterns had to be used for CODAR SeaSonde direction finding because the swapping out of system elements and the lack of funds precluded updating the antenna pattern within situ measurements. The recently proposed automatic calibrations for improved quality assurance for CODAR data were not implemented (Whelan et al. 2012). Further, WERA beam forming calibrated beyond the initial manufacturer’s setup and testing was also not performed (e.g., Flores-Vidal et al. 2013).

For total current velocities, CODAR SeaSonde exhibited a slightly better comparison with ADCP location C10 than did WERA. The complex correlation coefficient was 0.87 versus 0.85 (0.94 vs 0.92), and the average veering angle was −4° versus 6° (−5° vs 8°) for hourly (36-h low-pass filtered) CODAR SeaSonde–ADCP versus WERA–ADCP velocity vectors, respectively. The slightly poorer comparison for WERA was likely caused by the nonideal location of the C10 mooring (on the edge of the Fort De Soto WERA footprint). The large error in the radial currents for the Fort De Soto–C10 pair slightly increased the error of the total currents. However, overall, these total current comparison results were comparable to, or better than, previous comparisons for CODAR (e.g., Kaplan et al. 2005) and WERA (e.g., Parks et al. 2009; Savidge et al. 2011).

Despite differences in a variety of aspects between the direction-finding CODAR SeaSonde (long range, effective depth of 2.4 m, and integration time of 4 h) and the beam-forming WERA (median range, effective depth of 0.9 m, and integration time of 1 h), both types of HF radar systems demonstrated very good surface current mapping capability as gauged against the moored ADCP observations. The differences between the velocities measured with HF radar and ADCP were sufficiently small in this low-energy shelf that a substantial amount of the rmsd values may be accounted for by the measurement differences due to the horizontal, vertical, and temporal sampling between the ocean current sensors.

Finally, it is worth noting from the pragmatic perspective that robust performance was achieved with both systems. One year of continuous operation revealed both sites up and operational in excess of 97% of the time (Merz et al. 2012). Although the CODAR SeaSonde and WERA systems differ in design and complexity, once installed and operating, their day-to-day operation, maintenance requirements, and overall performance were found to be comparable. Ultimately, system selection depends on many criteria, including operational frequency selection and local radio frequency interference; measurement range and resolution considerations; initial and recurring operation and maintenance funding and system spares constraints; site selection, site spacing, and length of linear beach frontage available; environmental permitting and local construction requirements; installation design robustness; and IT and staffing considerations. An initial systemwide evaluation of such criteria is needed to ultimately determine a suitable system selection.

Acknowledgments

Support was provided by NOAA via the U.S. IOOS Office (Awards NA10NOS0120063 and NA11NOS0120141). The HF radar network was a contribution by the NOAA-funded Southeast Coastal Ocean Observing Regional Association (SECOORA). WERA equipment additions were through USF internal research and development funds. The success of the seagoing activity at USF is attributed to J. Law and J. Donovan, and D. Mayer assisted with real-time ADCP data acquisition and editing.

REFERENCES

  • Barrick, D. E., and Snider J. B. , 1977: The statistics of HF sea-echo Doppler spectra. IEEE Trans. Antennas Propag., 25, 1928, doi:10.1109/TAP.1977.1141529.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and Lipa B. J. , 1986: Correcting for distorted antenna patterns in CODAR ocean surface measurements. IEEE J. Oceanic Eng., 11, 304309, doi:10.1109/JOE.1986.1145158.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and Lipa B. J. , 1997: Evolution of bearing determination in HF current mapping radars. Oceanography, 10, 7275, doi:10.5670/oceanog.1997.27.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., Evans M. W. , and Weber B. L. , 1977: Ocean surface currents mapped by radar. Science,198, 138–144, doi:10.1126/science.198.4313.138.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., Lipa B. J. , and Crissman R. D. , 1985: Mapping surface currents with CODAR. Sea Technol.,26, 43–48.

  • Barth, A., Alvera-Azcárate A. , and Weisberg R. H. , 2008: Assimilation of high-frequency radar currents in a nested model of the West Florida shelf. J. Geophys. Res., 113, C08033, doi:10.1029/2007JC004585.

    • Search Google Scholar
    • Export Citation
  • Barth, A., Alvera-Azcárate A. , Gurgel K.-W. , Staneva J. , Port A. , Beckers J.-M. , and Stanev E. , 2010: Ensemble perturbation smoother for optimizing tidal boundary conditions by assimilation of high-frequency radar surface currents—Application to the German Bight. Ocean Sci., 6, 161178, doi:10.5194/os-6-161-2010.

    • Search Google Scholar
    • Export Citation
  • Chapman, R. D., and Graber H. C. , 1997: Validation of HF radar measurements. Oceanography, 10, 7679, doi:10.5670/oceanog.1997.28.

  • Chapman, R. D., Shay L. K. , Graber H. C. , Edson J. B. , Karachintsev A. , Trump C. L. , and Ross D. B. , 1997: On the accuracy of HF radar surface current measurements: Intercomparisons with ship-based sensors. J. Geophys. Res., 102, 18 73718 748, doi:10.1029/97JC00049.

    • Search Google Scholar
    • Export Citation
  • Chavanne, C. P., Flament P. , Luther D. S. , and Gurgel K.-W. , 2010: Observations of vortex Rossby waves associated with a mesoscale cyclone. J. Phys. Oceanogr., 40, 23332340, doi:10.1175/2010JPO4495.1.

    • Search Google Scholar
    • Export Citation
  • Cosoli, S., Mazzoldi A. , and Gačić M. , 2010: Validation of surface current measurements in the northern Adriatic Sea from high-frequency radars. J. Atmos. Oceanic Technol., 27, 908919, doi:10.1175/2009JTECHO680.1.

    • Search Google Scholar
    • Export Citation
  • Crombie, D. D., 1955: Doppler spectrum of sea echo at 13.56Mc./s. Nature, 175, 681682, doi:10.1038/175681a0.

  • De Paolo, T., and Terrill E. J. , 2007: Skill assessment of resolving ocean surface current structure using compact-antenna-style HF radar and the MUSIC direction-finding algorithm. J. Atmos. Oceanic Technol., 24, 12771300, doi:10.1175/JTECH2040.1.

    • Search Google Scholar
    • Export Citation
  • Emery, B. M., Washburn L. , and Harlan J. A. , 2004: Evaluating radial current measurements from CODAR high-frequency radars with moored current meters. J. Atmos. Oceanic Technol., 21, 12591271, doi:10.1175/1520-0426(2004)021<1259:ERCMFC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Emery, J. R., and Thompson R. E. , 2001: Data Analysis Methods in Physical Oceanography. 2nd ed. Elsevier, 654 pp.

  • Essen, H.-H., Gurgel K.-W. , and Schlick T. , 2000: On the accuracy of current measurements by means of HF radar. IEEE J. Oceanic Eng., 25, 472480, doi:10.1109/48.895354.

    • Search Google Scholar
    • Export Citation
  • Essen, H.-H., Breivik Ø. , Günther H. , Gurgel K.-W. , Johannessen J. , Klein H. , Schlick T. , and Stawarz M. , 2003: Comparison of remotely measured and modelled currents in coastal areas of Norway and Spain. J. Atmos. Ocean Sci., 9, 3864, doi:10.1080/1023673031000151412.

    • Search Google Scholar
    • Export Citation
  • Fernandez, D. M., and Paduan J. D. , 1996: Simultaneous CODAR and OSCR measurements of ocean surface currents in Monterey Bay. IGARSS ʼ96: 1996 International Geoscience and Remote Sensing Symposium; Remote Sensing for a Sustainable Future, Vol. 3, IEEE, 17491752, doi:10.1109/IGARSS.1996.516789.

  • Fernandez, D. M., Vesecky J. F. , and Teague C. C. , 2003: Calibration of HF radar systems with ships of opportunity. IGARSS ’03: Proceedings; 2003 IEEE International Geoscience and Remote Sensing Symposium, Vol. 7, IEEE, 42714273, doi:10.1109/IGARSS.2003.1295486.

  • Flores-Vidal, X., Flament P. , Durazo R. , Chavanne C. , and Gurgel K.-W. , 2013: High-frequency radars: Beamforming calibrations using ships as reflectors. J. Atmos. Oceanic Technol., 30, 638648, doi:10.1175/JTECH-D-12-00105.1.

    • Search Google Scholar
    • Export Citation
  • Graber, H. C., Haus B. K. , Chapman R. D. , and Shay L. K. , 1997: HF radar comparisons with moored estimates of current speed and direction: Expected differences and implications. J. Geophys. Res., 102, 18 74918 766, doi:10.1029/97JC01190.

    • Search Google Scholar
    • Export Citation
  • Gurgel, K.-W., Antonischki G. , and Schlick T. , 1997: A comparison of surface current fields derived by beam forming and direction finding techniques as applied by the HF radar WERA. IGARSSʼ97: 1997 International Geoscience and Remote Sensing Symposium; Remote Sensing—A Scientific Vision for Sustainable Development, T. I. Stein, Ed., Vol. 4, IEEE, 1805–1807, doi:10.1109/IGARSS.1997.609081.

  • Gurgel, K.-W., Antonischki G. , Essen H.-H. , and Schlick T. , 1999: Wellen Radar (WERA): A new ground-wave HF radar for ocean remote sensing. Coastal Eng., 37, 219234, doi:10.1016/S0378-3839(99)00027-7.

    • Search Google Scholar
    • Export Citation
  • Gurgel, K.-W., Schlick T. , Horstmann J. , and Maresca S. , 2010: Evaluation of an HF-radar ship detection and tracking algorithm by comparison to AIS and SAR data. 2010 International Waterside Security Conference (WSS), IEEE, 1–6, doi:10.1109/WSSC.2010.5730246.

  • Halle, C. M., and Largier J. L. , 2011: Surface circulation downstream of the Point Arena upwelling center. Cont. Shelf Res., 31, 12601272, doi:10.1016/j.csr.2011.04.007.

    • Search Google Scholar
    • Export Citation
  • Harlan, J., Terrill E. , Hazard L. , Keen C. , Barrick D. , Whelan C. , Howden S. , and Kohut J. , 2010: The Integrated Ocean Observing System high-frequency radar network: Status and local, regional, and national applications. Mar. Technol. Soc. J., 44, 122132, doi:10.4031/MTSJ.44.6.6.

    • Search Google Scholar
    • Export Citation
  • He, R., and Weisberg R. H. , 2002: Tides on the West Florida shelf. J. Phys. Oceanogr., 32, 34553473, doi:10.1175/1520-0485(2002)032<3455:TOTWFS>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Helzel, T., 2007: FMCW radar “WERA.” Helzel Messtechnik GmbH Internal Doc. 10/19/2007, 6 pp.

  • Helzel, T., and Kniephoff M. , 2010: Software beam forming for ocean radar: WERA features and accuracy. Oceans 2010, IEEE, 1–3, doi:10.1109/OCEANS.2010.5664374.

  • Helzel, T., Kniephoff M. , and Peterson L. , 2006: WERA: Remote ocean sensing for current, wave, and wind direction: Introduction to the principle of operation. Proc. US/EU-Baltic Int. Symp., Klaipeda, Lithuania, IEEE. [Available online at http://www.helzel.com/files/432/upload/Papers/WERA-Basics-Paper-6605.pdf.]

  • Helzel, T., Kniephoff M. , Peterson L. , Mariette V. , and Thomas N. , 2010: Accuracy and reliability of ocean radar WERA in beam forming or direction finding mode. 2011 IEEE/OES Tenth Current, Waves and Turbulence Measurements (CWTM), J. Rizoli White and A. J. Williams III, IEEE, 2124, doi:10.1109/CWTM.2011.5759518.

  • Heron, M. L., and Atwater D. P. , 2013: Temporal and spatial resolution of HF ocean radars. Ocean Sci. J., 48, 99103, doi:10.1007/s12601-013-0008-z.

    • Search Google Scholar
    • Export Citation
  • Holbrook, J. R., and Frisch A. S. , 1981: A comparison of near-surface CODAR and VACM measurements in the Strait of Juan de Fuca, August 1978. J. Geophys. Res., 86, 10 90810 912, doi:10.1029/JC086iC11p10908.

    • Search Google Scholar
    • Export Citation
  • Kaplan, D. M., Largier J. , and Botsford L. W. , 2005: HF radar observations of surface circulation off Bodega Bay (northern California, USA). J. Geophys. Res., 110, C10020, doi:10.1029/2005JC002959.

    • Search Google Scholar
    • Export Citation
  • Kim, S. Y., and Coauthors, 2011: Mapping the U.S. West Coast surface circulation: A multiyear analysis of high-frequency radar observation. J. Geophys. Res., 116, C03011, doi:10.1029/2010JC006669.

    • Search Google Scholar
    • Export Citation
  • Kirincich, A. R., de Paolo T. , and Terrill E. , 2012: Improving HF radar estimates of surface currents using signal quality metrics, with application to the MVCO high-resolution radar system. J. Atmos. Oceanic Technol., 29, 13771390, doi:10.1175/JTECH-D-11-00160.1.

    • Search Google Scholar
    • Export Citation
  • Kohut, J. T., and Glenn S. M. , 2003: Calibration of HF radar surface current measurements using measured antenna beam patterns. J. Atmos. Oceanic Technol., 20, 13031316, doi:10.1175/1520-0426(2003)020<1303:IHRSCM>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kohut, J. T., Glenn S. M. , and Chant R. J. , 2004: Seasonal current variability on the New Jersey inner shelf. J. Geophys. Res., 109, C07S07, doi:10.1029/2003JC001963.

    • Search Google Scholar
    • Export Citation
  • Kohut, J. T., Roarty H. , Randall-Goodwin E. , Glenn S. , and Lichtenwalner C. S. , 2012: Evaluation of two algorithms for a network of coastal HF radars in the Mid-Atlantic Bight. Ocean Dyn., 62, 953968, doi:10.1007/s10236-012-0533-9.

    • Search Google Scholar
    • Export Citation
  • Kosro, P. M., Barth J. A. , and Strub T. P. , 1997: The coastal jet: Observations of surface currents over the Oregon continental shelf from HF radar. Oceanography, 10, 5357, doi:10.5670/oceanog.1997.22.

    • Search Google Scholar
    • Export Citation
  • Kundu, P. K., 1976: Ekman veering observed near the ocean bottom. J. Phys. Oceanogr., 6, 238242, doi:10.1175/1520-0485(1976)006<0238:EVONTO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Laws, K. E., Fernandez D. M. , and Paduan J. D. , 2000: Simulation-based evaluations of HF radar ocean current algorithms. IEEE J. Oceanic Eng., 25, 481491, doi:10.1109/48.895355.

    • Search Google Scholar
    • Export Citation
  • Lipa, B. J., and Barrick D. E. , 1983: Least-squares methods for the extraction of surface currents from CODAR crossed-loop data: Application at ARSLOE. IEEE J. Oceanic Eng.,8, 226–253, doi:10.1109/JOE.1983.1145578.

  • Lipa, B. J., and Barrick D. E. , 1986: Extraction of sea state from HF radar sea echo: Mathematical theory and modeling. Radio Sci., 21, 81100, doi:10.1029/RS021i001p00081.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., and Weisberg R. H. , 2005a: Momentum balance diagnoses on the West Florida shelf. Cont. Shelf Res., 25, 20542074, doi:10.1016/j.csr.2005.03.004.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., and Weisberg R. H. , 2005b: Patterns of ocean current variability on the West Florida shelf using the self-organizing map. J. Geophys. Res., 110, C06003, doi:10.1029/2004JC002786.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., and Weisberg R. H. , 2007: Ocean currents and sea surface heights estimated across the West Florida shelf. J. Phys. Oceanogr., 37, 16971713, doi:10.1175/JPO3083.1.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., and Weisberg R. H. , 2012: Seasonal variability on the West Florida shelf. Prog. Oceanogr., 104, 8098, doi:10.1016/j.pocean.2012.06.001.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., Weisberg R. H. , and Shay L. K. , 2007: Current patterns on the West Florida shelf from joint self-organizing map analyses of HF radar and ADCP data. J. Atmos. Oceanic Technol., 24, 702712, doi:10.1175/JTECH1999.1.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., Weisberg R. H. , Merz C. R. , Lichtenwalner S. , and Kirkpatrick G. J. , 2010: HF radar performance in a low-energy environment: CODAR SeaSonde experience on the West Florida shelf. J. Atmos. Oceanic Technol., 27, 16891710, doi:10.1175/2010JTECHO720.1.

    • Search Google Scholar
    • Export Citation
  • Liu, Y., MacFadyen A. , Ji Z. G. , and Weisberg R. H. , 2011: Monitoring and Modeling the Deepwater Horizon Oil Spill: A Record-Breaking Enterprise. Geophys. Monogr., Vol. 195, Amer. Geophys. Union, 271 pp., doi:10.1029/GM195.

  • Liu, Y., Weisberg R. H. , Vignudelli S. , Roblou L. , and Merz C. , 2012: Comparison of the X-TRACK altimetry estimated currents with moored ADCP and HF radar observations on the West Florida shelf. Adv. Space Res., 50, 10851098, doi:10.1016/j.asr.2011.09.012.

    • Search Google Scholar
    • Export Citation
  • Merz, C. R., 2001: An overview of the Coastal Ocean Monitoring and Prediction System (COMPS). MTS/IEEE Oceans 2001: An Ocean Odyssey, Vol. 2, MTS and IEEE, 1183–1187, doi:10.1109/OCEANS.2001.968281.

  • Merz, C. R., Weisberg R. H. , and Liu Y. , 2012: Evolution of the USF/CMS CODAR and WERA HF radar network. Oceans 2012, IEEE, 1–5, doi:10.1109/OCEANS.2012.6404947.

  • Mooers, C. N. K., 1973: A technique for the cross spectrum analysis of pairs of complex-valued time series, with emphasis on properties of polarized components and rotational invariants. Deep-Sea Res., 20, 11291141, doi:10.1016/0011-7471(73)90027-2.

    • Search Google Scholar
    • Export Citation
  • Ohlmann, C., White P. , Washburn L. , Terrill E. , Emery B. , and Otero M. , 2007: Interpretation of coastal HF radar-derived surface currents with high-resolution drifter data. J. Atmos. Oceanic Technol., 24, 666680, doi:10.1175/JTECH1998.1.

    • Search Google Scholar
    • Export Citation
  • Paduan, J. D., and Rosenfeld L. K. , 1996: Remotely sensed surface currents in Monterey Bay from shore-based HF radar (Coastal Ocean Dynamics Application Radar). J. Geophys. Res., 101, 20 66920 686, doi:10.1029/96JC01663.

    • Search Google Scholar
    • Export Citation
  • Paduan, J. D., and Graber H. C. , 1997: Introduction to high-frequency radar: Reality and myth. Oceanography, 10, 3639, doi:10.5670/oceanog.1997.18.

    • Search Google Scholar
    • Export Citation
  • Paduan, J. D., and Washburn L. , 2013: High-frequency radar observations of ocean surface currents. Annu. Rev. Mar. Sci., 5, 115136, doi:10.1146/annurev-marine-121211-172315.

    • Search Google Scholar
    • Export Citation
  • Paduan, J. D., Kosro P. M. , and Glenn S. M. , 2004: A national coastal ocean surface current mapping system for the United States. Mar. Technol. Soc. J., 38, 102108, doi:10.4031/002533204787522839.

    • Search Google Scholar
    • Export Citation
  • Paduan, J. D., Kim K. C. , Cook M. S. , and Chavez F. P. , 2006: Calibration and validation of direction-finding high-frequency radar ocean surface current observations. IEEE J. Oceanic Eng., 31, 862875, doi:10.1109/JOE.2006.886195.

    • Search Google Scholar
    • Export Citation
  • Parks, A. B., Shay L. K. , Johns W. E. , Martinez-Pedraja J. , and Gurgel K.-W. , 2009: HF radar observations of small-scale surface current variability in the Straits of Florida. J. Geophys. Res., 114, C08002, doi:10.1029/2008JC005025.

    • Search Google Scholar
    • Export Citation
  • Pawlowicz, R., Beardsley B. , and Lentz S. , 2002: Classical tidal harmonic analysis including error estimates in MATLAB using T-TIDE. Comput. Geosci., 28, 929937, doi:10.1016/S0098-3004(02)00013-4.

    • Search Google Scholar
    • Export Citation
  • Roarty, H. J., and Coauthors, 2010: Operation and application of a regional high-frequency radar network in the Mid-Atlantic Bight. Mar. Technol. Soc. J., 44, 133145, doi:10.4031/MTSJ.44.6.5.

    • Search Google Scholar
    • Export Citation
  • Robinson, A. M., Wyatt L. R. , and Howarth M. J. , 2011: A two year comparison between HF radar and ADCP current measurements in Liverpool Bay. J. Oper. Oceanogr., 4, 3345.

    • Search Google Scholar
    • Export Citation
  • Savidge, D., Amft J. , Gargett A. , Archer M. , Conley D. , Voulgaris G. , Wyatt L. , and Gurgel K.-W. , 2011: Assessment of WERA long-range HF-radar performance from the user’s perspective. 2011 IEEE/OES Tenth Current, Waves and Turbulence Measurements (CWTM), J. Rizoli White and A. J. Williams III, IEEE, 3138, doi:10.1109/CWTM.2011.5759520.

  • Schmidt, R. O., 1986: Multiple emitter location and signal parameter estimation. IEEE Trans. Antennas Propag., 34, 276280, doi:10.1109/TAP.1986.1143830.

    • Search Google Scholar
    • Export Citation
  • Seim, H. E., Fletcher M. , Mooers C. N. K. , Nelson J. R. , and Weisberg R. H. , 2009: Towards a regional coastal ocean observing system: An initial design for the Southeast Coastal Ocean Observing Regional Association. J. Mar. Syst., 77, 261277, doi:10.1016/j.jmarsys.2007.12.016.

    • Search Google Scholar
    • Export Citation
  • Shay, L. K., Graber H. C. , Ross D. B. , and Chapman R. D. , 1995: Mesoscale surface current structure detected by HF radar. J. Atmos. Oceanic Technol., 12, 881900, doi:10.1175/1520-0426(1995)012<0881:MOSCSD>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Shay, L. K., Lentz S. J. , Graber H. C. , and Haus B. K. , 1998: Current structure variations detected by HF radar and vector measuring current meters. J. Atmos. Oceanic Technol., 15, 237256, doi:10.1175/1520-0426(1998)015<0237:CSVDBH>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Shay, L. K., Cook T. M. , Peters H. , Mariano A. J. , An P. E. , Soloviev A. , Weisberg R. H. , and Luther M. E. , 2002: Very high-frequency radar mapping of surface currents. IEEE J. Oceanic Eng., 27, 155169, doi:10.1109/JOE.2002.1002470.

    • Search Google Scholar
    • Export Citation
  • Shay, L. K., Martinez J. , Cook T. M. , Haus B. K. , and Weisberg R. H. , 2007: High-frequency surface current mapping using WERA. J. Atmos. Oceanic Technol., 24, 484503, doi:10.1175/JTECH1985.1.

    • Search Google Scholar
    • Export Citation
  • Stewart, R. H., and Joy J. W. , 1974: HF radio measurements of surface currents. Deep-Sea Res., 21, 10391049, doi:10.1016/0011-7471(74)90066-7

    • Search Google Scholar
    • Export Citation
  • Teague, C. C., Fernandez D. M. , Laws K. E. , Paduan J. D. , and Vesecky J. F. , 1998: Comparison of multifrequency phased-array and direction-finding HF radar systems during COPE-3. IGARSS ʼ98: 1998 IEEE International Geoscience and Remote Sensing Symposium Proceedings, T. I. Stein, Ed., Vol. 1, IEEE, 201203, doi:10.1109/IGARSS.1998.702852.

  • Teague, C. C., Vesecky J. F. , and Hallock Z. R. , 2001: A comparison of multifrequency HF radar and ADCP measurements of near-surface currents during COPE-3. IEEE J. Oceanic Eng., 26, 399405, doi:10.1109/48.946513.

    • Search Google Scholar
    • Export Citation
  • Ullman, D. S., Codiga D. L. , Hebert D. , Decker L. B. , and Kincaid C. R. , 2012: Structure and dynamics of the midshelf front in the New York Bight. J. Geophys. Res., 117, C01012, doi:10.1029/2011JC007553.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., 2011: Coastal ocean pollution, water quality and ecology: A commentary. Mar. Technol. Soc. J., 45, 3542, doi:10.4031/MTSJ.45.2.1.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., and He R. , 2003: Local and deep-ocean forcing contributions to anomalous water properties on the West Florida shelf. J. Geophys. Res., 108, 3184, doi:10.1029/2002JC001407.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., Black B. D. , and Yang H. , 1996: Seasonal modulation of the west Florida continental shelf circulation. Geophys. Res. Lett., 23, 22472250, doi:10.1029/96GL02184.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., He R. , Liu Y. , and Virmani J. , 2005: West Florida shelf circulation on synoptic, seasonal, and inter-annual time scales. Circulation in the Gulf of Mexico: Observations and Models, Geophys. Monogr., Vol. 161, Amer. Geophys. Union, 325–347, doi:10.1029/161GM23.

  • Weisberg, R. H., Barth A. , Alvera-Azcárate A. , and Zheng L. , 2009a: A coordinated coastal ocean observing and modeling system for the West Florida shelf. Harmful Algae, 8, 585598, doi:10.1016/j.hal.2008.11.003.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., Liu Y. , and Mayer D. A. , 2009b: West Florida shelf mean circulation observed with long-term moorings. Geophys. Res. Lett., 36, L19610, doi:10.1029/2009GL040028.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., Liu Y. , Merz C. R. , Virmani J. I. , and Zheng L. , 2012: A critique of alternative power generation for Florida by mechanical and solar means. Mar. Technol. Soc. J., 46, 1223, doi:10.4031/MTSJ.46.5.1.

    • Search Google Scholar
    • Export Citation
  • Weisberg, R. H., Zheng L. , Liu Y. , Murawski S. , Hu C. , and Paul J. , 2014: Did Deepwater Horizon hydrocarbons transit to the West Florida continental shelf? Deep-Sea Res. II, doi:10.1016/j.dsr2.2014.02.002, in press.

    • Search Google Scholar
    • Export Citation
  • Whelan, C., Emery B. , Teague C. , Barrick D. , Washburn L. , and Harlan J. , 2012: Automatic calibrations for improved quality assurance of coastal HF radar currents. Oceans 2012, IEEE, 1–4, doi:10.1109/OCEANS.2012.6405104.

  • Zaron, E., Chavanne C. , Egbert G. , and Flament P. , 2009: Baroclinic tidal generation in the Kauai Channel inferred from high-frequency radar Doppler currents. Dyn. Atmos. Oceans, 48, 93120, doi:10.1016/j.dynatmoce.2009.03.002.

    • Search Google Scholar
    • Export Citation
  • Zheng, L., and Weisberg R. H. , 2012: Modeling the West Florida coastal ocean by downscaling from the deep ocean, across the continental shelf and into the estuaries. Ocean Modell., 48, 1029, doi:10.1016/j.ocemod.2012.02.002.

    • Search Google Scholar
    • Export Citation
Save
  • Barrick, D. E., and Snider J. B. , 1977: The statistics of HF sea-echo Doppler spectra. IEEE Trans. Antennas Propag., 25, 1928, doi:10.1109/TAP.1977.1141529.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and Lipa B. J. , 1986: Correcting for distorted antenna patterns in CODAR ocean surface measurements. IEEE J. Oceanic Eng., 11, 304309, doi:10.1109/JOE.1986.1145158.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and Lipa B. J. , 1997: Evolution of bearing determination in HF current mapping radars. Oceanography, 10, 7275, doi:10.5670/oceanog.1997.27.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., Evans M. W. , and Weber B. L. , 1977: Ocean surface currents mapped by radar. Science,198, 138–144, doi:10.1126/science.198.4313.138.

    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., Lipa B. J. , and Crissman R. D. , 1985: Mapping surface currents with CODAR. Sea Technol.,26, 43–48.

  • Barth, A., Alvera-Azcárate A. , and Weisberg R. H. , 2008: Assimilation of high-frequency radar currents in a nested model of the West Florida shelf. J. Geophys. Res., 113, C08033, doi:10.1029/2007JC004585.

    • Search Google Scholar
    • Export Citation
  • Barth, A., Alvera-Azcárate A. , Gurgel K.-W. , Staneva J. , Port A. , Beckers J.-M. , and Stanev E. , 2010: Ensemble perturbation smoother for optimizing tidal boundary conditions by assimilation of high-frequency radar surface currents—Application to the German Bight. Ocean Sci., 6, 161178, doi:10.5194/os-6-161-2010.

    • Search Google Scholar
    • Export Citation
  • Chapman, R. D., and Graber H. C. , 1997: Validation of HF radar measurements. Oceanography, 10, 7679, doi:10.5670/oceanog.1997.28.

  • Chapman, R. D., Shay L. K. , Graber H. C. , Edson J. B. , Karachintsev A. , Trump C. L. , and Ross D. B. , 1997: On the accuracy of HF radar surface current measurements: Intercomparisons with ship-based sensors. J. Geophys. Res., 102, 18 73718 748, doi:10.1029/97JC00049.

    • Search Google Scholar
    • Export Citation
  • Chavanne, C. P., Flament P. , Luther D. S. , and Gurgel K.-W. , 2010: Observations of vortex Rossby waves associated with a mesoscale cyclone. J. Phys. Oceanogr., 40, 23332340, doi:10.1175/2010JPO4495.1.

    • Search Google Scholar
    • Export Citation
  • Cosoli, S., Mazzoldi A. , and Gačić M. , 2010: Validation of surface current measurements in the northern Adriatic Sea from high-frequency radars. J. Atmos. Oceanic Technol., 27, 908919, doi:10.1175/2009JTECHO680.1.

    • Search Google Scholar
    • Export Citation
  • Crombie, D. D., 1955: Doppler spectrum of sea echo at 13.56Mc./s. Nature, 175, 681682, doi:10.1038/175681a0.

  • De Paolo, T., and Terrill E. J. , 2007: Skill assessment of resolving ocean surface current structure using compact-antenna-style HF radar and the MUSIC direction-finding algorithm. J. Atmos. Oceanic Technol., 24, 12771300, doi:10.1175/JTECH2040.1.

    • Search Google Scholar
    • Export Citation
  • Emery, B. M., Washburn L. , and Harlan J. A. , 2004: Evaluating radial current measurements from CODAR high-frequency radars with moored current meters. J. Atmos. Oceanic Technol., 21, 12591271, doi:10.1175/1520-0426(2004)021<1259:ERCMFC>2.0.CO;2.