• Balsley, B. B., A. C. Riddle, W. L. Ecklund, and D. A. Carter, 1987: Sea surface currents in the equatorial Pacific from VHF radar backscatter observations. J. Atmos. Oceanic Technol., 4, 530535, https://doi.org/10.1175/1520-0426(1987)004<0530:SSCITE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., 1971: Theory of HF and VHF propagation across the rough sea, 2, application to HF and VHF propagation above the sea. Radio Sci., 6, 527533, https://doi.org/10.1029/RS006i005p00527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., 1972: First-order theory and analysis of MF/HF/VHF scatter from the sea. IEEE Trans. Antennas Propag., 20, 210, https://doi.org/10.1109/TAP.1972.1140123.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and B. J. Lipa, 1997: Evolution of bearing determination in HF current mapping radars. Oceanography, 10 (2), 7275, https://doi.org/10.5670/oceanog.1997.27.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bass, F. G., I. M. Fuks, A. I. Kalmycov, I. E. Ostrovsky, and A. D. Rosenberg, 1968a: Very high frequency radiowave scattering by a disturbed sea surface—Part I: Scattering from a slightly disturbed boundary. IEEE Trans. Antennas Propag., 16, 554559, https://doi.org/10.1109/TAP.1968.1139243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bass, F. G., I. M. Fuks, A. I. Kalmycov, I. E. Ostrovsky, and A. D. Rosenberg, 1968b: Very high frequency radiowave scattering by a disturbed sea surface—Part II: Scattering from an actual sea surface. IRE Trans. Antennas Propag., 16, 560568, https://doi.org/10.1109/TAP.1968.1139244.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Capon, J., 1969: High-resolution frequency-wavenumber spectrum analysis. Proc. IEEE, 57, 14081419, https://doi.org/10.1109/PROC.1969.7278.

  • Chen, J.-S., P. Hoffmann, M. Zecha, and C.-H. Hsieh, 2008: Coherent radar imaging of mesosphere summer echoes: Influence of radar beam pattern and tilted structures on atmospheric echo center. Radio Sci., 43, RS1002, https://doi.org/10.1029/2006RS003593.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, J.-S., C.-L. Su, Y.-H. Chu, G. Hassenpflug, and M. Zecha, 2009: Extended application of a novel phase calibration method of multiple-frequency range imaging to the Chung-Li and MU VHF radars. J. Atmos. Oceanic Technol., 26, 24882500, https://doi.org/10.1175/2009JTECHA1295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, J.-S., J. Furumoto, and M. Yamamoto, 2014: Three-dimensional radar imaging of atmospheric layer and turbulence structures using multiple receivers and multiple frequencies. Ann. Geophys., 32, 899909, https://doi.org/10.5194/angeo-32-899-2014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, M.-Y., T.-Y. Yu, Y.-H. Chu, W. O. J. Brown, and S. A. Cohn, 2007: Application of Capon technique to mitigate bird contamination on a spaced antenna wind profiler. Radio Sci., 42, RS6005, https://doi.org/10.1029/2006RS003604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chilson, P. B., T.-Y. Yu, R. G. Strauch, A. Muschinski, and R. D. Palmer, 2003: Implementation and validation of range imaging on a UHF radar wind profiler. J. Atmos. Oceanic Technol., 20, 987996, https://doi.org/10.1175/1520-0426(2003)20<987:IAVORI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chu, Y.-H., and C.-L. Su, 2015: Sea surface radar echoes collected by the Chung-Li VHF radar from 2015-10-28 to 2015-11-13. Chung-Li VHF Radar Station.

  • Cochin, V., P. Forget, B. Seille, and G. Mercier, 2005: Sea surface currents and wind direction by VHF radar: Results and validation. Europe Oceans 2005, Brest, France, Institute of Electrical and Electronics Engineers, 8573991, https://doi.org/10.1109/OCEANSE.2005.1513183.

    • Crossref
    • Export Citation
  • Cochin, V., V. Mariette, P. Broche, and R. Garello, 2006: Tidal current measurements using VHF radar and ADCP in the Normand Breton Gulf: Comparison of observations and numerical model. IEEE J. Oceanic Eng., 31, 885893, https://doi.org/10.1109/JOE.2006.886077.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Egbert, G. D., and S. Y. Erofeeva, 2002: Efficient inverse modeling of barotropic ocean tides. J. Atmos. Oceanic Technol., 19, 183204, https://doi.org/10.1175/1520-0426(2002)019<0183:EIMOBO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Forget, P., J. L. Devenon, J. C. De Maistre, M. Leveau, and P. Broche, 1990: VHF remote sensing for mapping river plume circulation. Geophys. Res. Lett., 17, 10971100, https://doi.org/10.1029/GL017i008p01097.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fukao, S., H. Luce, T. Mega, and M. K. Yamamoto, 2011: Extensive studies of large-amplitude Kelvin–Helmholtz billows in the lower atmosphere with VHF middle and upper atmosphere radar. Quart. J. Roy. Meteor. Soc., 137, 10191041, https://doi.org/10.1002/qj.807.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heron, M., and R. J. Rose, 1986: On the application of HF ocean radar to the observation of temporal and spatial changes in wind direction. IEEE J. Oceanic Eng., 11, 210218, https://doi.org/10.1109/JOE.1986.1145173.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jan, S., Y.-H. Wang, D.-P. Wang, and S.-Y. Chao, 2004: Incremental inference of boundary forcing for a three-dimensional tidal model: Tides in the Taiwan Strait. Cont. Shelf Res., 24, 337351, https://doi.org/10.1016/j.csr.2003.11.005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumar, S., T. N. Rao, V. K. Anandan, V. N. Sureshbabu, and T. Arunamani, 2016: Comparison of horizontal winds and turbulent eddy dissipation rates obtained by two different multi-receiver techniques—Spaced antenna and postset beam steering. J. Atmos. Sol.-Terr. Phys., 137, 4452, https://doi.org/10.1016/j.jastp.2015.11.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luce, H., G. Hassenpflug, M. Yamamoto, S. Fukao, and K. Sato, 2008: High-resolution observations with MU radar of a KH instability triggered by an inertia–gravity wave in the upper part of a jet stream. J. Atmos. Sci., 65, 17111718, https://doi.org/10.1175/2007JAS2346.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palmer, R. D., S. Gopalam, T.-Y. Yu, and S. Fukao, 1998: Coherent radar imaging using Capon’s method. Radio Sci., 33, 15851598, https://doi.org/10.1029/98RS02200.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Prandle, D., 1987: The fine-structure of nearshore tidal and residual circulations revealed by H.F. radar surface current measurements. J. Phys. Oceanogr., 17, 231245, https://doi.org/10.1175/1520-0485(1987)017<0231:TFSONT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saito, S., M. Yamamoto, H. Hashiguchi, and A. Maegawa, 2006: Observation of three-dimensional structures of quasi-periodic echoes associated with mid-latitude sporadic-E layers by MU radar ultra-multi-channel system. Geophys. Res. Lett., 33, L14109, https://doi.org/10.1029/2005GL025526.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shay, L. K., and Coauthors, 2000: VHF radar detects oceanic submesoscale vortex along Florida coast. Eos, Trans. Amer. Geophys. Union, 81, 209213, https://doi.org/10.1029/00EO00143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shrira, V. I., D. V. Ivonin, P. Broche, and J. C. de Maistre, 2001: On remote sensing of vertical shear of ocean surface currents by means of a single-frequency VHF radar. Geophys. Res. Lett., 28, 39553958, https://doi.org/10.1029/2001GL013387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Teague, C. C., J. F. Vesecky, and D. M. Fernandez, 1997: HF radar instruments, past to present. Oceanography, 10 (2), 4044, https://doi.org/10.5670/oceanog.1997.19.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, C.-Y., C.-L. Su, K.-H. Wu, and Y.-H. Chu, 2017: Cross spectral analysis of CODAR-SeaSonde echoes from sea surface and ionosphere at Taiwan. Int. J. Antennas Propag., 2017, 1756761, https://doi.org/10.1155/2017/1756761.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, T.-Y., R. D. Palmer, and D. L. Hysell, 2000: A simulation study of coherent radar imaging. Radio Sci., 35, 11291141, https://doi.org/10.1029/1999RS002236.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, T.-Y., J. Furumoto, and M. Yamamoto, 2010: Clutter suppression for high-resolution atmospheric observations using multiple receivers and multiple frequencies. Radio Sci., 45, RS4011, https://doi.org/10.1029/2009RS004330.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) VHF radar location (24°57′58″N, 121°00′30″E) and beam direction. The background chart is from Google Maps. (b) Arrangement of VHF radar antennas. (c) Cartesian coordinate system referring to (b). The z axis represents the radar beam direction; x and y directions are, respectively, horizontal and upward; and k denotes the wavenumber vector for beamforming.

  • View in gallery

    (a) Comparison of Capon and classical beamforming results for two situations of sea echoes received by the VHF radar. (b) Capon brightness distribution extending along the vertical dimension for locating the echo center. The brightness value is self-normalized, and the cross symbol indicates the located echo center.

  • View in gallery

    Range–time intensities of radar sea echoes observed by (a)–(c) 52-MHz VHF radar and (d) 4.58-MHz CODAR. The distance between the two radar sites is about 8 km.

  • View in gallery

    Typical spectra of sea echoes observed by 52-MHz VHF radar.

  • View in gallery

    Radial velocities observed by 4.58-MHz CODAR and 52-MHz VHF radars. (a) Positive and (b) negative Doppler components. The first-order backscatter component has been removed. Data time: 10–12 Nov 2015.

  • View in gallery

    (a) Brightness distribution (self-normalized) and echo center (dots) retrieved by the Capon method. (b) Histograms of echo center locations for three time periods: 28–30 Oct and 3–6 and 10–13 Nov 2015, respectively.

  • View in gallery

    (a) Radar echo intensity (bold line) and the seal levels measured at Zhuwei (solid thin line) and Shinchu (dotted line) harbors. The height range of sea level is between −250 and 250 cm, and echo intensity is present between 44 and 55 dB for the range cell between 17.3 and 17.6 km. The slanted long dotted lines indicate approximately the peaks of the oscillatory curves. (b) As in (a), but the simulated sea levels at the ranges of 15 and 20 km are shown (solid and dotted lines). (c) Simulated tidal current directions: rightward and upward correspond to eastward and northward, respectively. (d) Comparisons of temporal variations between echo intensity, echo center, and radial velocity.

  • View in gallery

    The 52-MHz VHF radar observations. (a) Difference between the sea surface wind-related backscatter frequency fBr and the theoretical value fB. Data time: 10–13 Nov 2015. (b) Radial velocity estimated by combining the positive and negative first-order Doppler means. The right two panels show the histograms of the values in the left panels, but the data time is 10–12 Nov 2015. The data with SNR < 0.25, appearing mostly at the range farther than 20-something km, have been discarded in the histogram.

  • View in gallery

    (a) Schematic plot of variation in sea level within the radar beam and (b),(c) temporal variations in echo intensity for the range cell between 17.3 and 17.6 km, during 28 and 30 Oct and 10 and 13 Nov 2015.

  • View in gallery

    (a) (top row) Spectral analysis of echo intensity, echo center, and radial velocity in the range cell between 17.9 and 18.2 km. (middle row) Histograms of spectral peaks in the range interval between 0.5 and 39.5 km. (bottom row) Spectral contributions of various tidal modes. (b) As in the middle and bottom rows in (a), but for another period.

  • View in gallery

    Bathymetric chart around the sea area observed by the VHF radar.

  • View in gallery

    (a) Spectral widths estimated from (top) positive and (bottom) negative first-order Doppler spectral lines. The curves overlaying on the color map are the 1-h mean echo intensity of several selected range cells. (b) Temporal variations in spectral width (black), estimated from positive first-order spectral lines, and echo intensity (red). Blue line connects the peak locations of oscillatory curves of the echo intensity between adjacent range cells. Data time: 10–13 Nov 2015.

  • View in gallery

    (a) Radial velocity estimated by combining the positive and negative first-order Doppler means. The curves overlaying on the color map are the 1-h mean echo intensity of several selected range cells. (b) (left) Histogram of correlation coefficient between radial velocity and echo intensity. Calculation of correlation coefficient is made with 1-h mean data series. (right) Leading time of radial velocity over echo intensity. The 3-h leading or delay represents a phase difference of 90° in the wavy curves. Data time of (a) and (b) is between 10 and 13 Nov 2015. (c),(d) As in (b), but present for the data periods of 3–6 Nov and 28–30 Oct 2015, respectively.

  • View in gallery

    (a)–(c) As in Figs. 13b–d, but for the relationship between echo center (absolute value) and echo intensity. (d) (left) The 1-h mean data series of echo center (black) and echo intensity (gray). The numbers shown on the right edge of the panel are the boundaries of the curves in the range cells; the first and second columns are for the black and gray curves, respectively. (right) Temporal variation in correlation coefficient, calculated from the data series shown in the left panel.

  • View in gallery

    (left) The echo center (solid circle) associating with altitudinal variation of a slanted sea level (dashed lines). The gray box indicates the fan-shaped radar beam pattern radiated by the vertical four-element transmitting antenna. (right) The echo center associating with altitudinal variation of a horizontal sea level in a slanted radar beam pattern.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 744 534 27
PDF Downloads 683 497 33

VHF Radar Observations of Sea Surface in the Northern Taiwan Strait

View More View Less
  • 1 Center for General Education, China Medical University, Taichung, Taiwan
  • | 2 Taiwan Ocean Research Institute, National Applied Research Laboratories, Kaohsiung, Taiwan
  • | 3 Graduate Institute of Hydrological and Oceanic Sciences, National Central University, Taoyuan, Taiwan
  • | 4 Department of Optoelectric Physics, Chinese Culture University, Taipei, Taiwan
  • | 5 Graduate Institute of Space Science, National Central University, Taoyuan, Taiwan
  • | 6 Taiwan Typhoon and Flood Research Institute, National Applied Research Laboratories, Taipei, Taiwan
© Get Permissions
Open access

Abstract

A VHF pulsed radar system was set up on the Taoyuan County seashore (24°57′58″N, 121°00′30″E; Taiwan) to observe the sea surface in the northern Taiwan Strait for the first time. The radar used a four-element, vertically polarized Yagi antenna to transmit the 52-MHz radar wave. The receiving linear array consists of four vertical dipole antennas that were located 3 m apart and attached with four independent and identical receivers. With the multichannel echoes, the direction of arrival (DOA) of the radar echoes were determined by using an optimization beamforming approach—the Capon method. Echo intensity was observed to vary principally in semidiurnal oscillation, which matched well the time series of tide gauge measurements and sea level simulations. In addition, the oscillatory characteristics of Doppler/radial velocity of the VHF radar were generally consistent with that of the HF coastal ocean dynamics applications radar (CODAR) nearby. Nevertheless, the contributions of various tidal modes to the parameters of DOA, echo intensity, radial velocity, and spectral width, varied with the range and time period (e.g., neap or spring tides). For example, the semidiurnal tides governed the variation in the echo center only in the range interval between ~15 and ~25 km from the seashore but dominated other parameters throughout the detectable range. Correlations and phase relationships between these parameters were diverse; they varied with time and had dramatic changes at around the distances of 3 and 10 km. Possible causes of these features were discussed, including sea surface wind, nearshore current, sea level height, and bathymetric effect.

Denotes content that is immediately available upon publication as open access.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Yen-Hsyang Chu, yhchu@jupiter.ss.ncu.edu.tw

Abstract

A VHF pulsed radar system was set up on the Taoyuan County seashore (24°57′58″N, 121°00′30″E; Taiwan) to observe the sea surface in the northern Taiwan Strait for the first time. The radar used a four-element, vertically polarized Yagi antenna to transmit the 52-MHz radar wave. The receiving linear array consists of four vertical dipole antennas that were located 3 m apart and attached with four independent and identical receivers. With the multichannel echoes, the direction of arrival (DOA) of the radar echoes were determined by using an optimization beamforming approach—the Capon method. Echo intensity was observed to vary principally in semidiurnal oscillation, which matched well the time series of tide gauge measurements and sea level simulations. In addition, the oscillatory characteristics of Doppler/radial velocity of the VHF radar were generally consistent with that of the HF coastal ocean dynamics applications radar (CODAR) nearby. Nevertheless, the contributions of various tidal modes to the parameters of DOA, echo intensity, radial velocity, and spectral width, varied with the range and time period (e.g., neap or spring tides). For example, the semidiurnal tides governed the variation in the echo center only in the range interval between ~15 and ~25 km from the seashore but dominated other parameters throughout the detectable range. Correlations and phase relationships between these parameters were diverse; they varied with time and had dramatic changes at around the distances of 3 and 10 km. Possible causes of these features were discussed, including sea surface wind, nearshore current, sea level height, and bathymetric effect.

Denotes content that is immediately available upon publication as open access.

© 2019 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Yen-Hsyang Chu, yhchu@jupiter.ss.ncu.edu.tw

1. Introduction

The pulsed radar operated at the very-high-frequency (VHF) band between 30 and 300 MHz is one of the most powerful instruments for not only observing the dynamic atmosphere but also for remote sensing of the sea surface for a long time (e.g., Bass et al. 1968a,b; Barrick 1971; Balsley et al. 1987; Shrira et al. 2001; Cochin et al. 2005, 2006). Atmospheric wind fields and ocean surface currents can be measured by the VHF radar based on the same physical quantity: Doppler frequency shift caused by a moving target. The Doppler spectral characteristics of the atmospheric echoes, which are the results of Bragg scattering from random fluctuations of the atmospheric refractive index, depend on many factors: three-dimensional wind vector, vertical gradients of atmospheric temperature and humidity, turbulence strength, raindrop size distribution for precipitating environment, and so on. By contrast, the spectral features of the sea surface echoes are primarily dominated by two major peaks resulting from Bragg resonances of the ocean surface waves at the scale of half the incident radar wavelength. The two spectral peaks represent the ocean waves moving toward and away from the radar, the so-called first-order spectral components. In addition, there are second-order spectral components appearing around the first-order spectral peaks, which can be employed to extract wave height and wave period (Barrick 1972). Like other sea radars at different frequency bands, the VHF radar has been applied to various studies on the sea surface: for example, sea surface current and wind direction (Heron and Rose 1986; Cochin et al. 2005), plume circulation off the river mouth (Forget et al. 1990), oceanic submesoscale vortex (Shay et al. 2000), and vertical shear of ocean surface currents (Shrira et al. 2001). Some of these studies took advantages of the superior spatial and temporal resolutions, as compared with the sea radars at lower-frequency bands, say, the HF coastal ocean dynamics applications radar (CODAR).

Plenty of advanced radar techniques have been applied to the pulsed VHF atmospheric radar, for example, digital transceiver, multiple receivers, and multiple carrier frequencies, to enhance its observational capability. For example, with the advantages of multireceiver and multifrequency radar imaging techniques, multiple irregularity layers and multiple echo centers in a range cell have been resolved (Chilson et al. 2003; Chen et al. 2009, 2014; Luce et al. 2008; Fukao et al. 2011). In this study, the Chung-Li VHF radar, which was designed originally for atmospheric observation, was modified and situated on the seashore of Taoyuan County (24°57′58″N, 121°00′30″E; Taiwan) to observe the sea surface in the northern half of the Taiwan Strait for the first time. The modified sea radar antenna system consists of a four-element, vertically polarized Yagi antenna for transmission of radar waves at the carrier frequency of 52 MHz, and a linear array with four vertical dipole antennas that were parallel to the coastline for receiving the sea echoes. The echo signals were processed by four independent and identical receivers to store the respective raw data series.

The echo center of radar returns can be derived with the multichannel raw data, which refers to a local echoing region contributing relatively strong radar returns. There have been several approaches for determining the echo center of the radar returns, including the algorithms of multiple signal classification (MUSIC) and beamforming (Teague et al. 1997; Barrick and Lipa 1997; Gurgel et al. 1999; Prandle 1987). In this study, we estimated the echo center by means of Capon’s method (Capon 1969). The Capon method is also a type of beamforming but is an optimization processing, which has been applied to the VHF atmospheric radar successfully (Palmer et al. 1998; Saito et al. 2006; Chen et al. 2007; Yu et al. 2010; Kumar et al. 2016); we will show its feasibility in determining the echo center of sea echoes. The main goal of this study is to validate the VHF radar-observed sea echoes, including periodic oscillations in the time series of echo intensity, Doppler/radial velocity, spectral width, and echo center. Comparisons with the observations of an HF CODAR, in situ measurements of local tidal gauges, and simulation results from the TPXO8-atlas model (Egbert and Erofeeva 2002) were made.

The Taiwan Strait that separates the Taiwan island and Mainland China is a continental marginal sea of about 360 km in length and 130–200 km in width. The deepest water of the strait is about 120 m with a mean depth of 60 m. Semidiurnal tides are predominant, which propagate southward along the strait and exhibit standing waves with an antinodal point at the middle of the Taiwan Strait (Jan et al. 2004). The amplitude of the semidiurnal tide (M2 constituent) reaches its maximum (can be above 2.5 m) in the middle part of the strait and diminishes rapidly toward both ends of the strait. The modified Chung-Li VHF radar employed exclusively for the sea surface observation was located at around the midway between the midpoint and the northern open end of the strait, where the semidiurnal tide is prevailing and the tidal amplitude can reach ~2.5 m sometimes.

This article is organized as follows. Section 2 describes the experimental setup and the signal analysis approaches of multireceiver techniques. Observational results of VHF and HF radars are compared and shown in section 3. Section 4 discusses the variations of echo intensity, echo center, and radial velocity and their relationships at different ranges and time durations. The sea level and tidal current measured or simulated are also described. Conclusions are stated in section 5.

2. Experimental setup and Capon’s method

a. Experimental setup

The duration of the experiment was from mid-October to mid-November 2015 (Chu and Su 2015). Figure 1 shows the deployment of the Chung-Li VHF sea radar as well as the arrangement of the antennas. The cross symbol in Fig. 1a denotes the location of the VHF sea radar. A four-element Yagi antenna, as shown in Fig. 1b, was set to transmit vertically polarized radar waves (52 MHz). Four vertical dipole antennas for receiving the sea echoes were arranged in a linear array with a spacing of 3 m between each adjacent antenna and were parallel to the coastline. The radar beam was transmitted horizontally in the direction of 50° north by west, and its half-power full width in the horizontal dimension was about 90°. The coordinate system for beamforming is illustrated in Fig. 1c, where the z axis is the boresight direction of the transmitted beam, the xz plane is horizontal, and the y direction is upward. The wavenumber vector k denotes the direction of beamforming.

Fig. 1.
Fig. 1.

(a) VHF radar location (24°57′58″N, 121°00′30″E) and beam direction. The background chart is from Google Maps. (b) Arrangement of VHF radar antennas. (c) Cartesian coordinate system referring to (b). The z axis represents the radar beam direction; x and y directions are, respectively, horizontal and upward; and k denotes the wavenumber vector for beamforming.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Radar parameters are given as follows. The interpulse period was 0.001 s, and coherent integration was 256, resulting in a sampling time of 0.256 s. The radar pulse length was 2 μs, and a matched filter was employed for reception, corresponding to a 500-KHz receiver bandwidth and giving a range resolution of ~300 m. The number of sampling range cells was 300, and the sampling step was 2 μs, providing a range cell of 300 m. For each cell, 128 temporal data points were taken for Fourier transform to produce the Doppler spectrum, and three of the Doppler spectrums were averaged. As a result, the total echo intensity, calculated with the raw data after removing the dc component, and the Doppler velocity and spectral width of the first-order sea echoes were obtained at a time resolution of about 128 × 0.256 × 3 = 98.304 s. Gaussian fitting around the first-order Doppler components was executed, and the peak location and standard deviation of the fitted Gaussian curve were assigned as the Doppler velocity and spectral width, respectively.

b. Beamforming and Capon’s method

For the atmosphere and sea surface, many methods have been proposed for determining the direction of arrival (DOA) of the radar returns received by a set of spaced receivers (referring to the references mentioned in the introduction). In this study, Capon’s method (Capon 1969) was utilized to retrieve the angular power distribution (brightness distribution) of the radar returns; then the echo centers (e.g., DOAs) were determined from the brightness distribution.

Previous studies have suggested that the Capon method is more robust than the linear-phase modulation method for the atmosphere (Palmer et al. 1998; Yu et al. 2000), and the former is more capable and effective than the latter in resolving multiple targets in the radar volume. Two examples of beamforming with classical (thin curve) and Capon (bold curve) methods are shown in Fig. 2a, in which the normalized brightness values are present. Notice that the angle of 50° was the geographic north (see Fig. 1). Because four receiving antennas were aligned along the coastline, only the brightness distribution in the horizontal dimension can be retrieved. In the left panel of Fig. 2a, one echo center was located at about 5° south of the beam direction, and it is evident that the Capon’s brightness distribution was more concentrated than the classical/Fourier result. The brightness values in the angular regions beyond about ±30°, those increased with the angles, were not reliable. These unreliable brightness values could be associated with the cluster echoes from the seashore or the effect of grating lobes of beamforming. The other example shown in the right panel of Fig. 2a is a two-echo center situation, in which the two centers separated by about 25° and was resolved unambiguously by the Capon method. By contrast, the classical/Fourier beamforming can hardly identify the two centers.

Fig. 2.
Fig. 2.

(a) Comparison of Capon and classical beamforming results for two situations of sea echoes received by the VHF radar. (b) Capon brightness distribution extending along the vertical dimension for locating the echo center. The brightness value is self-normalized, and the cross symbol indicates the located echo center.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

To determine all echo centers in the brightness distribution automatically and effectively, we employed the contouring process proposed by Chen et al. (2008). To use the contouring process, the brightness distribution was extended in the vertical dimension with a suitable Gaussian function, where the peak value of the Gaussian function was the computed brightness value and the standard deviation of the Gaussian function was assigned arbitrarily, say, 1°. Figure 2b shows the locating results of the two examples given in Fig. 2a, where the cross sign denotes the echo center estimated, the origin is the beam center, and the beam direction is inward to the paper. The echo centers estimated with this process for several days will be examined later. We noticed that airplane or vessel echoes were found intermittently, but they did not bias the statistical characteristics of echo centers.

3. Observations

In this section, some apparent features of echo intensity, Doppler/radial velocity, and echo center observed by the VHF sea radar are shown. Comparisons with the outputs of an HF CODAR nearby are also made.

a. Echo intensity

Figures 3a–c show the range–time variations in the echo intensities observed by the VHF sea radar for the period between 28 October and 13 November 2015. In general, the echoes diminished rapidly beyond the range of 20 km. Weak interferences occurred frequently throughout the observational range and period, as emerged from the sea echoes at distances farther than ~15 km. Some strong interferences happened sporadically, as seen around noon on 29 October and 5 November. Moreover, airplane or vessel echoes appeared intermittently. These interferences did not compromise the regular variation pattern of the intense echoes over time and range. As seen, the maximum detectable range of the sea echoes varied in a period of about a half day, with about 5-km difference in range between the peak and the trough. A later investigation shows that the echo intensity oscillated in a semidiurnal period almost through the detectable range. It is thus believed that the semidiurnal lunar tide was responsible for the semidiurnal oscillation in the echo intensity.

Fig. 3.
Fig. 3.

Range–time intensities of radar sea echoes observed by (a)–(c) 52-MHz VHF radar and (d) 4.58-MHz CODAR. The distance between the two radar sites is about 8 km.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Figure 3d shows the range–time–intensity distribution of the HF CODAR echoes collected between 10 and 12 November 2015, at the DATN station (see Fig. 1a). The HF CODAR is a frequency-modulated continuous wave (FMCW) radar operating at a center frequency of 4.58 MHz with a bandwidth of 40.439 kHz and located at about 8 km northeast from the VHF sea radar (Wang et al. 2017). Compared with the VHF echoes received during the same period (Fig. 3c), the features of semidiurnal oscillations in both echo intensity and detectable range were absent in the HF CODAR sea echoes. This could be due to the much coarser range resolution (~3.75 km) of the CODAR echoes. Nevertheless, the radial velocity of the CODAR echoes indeed oscillated in semidiurnal period; that will be shown in the following section.

b. Doppler velocity

Figure 4 shows an example of the power spectra observed by the VHF sea radar, where the first-order sea echoes having salient spectral peaks at around −0.7 and 0.7 Hz (indicated by the vertical dashed lines), corresponding to a radial velocity of about 2 m s−1, were clearly seen. The echoes detected at a distance of around 15 km in the form of a horizontal line were caused by an airplane. Moreover, the second-order sea echoes can be identified around the first-order spectral peaks; their characteristics and uses are not exhibited in this paper.

Fig. 4.
Fig. 4.

Typical spectra of sea echoes observed by 52-MHz VHF radar.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

In general, the spectral characteristics of the VHF sea radar were similar to those of the HF CODAR. A comparison of the radial velocities, estimated by removing the first-order backscatter components, between the two types of radars were made, as shown in Fig. 5, where the values derived from positive (advancing) and negative (receding) Doppler spectra are compared, respectively. As seen, a common feature of these radial velocities was a semidiurnal variation. The range–time variations in radial velocities of the two radars, derived from positive or negative Doppler spectra, were in principle consistent. A remarkable difference is that the range-dependent characteristic of the radial velocities observed by the VHF radar was more prominent within the range of about 10 km than those of the HF CODAR. This could be owing to a smaller wave scale (~3 m) and a shorter range cell (300 m) taken by the VHF radar, compared with the HF CODAR operations. It is thus possible that the VHF radar can observe the variation in sea surface current at a smaller scale, although the detectable range of the VHF radar is shorter than that of the HF CODAR.

Fig. 5.
Fig. 5.

Radial velocities observed by 4.58-MHz CODAR and 52-MHz VHF radars. (a) Positive and (b) negative Doppler components. The first-order backscatter component has been removed. Data time: 10–12 Nov 2015.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

c. Echo center

Figure 6a exhibits an example of the brightness distribution retrieved by the Capon method with the four channel echoes. The corresponding power spectra have been shown in Fig. 4. The ranges on the ordinate and abscissa are, respectively, the distances from the VHF radar in the beam axis and transverse beam directions. The angular extent was set between −50° and 50° with respect to the beam axis direction, and an imaging step of 0.5° was given in the processing of the Capon method. Small open circles indicate the locations of brightness centers (i.e., echo centers) in the range cells. Occasionally, there was more than one echo center in a range cell. The occurrence of multiple echo centers could be associated with different isolate and discrete targets such as airplanes and vessels or caused by external radio interference. Moreover, the ground clutters and the possible interference from the HF CODAR might also compromise the analysis result. In light of these considerations, we have limited the imaging within an angular interval between −50° and 50°.

Fig. 6.
Fig. 6.

(a) Brightness distribution (self-normalized) and echo center (dots) retrieved by the Capon method. (b) Histograms of echo center locations for three time periods: 28–30 Oct and 3–6 and 10–13 Nov 2015, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

As seen in Fig. 6a, the echo centers were very close to the beam axis direction (indicated by the solid white line) within the range of ~15 km. As the range increased, the echo center locations shifted dramatically away from the beam axis from about 5° at ~15 km to about 20° at ~22 km and stayed at around 20° beyond ~22 km. An airplane was detected at a distance of around 15 km, and its location was at around 10° south of the beam axis direction. Figure 4 shows that the airplane echoes were characterized by a nearly horizontal line in the Doppler spectra.

Figure 6b shows the histograms of echo center locations as a function of range for the three consecutive periods of observations displayed in Figs. 3a–c, respectively. The number in each range cell has been self-normalized and is presented in gray palette. In general, a dramatic change in the echo center location took place between 15 and 25 km, and there was no remarkable difference in the patterns of echo center distributions between the three time periods. For the range farther than about 25 km, the SNRs were very low and were ignored in the study. Considering that the system phase differences of the radar-receiving channels were not calibrated precisely in this experiment, the estimated locations of echo centers were only relative values. In the following, we examine mainly the relative variation in echo center location. However, a short discussion will also be made for the dramatic change in the echo center location occurring between 15 and 25 km.

4. Further examination and discussion

In the previous section, the capabilities of detecting sea surface characteristics using a VHF radar system are demonstrated. In this section, more characteristics of the sea echoes observed by the VHF radar system are discussed and examined in greater detail.

a. Oscillatory characteristics of radar echo intensity, echo center, and radial velocity

As exhibited in Fig. 5, the radial velocities oscillated in a regular semidiurnal period through the detectable range, and the maximum detectable range of the sea echo intensity shown in Fig. 3 also varied in a semidiurnal period. In view of this, it is reasonable to believe that the sea echo intensity at respective range cells also had the characteristic of semidiurnal oscillation.

An example taken at the range cell of 17.3–17.6 km in the period from 28 October to 13 November 2015 was examined and is shown in Fig. 7. In Fig. 7a, the time series of sea echo intensity (bold curves) and the tide gauge–measured sea levels, made respectively at Zhuwei harbor (25°07′05″N, 121°14′36″E; about 26 km north the VHF radar site; solid thin curves) and Shinchu harbor (24°50′55″N, 120°55′14″E; about 18 km south the radar site; dot series), are compared, in which each data point represents 1-h average. As seen, the semidiurnal oscillations of the sea echo intensity were in harmony with those of the local tide gauge–measured sea levels. In this example, the sea echo intensity varied between 44 and 55 dB, and the tide gauge–measured sea levels changed their amplitude between −250 and 250 cm. Two slanted dashed lines indicate the daily phase shift of the oscillation; the oscillation is associated mainly with the M2 tidal current. In addition to the tide gauge–measured sea level data, the sea levels at 15 (thin curves) and 20 km (dot series) simulated by the TPXO8-atlas model (Egbert and Erofeeva 2002) were also compared with the sea echo intensity, as shown in Fig. 7b. It is obvious that the semidiurnal oscillations of the radar-measured echo intensity were also in accord with the model-simulated sea levels.

Fig. 7.
Fig. 7.

(a) Radar echo intensity (bold line) and the seal levels measured at Zhuwei (solid thin line) and Shinchu (dotted line) harbors. The height range of sea level is between −250 and 250 cm, and echo intensity is present between 44 and 55 dB for the range cell between 17.3 and 17.6 km. The slanted long dotted lines indicate approximately the peaks of the oscillatory curves. (b) As in (a), but the simulated sea levels at the ranges of 15 and 20 km are shown (solid and dotted lines). (c) Simulated tidal current directions: rightward and upward correspond to eastward and northward, respectively. (d) Comparisons of temporal variations between echo intensity, echo center, and radial velocity.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Figure 7c displays the model-simulated velocity vectors of tidal currents at the range of 20 km. Because the Taiwan Strait is a long and narrow water channel that is aligned in the northeast–southwest direction, this orientation confines the tidal current to advance and recede along the northeast–southwest direction.

Figures 7a–c show that there was an explicit 90° phase difference in the oscillations between sea level and tidal current velocity, which is one of the fundamental characteristics of flood and ebb tides. The two slanted dashed lines superimposed in Fig. 7a indicate an approximately 50-min phase delay of daily variation in the periodic oscillation. This phase difference combined with the relation between sea level and tidal current velocity provide solid evidences to suggest the leading role of the semidiurnal tides in the VHF radar-measured sea echo intensity.

Semidiurnal oscillation occurred not only in the radar echo intensity and radial velocities but also the echo center, as displayed in Fig. 7d for the period between 10 and 13 November 2015. Variations in echo intensity and echo center were observed to be in phase, meaning the closer the echo center is to the beam axis, the higher the echo intensity will be. It suggests that a larger beam weighting for an echo center closer to the beam axis results in a higher echo intensity. However, the echo intensity is also governed by other factors such as sea level height and amplitude of the wave component that corresponds to the Bragg backscattering; this issue is discussed later in more detail.

The radial velocities have been demonstrated in Fig. 5 to oscillate in a semidiurnal period throughout the detectable range. However, the positive and negative radial velocities observed by either HF CODAR or VHF radars were generally out of phase, and those detected by the VHF radar had more diverse phase relationships throughout the range and time. The lower two panels of Fig. 7d show an example of the VHF radar. As seen, the oscillations of the two radial velocities were almost in phase during the first day (10 November) and shifted to a larger phase difference during the second and third days (11–12 November), then became almost in phase again during the final day (13 November). It seems that the radial velocities derived independently from positive or negative Doppler components did not always represent the radial velocity of the sea or tidal current. In a previous study (Bass et al. 1968b), the true first-order backscatter frequency is also in relation to the sea surface wind:
fBrgπλ+2|Vwcosθ|λ=fB+2|Vwcosθ|λ,
where fB is the theoretical first-order backscatter frequency without sea surface wind, g is the gravity acceleration, Vw is the surface wind speed, θ is the angle between wind direction and radar beam axis (directed rightly to the ocean), and λ is the radar wavelength.
Giving a radial current velocity U, the observed positive and negative first-order Doppler frequency, fD+ and fD, are connected with U, fB, and fBr in the following expression:
fD±=±fBr+2Uλ=±(fB+2|Vwcosθ|λ)+2Uλ.
Assuming U remains unchanged in the time duration for an estimate of fD, which was about 98.304 s in the computation, U can be calculated by (fD+ + fD)λ/4, and the surface wind-related term fBr can be estimated with (fD+fD)/2 because
fD+fD=2(fB+2|Vwcosθ|λ)=2fBr.
Figure 8 shows the calculated results, in which the values of fBrfB, U, and their histograms along the range are present. Obviously, the value of fBrfB varied with time and range and oscillated in a semidiurnal period. We noticed that severe interference occurred after 72 h, which was during the final day; therefore, the resultant values on 13 November were discarded in presentation of the histograms shown in the right panels. The histogram reveals that the values of fBrfB were mostly larger than zero, which is expected because the sea surface wind always adds a positive component to fB to give a larger fBr, as expressed in (1). On the other hand, the radial current velocity U was biased to negative value at the range farther than about 5 km (ignore the low-SNR data beyond ~25 km); U is expected to vary symmetrically around zero if the tidal current is the only cause of U variation, as simulated in Fig. 7c. According to the biased U values, therefore, a residual current might exist and contribute to the radial velocity. To determine the exact sea surface current velocity and direction, two radar stations for reception of the sea echoes are needed in the future, like the HF CODAR system.
Fig. 8.
Fig. 8.

The 52-MHz VHF radar observations. (a) Difference between the sea surface wind-related backscatter frequency fBr and the theoretical value fB. Data time: 10–13 Nov 2015. (b) Radial velocity estimated by combining the positive and negative first-order Doppler means. The right two panels show the histograms of the values in the left panels, but the data time is 10–12 Nov 2015. The data with SNR < 0.25, appearing mostly at the range farther than 20-something km, have been discarded in the histogram.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

b. Range-dependent oscillation in radar echo intensity, echo center, and radial velocity

In the case of Fig. 7, the radar echo intensity, echo center, and radial velocity were demonstrated to be associated with the semidiurnal tides. The variation in echo intensity was almost the same phase with the sea levels observed and simulated, and in fact, its oscillating amplitude also varied with the lunar day. As illustrated in Fig. 9a, a raised (lowered) sea level not only increases (decreases) the radar cross section within the radar beam but also raises (lowers) the vertical scattering point toward (away from) the radar beam axis; both give larger (smaller) weightings to the echoes. Such an effect of sea level on the radar echo intensity can be verified from the daily variation of echo intensity displayed in Figs. 9b and 9c. As seen, the summits of oscillation decreased with time after a full moon, then increased eventually, and reached its maximum around the day of a new moon, which can be attributed to the processing of a lower and lower sea level after a full moon and a higher and higher sea level as the spring tide approaches.

Fig. 9.
Fig. 9.

(a) Schematic plot of variation in sea level within the radar beam and (b),(c) temporal variations in echo intensity for the range cell between 17.3 and 17.6 km, during 28 and 30 Oct and 10 and 13 Nov 2015.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

The features seen in Fig. 7 were also found to occur in other range cells. A range-dependent oscillation of the parameters can be examined by spectral analysis of the temporal data series at all range cells. First, the echo intensity, echo center, and radial velocity were averaged within 0.5 h, respectively, then Fourier transform was applied to these mean data series. Results of two datasets are exhibited in Figs. 10a and 10b, respectively. The upper three panels of Fig. 10a show the normalized spectral power of the range cell between 17.9 and 18.2 km, and only the first 32 amplitudes at positive frequency components are present because the input data are real values. The frequency scale in the abscissa was enlarged 3600 times as a result of the unit being hours instead of seconds. The duration of data was 78 h, but only the first 64 h of data were taken to produce 128 estimates for Fourier transform. As a result, the true frequency bin was about 4.340 278 × 10−6 Hz [= (1/64 h)/3600 s h−1]. In the plot, the four vertical dashed lines from left to right denote the oscillatory periods of 24-, 12-, 6-, and 4-h periods, respectively. The periods of semidiurnal and diurnal tides are about 12 h 25 min and 24 h 50 min, respectively; therefore, the spectral peaks associating with the semidiurnal and diurnal tides are located slightly at left of the dashed lines.

Fig. 10.
Fig. 10.

(a) (top row) Spectral analysis of echo intensity, echo center, and radial velocity in the range cell between 17.9 and 18.2 km. (middle row) Histograms of spectral peaks in the range interval between 0.5 and 39.5 km. (bottom row) Spectral contributions of various tidal modes. (b) As in the middle and bottom rows in (a), but for another period.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

The second-row panels of Fig. 10a are the histograms of spectral peaks at various frequency components, which were integrated from all range cells. To determine the spectral peaks effectively, the locating process proposed by Chen et al. (2008) was employed. In the locating process, 16 contour levels were given, and the peak levels below 25% of the maximum peak level were discarded; this threshold can remove some minor peaks due to noise. Such resultant histograms disclosed remarkable occurrences of diurnal and semidiurnal tides in echo intensity, echo center, and radial velocity. Moreover, the overtides with 6- and 8-h periods, locating at 0.167 and 0.125 frequency components, were also found in echo intensity and echo center but were not apparent in radial velocity. The semidiurnal tides are known to be the major tides in the sea area observed by the VHF radar, and the diurnal tides take second place. The spectral analysis results demonstrated here were in agreement with this scenario. Unexpectedly, there was a peak denoting the period of 5 h in the spectra of echo intensity.

The data shown in Fig. 10a were collected during the period of neap tide. On the other hand, Fig. 10b exhibits the period around the spring tide. According to the histogram of spectral peaks, the diurnal and semidiurnal tides were again the two most significant modes. However, the overtides were not exactly similar to the previous period. For example, the 4-h-period tides were detected in echo power, echo center, and radial velocity, which were not observed or were not important in the previous period. Moreover, the 6- and 8-h overtides were weak in echo intensity, and the 6-h overtide was remarkable in radial velocity and moderate in echo center; these features were different from the previous period.

The histogram of spectral peaks can reveal the roles of different tidal modes. Nevertheless, significance of different tidal modes varied with range. Therefore, respective contributions of diurnal, semidiurnal, 8-h (1/3 day), 6-h (1/4 day), and 4-h (1/6 day) tides were estimated for each range cell, as shown in the third- and bottom-row panels in Fig. 10. The contribution was estimated as follows: the spectral amplitudes of various tides were added up, respectively, and divided by the total amount of spectral amplitude summed from the 64 frequency bins, resulting in a value of the contribution in the abscissa. According to the tidal period, the spectral amplitudes of three frequency bins between 2 and 4, 5 and 7, 8 and 10, 10 and 12, and 16 and 18 were summed, respectively, to represent the diurnal, semidiurnal, 8-, 6-, and 4-h tides. The vertical dashed lines located at the value of ~0.0469 indicate the white noise level calculated from 3 divided by 64.

In the contribution profiles of echo intensity, the semidiurnal tides were the most significant for the range below ~25 km. The diurnal period was also substantial, but during the spring tide period between 10 and 13 November, it occupied less spectral power; this indicates that the semidiurnal tides were much more significant during the spring tide period. Farther than the range of 25 km, the diurnal tides were the most important in the first period (neap tide) and were comparable with the semidiurnal tides in the second period (spring tide). However, the echo intensity diminished quickly beyond the range of 20-something km according to the range–time intensity shown in Fig. 3. We suspect that the diurnal variation of echo intensity beyond the range of 25 km could be the daily variation of noise level, possibly caused by sun radiation and abundant electric waves from human activities. Since the noise governed the radar returns at farther ranges, we should ignore the results in the distance beyond about 25 km. Within the range of 10 km, the contributions of overtides were also detected, in which the period of 8 h (1/3 day; the curve in blue) seemed more significant among the overtides. The overtides are produced during the tidal waves traveling from a deeper sea to the seashore. Indeed, the bathymetric chart shown in Fig. 11 illustrates a rapid descending of water depth within the range of 10 km from the seashore, supporting the production of overtides in the sea area.

Fig. 11.
Fig. 11.

Bathymetric chart around the sea area observed by the VHF radar.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

The significance of semidiurnal tides was also verified from the contribution profile of radial velocity, as shown in the rightmost column of Fig. 10. It is noteworthy in the second period that the spectral contribution of the diurnal tides in radial velocity was as low as the overtides; this indicates again the major role of the semidiurnal tides in the spring tide period.

One of the advantages of using a multireceiver radar system is to determine the echo center. Such an obtained echo center was also found to vary with the tides, and one case has been shown in Fig. 7d. The range-dependent contribution of various tidal modes to echo center is shown at the central column and at the third and bottom rows of Fig. 10. Ignoring the far range (>25 km) having a low noise level, the value of echo center varied with the diurnal and semidiurnal tides more clearly beyond the range of 15 km, which was quite different from the range variations of echo intensity and radial velocity. We will discuss this issue in more detail in section 4e.

c. Relationships of spectral width and echo intensity

As shown in Fig. 9a, a change in sea level height leads to a variation in radar cross section within the range cell and then causes a change in echo intensity. One more factor that may alter the echo intensity is the roughness of the sea surface, which could be driven by surface wind and tidal current, and could be indicated by the spectral width. Figure 12a exhibits the relationship between echo intensity and spectral width during 10 and 13 November 2015. The spectral width is present in color, and the echo intensity is displayed in curves only for several selected range cells (1, 6, 11, …, 81; corresponding to 0.515, 2.15, 3.615, …, 24.815 km in range). For the sake of inspecting the oscillations easily, the echo intensity curves were self-normalized and expanded into a scale of 5 km in range. Evidently, the increase in spectral width is in accord with both increased and decreased periods of echo intensity and appeared in a temporal period of about 6 h. This feature can be attributed partly to the fact that the sea surface is more uneven during the rising and falling durations of the sea level when the tidal currents advance into and recede from the strait, leading to rising and falling radar returns, correspondingly. Inspecting the other two periods of radar data, the features were alike (not shown).

Fig. 12.
Fig. 12.

(a) Spectral widths estimated from (top) positive and (bottom) negative first-order Doppler spectral lines. The curves overlaying on the color map are the 1-h mean echo intensity of several selected range cells. (b) Temporal variations in spectral width (black), estimated from positive first-order spectral lines, and echo intensity (red). Blue line connects the peak locations of oscillatory curves of the echo intensity between adjacent range cells. Data time: 10–13 Nov 2015.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Readers may notice the echo intensity at the range cell 6 (~2.15 km in range) was almost 180° out of phase with the others. Figure 12b displays the temporal variations of echo intensity (red curve) and spectral width (black curve) between 0.5 and 3.5 km, in which the blue line segments reveal the trend of phase shift in the echo intensity curves between adjacent range cells. Remarkably, there was a phase deflection in the echo intensity curves around 2 km. The causes of such phase deflection were not examined yet in this study. More in situ and radar observational evidence, such as nearshore sea current, wave front, and surface wind, are needed to examine this feature specifically.

d. Relationships of radial velocity and echo intensity

Another characteristic deserving further examination is the phase difference of 90° between the radar echo intensity and the radial velocity simulated purely from the tidal current, as provided for the case in Fig. 7. This kind of phase relationship, however, may not be valid when a residual current mixes with the tidal current, as examined in Fig. 13. Figure 13a shows the radial velocity in color and displays the echo intensity in curves only for several selected range cells (analogous to Fig. 12a). The two panels of Fig. 13b exhibit the histograms of correlation coefficients and phase differences between radial velocity and echo intensity, respectively, for the data shown in Fig. 13a. The correlation coefficient and phase difference were obtained via the following processes. First, the 1-h mean was computed, then each correlation coefficient was calculated with 24 data points of the two mean series and by shifting one data point (1 h) for each calculation. Such calculation represents the zero-lag correlation of the two data series. The zero-lag correlation coefficients of respective range cells are shown in the left panel of Fig. 13b. For each process of 24-h data series, we also shifted the radial velocity forward to 11 h and backward to −12 h to produce 24 correlation coefficients. The time shift having the maximum value of the correlation coefficient can reveal the phase difference between the two data series, in which a positive time shift indicates a leading phase of radial velocity over echo intensity. In the present calculation, the 3-h time shift was equivalent to 90° leading phase because the oscillatory periods of the two data series were about 12 h. Such obtained time shifts are shown in the right panel of Fig. 13b. Notice that the time shifts of 0 and −12 h, 1 and −11 h, and so on are the same phase difference.

Fig. 13.
Fig. 13.

(a) Radial velocity estimated by combining the positive and negative first-order Doppler means. The curves overlaying on the color map are the 1-h mean echo intensity of several selected range cells. (b) (left) Histogram of correlation coefficient between radial velocity and echo intensity. Calculation of correlation coefficient is made with 1-h mean data series. (right) Leading time of radial velocity over echo intensity. The 3-h leading or delay represents a phase difference of 90° in the wavy curves. Data time of (a) and (b) is between 10 and 13 Nov 2015. (c),(d) As in (b), but present for the data periods of 3–6 Nov and 28–30 Oct 2015, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Figure 13a shows that, ignoring the low-SNR data at the range farther than about 25 km and discarding temporarily the curve around the range cell 6, the radial velocities were positive at ascending and descending wings of the echo intensity curves, respectively, within and beyond the range of about 10 km. It is evident that the relationship between radial velocity and echo intensity varied with range. More detailed features can be found in Fig. 13b. Inspecting the histogram of time shifts in the right panel, there was a very sharp change in the time shift for the range within about 3 km, which could be associated with the phase deflection in the echo intensity, as revealed in Fig. 12b. As the distance became larger than 3 km, the time shifts changed from 3 to −3 h, meaning a change of phase difference from 90° to −90°. The zero-lag correlation coefficients for a phase difference between 90° and −90° should be larger than zero, and the left panel of Fig. 13b reveals this characteristic.

Inspecting the radar data in the other two periods, as given in Figs. 13c and 13d, the relationship between radial velocity and echo intensity was more complicated. Between 3 and 20 km (approximately), most of the time shifts were larger than 3 h or smaller than −3 h, meaning the phase differences of the two data series were between 90° and 270°, so the zero-lag correlation coefficients were mostly negative. This feature was quite different from that of the previous dataset shown in Fig. 13b. By contrast, within 3 km, the variations of correlation coefficients and time shift still had dramatic changes; this was similar to the previous dataset.

e. Relationship of echo center and echo intensity

In view of a possible causal relationship between echo center and echo intensity, their correlation and phase difference were also examined, as shown in Fig. 14. The calculation was the same as that for the relationship between radial velocity and echo intensity. However, the absolute value of echo center, which indicates an angular distance from the radar beam axis, was taken in the calculation to disclose the dependence of echo intensity on echo center, irrespective of the positive or negative value of echo center with respect to the radar beam axis. As shown in the left three panels of Figs. 14a–c, the histograms of correlation coefficients revealed a range-dependent feature. We should ignore the outcomes at the range farther than about 25 km, where the SNR of radar echoes decreased quickly (generally lower than 0.25). In general, negative correlations appeared mostly at around 15 km for the three time periods, mixing with few positive values. A negative correlation indicates 90° to 270° phase difference between the two data series, corresponding to a time shift between 3 and 9 h, or between −3 and −9 h, as found clearly in the right panels of Figs. 14a–c. This feature will be discussed again later.

Fig. 14.
Fig. 14.

(a)–(c) As in Figs. 13b–d, but for the relationship between echo center (absolute value) and echo intensity. (d) (left) The 1-h mean data series of echo center (black) and echo intensity (gray). The numbers shown on the right edge of the panel are the boundaries of the curves in the range cells; the first and second columns are for the black and gray curves, respectively. (right) Temporal variation in correlation coefficient, calculated from the data series shown in the left panel.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Within the range of about 10 km, however, the correlation coefficient and time shift changed widely. Particularly during 28 and 30 October, both positive and negative correlations occurred distinctly between 5 and 10 km. Inspecting Fig. 14d, which displays the 1-h mean data series (left) and correlation coefficients (right) over the time duration of 28 and 30 October, the correlation value changed from positive to negative. In view of these results, there seemed to be some factors manipulating the location of echo center and giving various contributions at different ranges and times. One of these factors is the sea surface wind. As known, the targets responsible for the radar sea echoes are the ocean waves at the wavelength of Bragg scale; it is about 3 m for a backscattering at 52-MHz frequency. Such small-scale sea waves are likely to be driven and propelled by the sea surface wind so that their wave fronts may be partial to be perpendicular to the wind direction. Assuming that the radar echoes return mainly from the bearing perpendicular to the wave fronts of these small-scale waves, the location of echo center could be partial to the wind direction. The magnitude and direction of the sea surface wind might change with range, time, and azimuthal direction (Heron and Rose 1986), resulting in diversity of echo center, accordingly. Since the sea surface wind within the range of several kilometers from the seashore are also associated with the land and sea breeze, and other mechanisms such as nearshore currents and swirls influence the wave front direction and echo intensity as well, these factors could cause the variations in echo center and echo intensity to be nonperiodic or incoherent within the range of several kilometers from the seashore. In situ measurements operated coordinately with the radar observation are necessary in the future to verify these possibilities.

Next, we propose two more situations sketched in Fig. 15 to discuss the variation of echo center at the range of around 15 km. A discussion with Fig. 9a suggests that the rising and falling sea levels may produce vertical variations in echo center. This was thought initially not to influence the azimuthal echo center retrieved by the present radar configuration and beamforming method. However, for a broad radar beam that covers a wide angular range, it is possible for a slightly slanted sea level to occur in the radar beam when the tidal currents come into and withdraw from the strait, as sketched in the left panel of Fig. 15. For the four-element dipole antenna employed in this study, the half-power full width on the horizontal plane was about 90°, giving a horizontal extent of about 15 and 30 km at the ranges of 10 and 20 km, respectively. When the slightly slanted sea level rises and falls in the radar volume, it shifts the azimuthal and vertical locations of echo center accordingly because of a nonuniform beam weighting (larger closer to the beam center). Another situation is described in Fig. 15b, where the sea level is not slanted but the radar beam pattern is tilted owing to some reasons such as a nonvertically stood transmitting antenna or nonhorizontal ground, the location of echo center could also vary with the rising and falling sea levels. Both scenarios result in oscillations of echo intensity and echo center with the same period of sea level and produce a negative correlation between the two data series, as found around the range of 15 km in Fig. 14. In conclusion, the tidal currents made the sea level oscillatory then caused variations in echo intensity and echo center around the range of 10-something km.

Fig. 15.
Fig. 15.

(left) The echo center (solid circle) associating with altitudinal variation of a slanted sea level (dashed lines). The gray box indicates the fan-shaped radar beam pattern radiated by the vertical four-element transmitting antenna. (right) The echo center associating with altitudinal variation of a horizontal sea level in a slanted radar beam pattern.

Citation: Journal of Atmospheric and Oceanic Technology 36, 2; 10.1175/JTECH-D-18-0110.1

Finally, the dramatic shift in the location of echo center happening between 15 and 25 km, as seen in Fig. 6, is worthy of a short discussion. It is well known that the wave fronts of ocean waves will be refracted and tend to be parallel to the contour of water depth (or coastline) as the waves propagate from deep water to shallow water. Although this kind of wave front refraction is more significant for a longer wave, the 3-m short waves that ride on the longer waves may become the tracers of the longer waves. If the radar echoes return mainly from the bearing perpendicular to the wave fronts, the echo centers at the distances closer to the seashore are anticipated to get closer to the beam axis direction that is outward the sea and perpendicular to the coastline. The wave front of 3-m sea wave may not be parallel to the coastline exactly because of sea surface wind, swirl, or nearshore current, but the change in echo center location along the range, as exhibited in Fig. 6, seems to expose the shallow-water effect on the wave front. Referring to the bathymetric chart in Fig. 11, the refraction of the wave front could take place at 10-something km, supporting the above scenario.

5. Conclusions

This is the first time the Chung-Li VHF radar system was employed to observe the sea surface in the northern Taiwan Strait. Four independent receiving antennas were set up to receive the radar echoes and determine the echo center by using an optimal beamforming approach, the Capon method, which has been verified to be feasible for bearing determination of sea echoes and is superior to the classical/Fourier beamforming. Broad examinations on the parameters of radar echo intensity, radial velocity, and echo center and their relationships within the detectable range and time intervals have been made. A primary feature of these parameters is the semidiurnal oscillation, which corresponds with the known characteristic of the sea area observed. Comparisons between the sea echoes collected by the VHF and HF CODAR radars, and between the variations in radar echo intensity and tide gauge–measured and model-simulated sea levels, have demonstrated a significant role of the semidiurnal tides in the VHF radar echoes. The semidiurnal tides not only regulated sea level and echo intensity but also increased the spectral width of sea echoes during the rising and falling periods of sea level and echo intensity. It is speculated that the sea surface gets rougher during rising and ebb tides, resulting in a 6-h period of variation in spectral width.

More investigations have shown, however, that significances of various tidal modes changed with the distance from the radar site as well as with the time period (e.g., neap or spring tides). For example, the semidiurnal tides governed the variation in echo center only in the range interval between about 10 and 25 km but dominated the variations in echo intensity and radial velocity throughout the detectable range. In addition, the diurnal tides contributed more in the neap tide than in the spring tide period.

It is also disclosed that, bordering at about the distances of 10 and 3 km from the seashore, there had been different features of correlations between the oceanic parameters examined in this study. We have discussed widely the possible causes of these changeable correlations, such as sea surface wind, nearshore current, sea level height, and bathymetric effect on the refraction of wave front. Nevertheless, a deeper understanding of the causal relationship between these parameters needs more comprehensive observations, including in situ measurements, which is expected in our future observations.

Based on the present initial results, the Chung-Li VHF radar, established originally for atmospheric study, has demonstrated its capabilities of remote sensing the sea surface. Particularly, the VHF radar system can assign a range cell and a sampling time much smaller than that of the HF CODAR, giving observations of small-scale sea parameters at higher spatial and temporal resolutions. It is anticipated that the echo center determined by the beamforming of the VHF radar can be used to improve the estimate of sea surface current velocity, providing that the system phase differences between the independent receivers are calibrated accurately to obtain a more precise location of echo center and more than two VHF radars are set up at different locations like the HF CODAR systems. In addition, some sea parameters, such as oceanic submesoscale vortex, sea surface wind, and sea wave height and period, can be examined by the VHF radar. Even an early warning of tsunami based on the sensitivity of echo intensity, radial velocity, and echo center of the VHF radar to the sea level and sea current can be expected.

Acknowledgments

This research was partly supported by the Taiwan Ocean Research Institute, National Applied Research Laboratory (NARL), and also partly supported by the Ministry of Science and Technology, Taiwan, under Grants MOST105-2111-M-039-002-MY2 and MOST107-2111-M-039-001-MY3. The Chung-Li VHF radar was set up by Graduate Institute of Space Science, National Central University, Taiwan.

REFERENCES

  • Balsley, B. B., A. C. Riddle, W. L. Ecklund, and D. A. Carter, 1987: Sea surface currents in the equatorial Pacific from VHF radar backscatter observations. J. Atmos. Oceanic Technol., 4, 530535, https://doi.org/10.1175/1520-0426(1987)004<0530:SSCITE>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., 1971: Theory of HF and VHF propagation across the rough sea, 2, application to HF and VHF propagation above the sea. Radio Sci., 6, 527533, https://doi.org/10.1029/RS006i005p00527.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., 1972: First-order theory and analysis of MF/HF/VHF scatter from the sea. IEEE Trans. Antennas Propag., 20, 210, https://doi.org/10.1109/TAP.1972.1140123.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barrick, D. E., and B. J. Lipa, 1997: Evolution of bearing determination in HF current mapping radars. Oceanography, 10 (2), 7275, https://doi.org/10.5670/oceanog.1997.27.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bass, F. G., I. M. Fuks, A. I. Kalmycov, I. E. Ostrovsky, and A. D. Rosenberg, 1968a: Very high frequency radiowave scattering by a disturbed sea surface—Part I: Scattering from a slightly disturbed boundary. IEEE Trans. Antennas Propag., 16, 554559, https://doi.org/10.1109/TAP.1968.1139243.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bass, F. G., I. M. Fuks, A. I. Kalmycov, I. E. Ostrovsky, and A. D. Rosenberg, 1968b: Very high frequency radiowave scattering by a disturbed sea surface—Part II: Scattering from an actual sea surface. IRE Trans. Antennas Propag., 16, 560568, https://doi.org/10.1109/TAP.1968.1139244.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Capon, J., 1969: High-resolution frequency-wavenumber spectrum analysis. Proc. IEEE, 57, 14081419, https://doi.org/10.1109/PROC.1969.7278.

  • Chen, J.-S., P. Hoffmann, M. Zecha, and C.-H. Hsieh, 2008: Coherent radar imaging of mesosphere summer echoes: Influence of radar beam pattern and tilted structures on atmospheric echo center. Radio Sci., 43, RS1002, https://doi.org/10.1029/2006RS003593.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, J.-S., C.-L. Su, Y.-H. Chu, G. Hassenpflug, and M. Zecha, 2009: Extended application of a novel phase calibration method of multiple-frequency range imaging to the Chung-Li and MU VHF radars. J. Atmos. Oceanic Technol., 26, 24882500, https://doi.org/10.1175/2009JTECHA1295.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, J.-S., J. Furumoto, and M. Yamamoto, 2014: Three-dimensional radar imaging of atmospheric layer and turbulence structures using multiple receivers and multiple frequencies. Ann. Geophys., 32, 899909, https://doi.org/10.5194/angeo-32-899-2014.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chen, M.-Y., T.-Y. Yu, Y.-H. Chu, W. O. J. Brown, and S. A. Cohn, 2007: Application of Capon technique to mitigate bird contamination on a spaced antenna wind profiler. Radio Sci., 42, RS6005, https://doi.org/10.1029/2006RS003604.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chilson, P. B., T.-Y. Yu, R. G. Strauch, A. Muschinski, and R. D. Palmer, 2003: Implementation and validation of range imaging on a UHF radar wind profiler. J. Atmos. Oceanic Technol., 20, 987996, https://doi.org/10.1175/1520-0426(2003)20<987:IAVORI>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chu, Y.-H., and C.-L. Su, 2015: Sea surface radar echoes collected by the Chung-Li VHF radar from 2015-10-28 to 2015-11-13. Chung-Li VHF Radar Station.

  • Cochin, V., P. Forget, B. Seille, and G. Mercier, 2005: Sea surface currents and wind direction by VHF radar: Results and validation. Europe Oceans 2005, Brest, France, Institute of Electrical and Electronics Engineers, 8573991, https://doi.org/10.1109/OCEANSE.2005.1513183.

    • Crossref
    • Export Citation
  • Cochin, V., V. Mariette, P. Broche, and R. Garello, 2006: Tidal current measurements using VHF radar and ADCP in the Normand Breton Gulf: Comparison of observations and numerical model. IEEE J. Oceanic Eng., 31, 885893, https://doi.org/10.1109/JOE.2006.886077.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Egbert, G. D., and S. Y. Erofeeva, 2002: Efficient inverse modeling of barotropic ocean tides. J. Atmos. Oceanic Technol., 19, 183204, https://doi.org/10.1175/1520-0426(2002)019<0183:EIMOBO>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Forget, P., J. L. Devenon, J. C. De Maistre, M. Leveau, and P. Broche, 1990: VHF remote sensing for mapping river plume circulation. Geophys. Res. Lett., 17, 10971100, https://doi.org/10.1029/GL017i008p01097.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fukao, S., H. Luce, T. Mega, and M. K. Yamamoto, 2011: Extensive studies of large-amplitude Kelvin–Helmholtz billows in the lower atmosphere with VHF middle and upper atmosphere radar. Quart. J. Roy. Meteor. Soc., 137, 10191041, https://doi.org/10.1002/qj.807.

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

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Heron, M., and R. J. Rose, 1986: On the application of HF ocean radar to the observation of temporal and spatial changes in wind direction. IEEE J. Oceanic Eng., 11, 210218, https://doi.org/10.1109/JOE.1986.1145173.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jan, S., Y.-H. Wang, D.-P. Wang, and S.-Y. Chao, 2004: Incremental inference of boundary forcing for a three-dimensional tidal model: Tides in the Taiwan Strait. Cont. Shelf Res., 24, 337351, https://doi.org/10.1016/j.csr.2003.11.005.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kumar, S., T. N. Rao, V. K. Anandan, V. N. Sureshbabu, and T. Arunamani, 2016: Comparison of horizontal winds and turbulent eddy dissipation rates obtained by two different multi-receiver techniques—Spaced antenna and postset beam steering. J. Atmos. Sol.-Terr. Phys., 137, 4452, https://doi.org/10.1016/j.jastp.2015.11.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Luce, H., G. Hassenpflug, M. Yamamoto, S. Fukao, and K. Sato, 2008: High-resolution observations with MU radar of a KH instability triggered by an inertia–gravity wave in the upper part of a jet stream. J. Atmos. Sci., 65, 17111718, https://doi.org/10.1175/2007JAS2346.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Palmer, R. D., S. Gopalam, T.-Y. Yu, and S. Fukao, 1998: Coherent radar imaging using Capon’s method. Radio Sci., 33, 15851598, https://doi.org/10.1029/98RS02200.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Prandle, D., 1987: The fine-structure of nearshore tidal and residual circulations revealed by H.F. radar surface current measurements. J. Phys. Oceanogr., 17, 231245, https://doi.org/10.1175/1520-0485(1987)017<0231:TFSONT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Saito, S., M. Yamamoto, H. Hashiguchi, and A. Maegawa, 2006: Observation of three-dimensional structures of quasi-periodic echoes associated with mid-latitude sporadic-E layers by MU radar ultra-multi-channel system. Geophys. Res. Lett., 33, L14109, https://doi.org/10.1029/2005GL025526.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shay, L. K., and Coauthors, 2000: VHF radar detects oceanic submesoscale vortex along Florida coast. Eos, Trans. Amer. Geophys. Union, 81, 209213, https://doi.org/10.1029/00EO00143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shrira, V. I., D. V. Ivonin, P. Broche, and J. C. de Maistre, 2001: On remote sensing of vertical shear of ocean surface currents by means of a single-frequency VHF radar. Geophys. Res. Lett., 28, 39553958, https://doi.org/10.1029/2001GL013387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Teague, C. C., J. F. Vesecky, and D. M. Fernandez, 1997: HF radar instruments, past to present. Oceanography, 10 (2), 4044, https://doi.org/10.5670/oceanog.1997.19.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Wang, C.-Y., C.-L. Su, K.-H. Wu, and Y.-H. Chu, 2017: Cross spectral analysis of CODAR-SeaSonde echoes from sea surface and ionosphere at Taiwan. Int. J. Antennas Propag., 2017, 1756761, https://doi.org/10.1155/2017/1756761.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, T.-Y., R. D. Palmer, and D. L. Hysell, 2000: A simulation study of coherent radar imaging. Radio Sci., 35, 11291141, https://doi.org/10.1029/1999RS002236.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, T.-Y., J. Furumoto, and M. Yamamoto, 2010: Clutter suppression for high-resolution atmospheric observations using multiple receivers and multiple frequencies. Radio Sci., 45, RS4011, https://doi.org/10.1029/2009RS004330.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save