Recent Changes - Search:

radlab home

about

people

data

publications

support

funding

contact

FCC

ROW archive

radlab wiki

.

High Frequency Doppler Radio Scatterometers for mapping ocean currents

Physics and principles of operation

©2002 Pierre Flament

See also: physical parameters.

High Frequency Doppler Radio Scatterometers (HFDR) are simple in concept: electromagnetic waves (EM) sent to the ocean are backscattered on surface waves of exactly half the radio wavelength, just like X-rays are scattered in crystals.

Principle of Bragg scattering (click to enlarge). Incident electromagnetic (EM) wave in red and gray shading. Top: coherent superposition of reflected EM waves when the ocean wave has exactly 1/2 the EM wavelength. Bottom: incoherent superposition of reflected EM waves when the ocean wave has 0.9/2 the EM wavelength. The Radio Oceanography Laboratory logo is inspired by Bragg scattering.

Since the ocean is generally covered by waves of many different wavelengths and directions (continuous spectrum), there are always trains of waves propagating toward and away from the transmitter. The return signal from either train will be Doppler-shifted by the wave velocity, which is known exactly by the gravity wave dispersion relationship. Thus the spectrum of the return echoes consists of two peaks, symmetric with respect to the transmit frequency, in the absence of currents.

Typical backscattered spectrum (click to enlarge). The main spectral peaks correspond to the reflected signal Doppler-shifted by the motion of the Bragg waves (after Gurgel, 1999).

If the waves ride on an ocean current, the return signal is further Doppler-shifted by the radial component of the current, which can be readily estimated. With two instruments some distance apart along the coast, vector currents can be computed.

This is a precise concept, which surprisingly has taken more than three decades to be accepted by the oceanographic community (a list of early references has been compiled by Robert Stewart). Yet unlike Acoustic Doppler Current Profilers (ADCP), the scattering targets are well known, and well understood theoretically.

Wind direction can be inferred from the relative amplitude of the main peaks, and multiple scattering allows the extraction of spectral information on the wave field.

Although they have historically been called "radars", they have nothing in common with microwave navigation radars, which track targets such as ships and aircrafts using high energy pulses. They work instead by analyzing the reflection on the ocean of continuous radio waves (typically at 5 to 30 MHz, or 10 to 60 m wavelength); the technology is similar to Citizen Band radios, using very low power (typically 5 to 30 W), therefore presenting no danger to the public. Various HF radio instruments differ on several aspects:

  • whether they use ground wave or sky wave propagation
  • how the signal is modulated
  • how are the antennas configured
  • how are the echo directions estimated

(1) Do they use ground wave or sky wave?

At HF, EM wave can propagate in a "trapped" ground wave along the conductive surface of the ocean and a "free" wave in the atmosphere.

Sketch of ground and sky wave propagation

The useable range of the ground wave depends on attenuation, and increases with decreasing frequency, approximately range(km) * frequency(MHz) = 2000, thus 32 MHz = 60 km, 16 MHz = 125 km, 8 MHz = 250 km; the actual range varies greatly depending on EM background noise, and on the type of signal modulation. Attenuation increases and thus range decreases as salinity decreases, and as sea state increases. Background noise from distant sources such as lightnings, industrial noise and broadcast stations, is affected by the D-layer of the ionosphere, which has a strong diurnal cycle.

The free wave propagates to much longer distances and circles the earth, bouncing between the ionosphere and the earth surface ("sky" wave). However it is subject to multiple path propagation, and Doppler shifts by internal waves in the ionosphere. Therefore interpreting distant echos from the ocean is difficult, because Doppler shifts due to ocean currents are masked by the larger ionospheric Doppler shifts. Powerful "over-the-horizon" military radars, which have occasionally been used to monitor ocean wind and waves, are based on sky wave reflections. All modern ocean current-mapping HF radios are based on very low power ground waves.

(2) How the signal is modulated?

Modulation is necessary to achieve range resolution. Just like ADCPs which may use short pulses ("pings"), broad band chirps, or pseudo-random code phase modulation, so do HF radios. The more recent their design, the more complex the signal encoding. The intent is to lower the instantaneous transmit energy as much as possible, by transmitting with a larger duty cycle, ideally 100%.

The technology of short pulses is simplest, but there is a price to pay: the instantaneous power needed to achieve range resolution through short pulses can be quite large (often 1 kW or more), creating a risk to personnel; there is always a blank area in front of the transmitter, as the receiver must wait for the direct transmit signal to die out before listening to the echoes; and range resolution is limited by how short one is willing to make the pulse, thus by how large instantaneous power is acceptable. Instruments based on very low power continuous wave (CW) radios are immune from these problems.

Range resolution can be obtained by "chirping" the signal, i.e. ramping the frequency. This allows to lower the power, without loosing resolution. The spatial resolution in this case is not governed by the pulse length, but by the bandwidth of the chirps. Except for the very high grade components needed (dynamic range reaching 130 dB), this is a simple and easy to configure technology, easy to debug with standard lab equipment such as oscilloscopes and spectrum analyzers.

In any system, the strong direct signal from the transmit to the receive antennas or circuit must be rejected as much as possible, to avoid saturating the receiver when listening to the target echoes. In continuous wave chirped instruments, inevitably the receiver must listen to the sea echoes while the transmitter is still active. Of course the simplest is to ensure physical separation between transmit and receive antennas, at the expense of long cables, of a few hundred meters or more.

A second approach is to superimpose a repeated gate, typically 1 kHz, over long continuous chirps; through this artifact, silent transmit periods are created during which the echoes can be listened to, while maintaining spectral and spatial resolution. This is the principle of the US CODAR/SeaSonde and UK PISCES; its advantage is that there is no need for physical separation of the TX and RX antennas, and that 360 deg. coverage can be obtained with isotropic antennas. However this approach complicates the signal generation, amplification, detection and spectral estimation.

A third approach is to form an azimuthal beam on the transmit antenna, and ensure that the receive antenna lies in a null of the transmit pattern. With 2 or 4 properly phased antennas, a null is created toward the axis of the RX antennas, with better than 80 dB rejection. The transmitted beam can also be enhanced forward, improving the illumination of the ocean, and reducing signal reflection on the back topography (power lines, cliffs,...).

(3) how are the antennas configured and the echo direction estimated

Receive antennas can be monopoles or loops, and can be configured in beam steering mode, or in goniometric mode.

In the goniometric (or direction-finding) mode, four monopole antennas in a square (the old NOAA-CODAR, the WERA in DF configuration), or two orthogonal coils and a single monopole (the new CODAR-SeaSonde), are used to receive the signal. For each range, it is assumed that one set of current-induced spectral shift come from only one direction, which is estimated through a least-square procedure.

This solution achieves extreme electronic simplicity, but has two major shortcomings: drastic assumption on ocean current patterns (this is often overlooked), and impossibility to estimate wave spectrum since the second order backscatter spectrum is not always easily separated from first order spectrum.

In beam steering mode, a rake of 8, 12, 16, ... antennas at half wavelength spacing (thus 10 m at 15 MHz, 20 m at 7.5 MHz) is used to create a synthetic large aperture antenna; the longer the rake, the narrower the beam; beam width in radians scale as 2/(n-1), thus for 16 antennas, about 7.5 deg. The beam of the antenna array is steered by introducing variable phase delays in individual antenna channels.

This steering can be done in hardware, by switching sequentially different lengths of coaxial cable for each antenna (COSRAD), or in software, by recording all antennas in parallel, and combining their signals with appropriate phase shifts in software. Hardware phasing has a shortcoming: one must sequentially look in different directions, thus reducing the size of the statistical average, or reducing the temporal resolution to obtain the same number of degrees of freedom. Software phasing has also a shortcoming: noisy interferences which may saturate the demodulators, cannot be rejected from the acquired signal, since individual antennas are omnidirectional.

Beam steering achieves optimum signal sampling and allows estimation of all possible ocean parameters, at the expense of real estate. A 16 antenna rake at 8 MHz would measure 320 m long, not easy to find along a beach! Compromises between radio frequency thus antenna spacing, governing maximum range, and array length governing angular resolutions, must be found depending on the requirement of each application/deployment. Not to speak about cable handling: at 8 MHz, a 16 antenna array requires handling about 8 km of finger-thick RG-213 coaxial cable.

The two methods can of course be combined or used in parallel.

Edit - History - Print - Recent Changes - Search
Page last modified on February 04, 2023, at 09:06 pm