Rotating scan radar

The concentional radar developed for aircraft detection produces an image of a wide geographic area on a plan position indicator (PPI) oscilloscope. Radio energy is transmitted from an antenna rotating in a circular or sector scan; this energy is partially reflected from targets and terrain features, received by the same rotating antenna, and displayed on the PPI. The PPI is a polar display with the transmitter position at the centre; range or distance from the transmitter for each reflection is measured by the time taken between transmission and reception, and is displayed radially on the PPI. An example is presented in Figure 1 which shows the image produced on the PPl of a forward-looking, sector-scan, navigation radar in an aircraft flying in April 1971 over ice fields in Davis Strait off Baffin Island. It is interesting to note that the first discovery of an ice island is attributed to this type of radar carried on a U.S.A.F. aircraft, on 14 August 1946, approximately 500 km north of Alaska.

Fig. 1.

Sector-scan navigation radar display in Argus aircraft over ice in Davis Strait off Baffin Island in April 1971.

The ability to distinguish adjacent objects depends on the resolution of the radar, A limitation of this conventional rotating-scan radar is its very coarse azimuthal resolution. Although good range resolution can be obtained readily by suitably modulating the transmitted energy (wide bandwidth), the azimuthal resolution is determined by the angular width of the radio beam which in turn is determined by the size of the antenna. In practice, where the antenna must be rotated mechanically, it cannot usually be large enough to produce the high degree of azimuthal resolution desired for most remote-sensing purposes. Furthermore, because of the angular spread of the antenna beam, this azimuthal resolution of the image becomes coarser with increasing range from the transmitter.

SLR and its applications to ice studies

Side-looking radar, or SLR, is an airborne system using a different type of antenna scan system. The antenna is carried under the aircraft with its beam looking in a fixed sideways direction, at right angles to the aircraft track. Scanning is achieved by the forward motion of the aircraft so that the radar image obtained is in a Cartesian format rather than the polar format obtained with the conventional rotating scan. The SLR image is produced in a continuous strip as the aircraft moves along, and this image is usually recorded on film. Several studies of the application of such radars to ice studies have been reported, (Reference AndersonAnderson, 1966; Reference GuinardGuinard, [1970]; Reference Johnson and FarmerJohnson and Farmer, 1971; Reference Larrowe, Larrowe, Innes, Rendleman and PorcelloLarrowe and others, 1971; Reference BiacheBiache and others, 1971; Loshchilov and Voevodin, 1972; Reference Ketchum and ToomaKetchum and Tooma, 1973; Reference Jirberg, Jirberg, Schertler, Gedney, Mark, Santeford and SmithJirberg and others, 1974; Reference Gorbunov and LosevGorbunov and Losev, 1974; Reference DunbarDunbar, 1975; Parashar and others, in press).

There are two basic SLR systems which differ markedly in the image resolution obtained, and incidentally, in the complexity and cost of the radar. The simplest type is the non-coherent or real-aperture SLR system, in which the azimuthal, or along-track, resolution is determined by the along-track beam-width of the antenna. The longest possible antenna is carried under the aircraft to obtain the narrowest possible along-track beam-width. As in conventional radars, the range resolution, or cross-track resolution, is determined by the modulation (bandwidth) applied to the transmitted energy (usually narrow pulses). Although it is relatively easy to achieve high cross-track resolution, the transmitter modulation is usually chosen so that this resolution approximates the more limited along-track resolution determined by the angular beam of the antenna, As in conventional radars, because of the angular spread of the antenna beam in the along-track direction, the along-track resolution of the resulting imagery is degraded linearly with increasing range from the transmitter.

Several examples of imagery obtained with an X-band real-aperture system are shown in Figures 2 through 5, each of which illustrates points of possible interest to the glaciologist. The SLR used was a Motorola AN/APS₉₄D system flown by the Canadian Forces on an Argus aircraft. This system can image on both sides of the aircraft with possible range swaths 25, 50 or 100 km per side. All the imagery contains a non-imaged blind strip down the center of the track, which is outside the field of view of the antenna and is about twice as wide as the altitude of the aircraft. Range resolution is constant at about 30 m and along-track resolution is approximately 8 m per kilometer of range.

Figure 2, taken during Operation SKYLAB, shows part of the ice-covered Gulf of St. Lawrence and illustrates a number of interesting points. At the time this image was generated the north-western part of the Gulf was undergoing extensive ice formation, with the ice being advected to the south-east (Campbell and others, in press). The Quebec north shore west of Sept îles is visible in the upper left-hand corner of the image. The dark area between the shore and the first radar returns from ice contains both open water in the form of a shore lead and newly formed sea ice. The lack of ability to distinguish between open water and new ice is characteristic of SLR imagery, there being little reflected energy because both have flat, specularly reflecting surfaces. Some shore-fast ice is visible at the lower edge of the shore.

Fig. 2.

AN/APS-₉₄D SLR image. Gulf of St. Lawrence near north shore, west of Sept Îlies, taken on 13 January 1974. Coverage is 25 km across track and 42 km along track.

The demarcation between this smooth ice and the floes in the central portion of the image is clearly visible. The smooth floes consisting of grey ice show a radar return from the edges only. Most of the ice on the right-hand side of the track is of this type. The bottom half of the left-hand image shows a very strong radar return, where the ice is older and has a rough surface. Some surface structure within the Hoes is clearly visible. A distinct characteristic of wide-area SLR imagery is the tendency for the amplitude of radar returns to decrease at longer cross-track ranges, as for example in the central area of the right-hand image, where the ice is uniform from the central blind strip to the tanker. This is due to the fact that the back-scatter radar energy decreases substantially as the slant angle of the radar beam increases from the nadir.

Two ships can be seen on the right-hand image, a tanker and the Department of the Environment océanographie research vessel CSS Dawson whose track through the ice shows a strong return.

Figure 3 shows a map of the north-eastern tip of Ellesmere Island and a portion of Greenland. Two areas have been outlined, one showing parts of the Kennedy and Robeson Channels and the other, the Grant Ice Cap in the United States Range. These areas correspond to the SLR images shown in Figures 4 and 5 respectively. The Kennedy Channel portion of Figure 4 shows a large variety of multi-year ice structures mixed with first-year ice. In particular, note the relatively weak (dark) return from the newer ice which has formed in the large lead running horizontally on the bottom of the image. Two other characteristic effects in SLR imagery are evident in Figure 4. Because of the slanting angle of view, very high returns are obtained from the steep, rugged shore lines. Also as a result of this slanting view, there is a severe shadowing effect behind these high geographical features. In particular, note that the ice which extends into Lady Franklin Bay on the left-hand side of the image is not imaged because it is in the radar shadow of Cape Baird. Another striking example of this shadowing effect appears behind the small island off Kap Morton in the lower right-hand corner.

Fig. 3.

Map of the north-eastern tip of Ellesmere Island and portion of adjacent Greenland, showing areas imaged in Figures 4 and 5 .

The Grant Iec Cap area indicated on the map of Figure 3 is shown imaged in Figure 5. The glacial tongues extending downwards from the ice cap can be seen clearly. They have a milky appearance which is characteristic of thick glacier ice in radar images. There is also some evidence that the radar can detect subsurface structure in this type of icc, although it would not be possible to establish this without extensive ground-truth information. Loshchilov (personal communication) did obtain radar imagery from glaciers showing subsurface structure which could be associated with foliation planes. 1 he regular band structure across the image is due to uncompensated aircraft roll motion, and is to be ignored.

An operational application of this type of radar to fresh-water icc reconnaissance is shown in Figure 6. This imagery was obtained by the NASA Lewis Research Center during February 1974 over the western end of Lake Superior as part of a demonstration program for the extension of the navigable season in the St. Lawrence River and the Great Lakes. Above the SLR image is an ice chart indicating the various ice types as interpreted from the radar image. In this system, after processing of the SLR images and preparation of the interpretive ice chart, the composite was sent by facsimile to a number of radio stations for transmission by radio link to vessels on the lakes. It is interesting to note the large number of ship tracks extending from Two Harbors in the most direct line to the THIN-MED ice. As in Figure 2, open water, which is indicated by the dashed lines in the water of the ice chart, cannot be distinguished from uniform thin ice in the Point Detour region in the SLR image.

Fig. 4.

AN/APS₉₄D SLR image of first-year and multi-year ice located in Kennedy and Robeson Channels taken on 13 January 1973. Coverage across track is 50 km and 54 km along track.

Fig. 5.

AN/APS-₉₄D SLR image of glacier ice off the Grant he Cap, United States Range, Ellesmere Island, taken on. 13 January 1973. Coverage across track is 50 km and 63 km along track.

Fig. 6.

AN/APS-₉₄)C SLR image of fresh-water ice taken on 6 February 1974 over the western part of Lake Superior, together with an interpretation of the ice types, (photograph provided by Dr H. Mark,. VASA Lewis Research Center).

A number of further technical points must be borne in mind when interpreting SLR imagery. The brightness of the radar return from a given area is determined by its radar backscatter coefficient σ, which represents the energy scattered back to the radar by that area, expressed as a ratio of the radar energy illuminating it. The coefficient σ is dependent on the degree of surface roughness on the scale of a wavelength (3 cm at X-band); thus σ can be low for smooth water or smooth ice, and it can be high for ridged and broken ice surfaces. It is for this reason that young, smooth ice appears dark and indistinguishable from open water on many SLR images, while the brightest returns are produced by older ice which tends to be ridged, weathered and covered with mottled layers of crystallized snow. It is also the reason why the tonal and textural quality of the imagery obtained is very dependent upon the radar wavelength used. As reported by Ketchum and Tooma (1973), SLR imagery at wavelengths longer than 3 cm tends to lose the grey tones on the ice returns; they showed L-band (30 cm) imagery, for example, which was mostly black with only the major ice fractures and ridges showing as bright lines.

A point which has been noted Jirberg and others (1973) is the fact that with fresh-water ice, significant radar returns can be obtained from the ice-water interface as well as from the air-ice interface, due to penetration of the ice by the microwave radiation. Thus it is that fresh-water ice types with a rough (not mushy) bottom profile such as consolidated brash, can produce bright returns on the imagery.

The backscatter coefficient σ and thus the nature of the imagery, is sensitive to the polarization of the illuminating wave, as was demonstrated by Ketchum and Tooma (1973). The AN/APS₉₄ system which produced the images shown in Figures 2 through 6 is horizontally polarized for both the transmitted and received wave (H-H). An example is shown in Figure 7 of dual-polarized imagery, taken in the Baie des Chaleurs of the Gulf of St. Lawrence with a Westinghouse 35 GHz system for the Canadian Ministry of Transport. This illustrates the substantial differences in image tone which can be obtained between H-H and H-V system modes. Note the greater visibility of an ice fracture running near the shore line (sec arrow) on the H-H image; also more visible on this image is the structure of the newer grey ice in the upper left. On the cross-polarized image, however, note the improved visibility of the structure of the older grey-white ice in the lower right.

Fig. 7.

AN/APQ-₉₇ Westinghouse dual polarized 35 GHz SLR image taken over Baie des Chaleurs, Gulf of St. Lawrence on 1 March 1969. The top image is horizontally polarized (H-H) and the bottom image is cross polarized (H-V).

A further point regarding the imagery of Figure 7 is that it is taken with a system having a much shorter wavelength (I cm) than that used for Figures 2 to 6. The structure of younger ice, often not visible with X-band or longer wavelength systems, is clearly visible with tin’s shorter wavelength system.

The second, more complex and costly type of SLR is the coherent synthetic-aperture system. Here, the along-track beam-width of the antenna does not have to be very narrow lo achieve high image resolution; in fact, the opposite is true. Consider a point on the ground being illuminated by many successive radar pulses as the antenna beam passes over it. If the phases of the radar reflections from that point are recorded as the antenna beam traverses the point, from one edge of the beam to the other, and if that phase history is subsequently processed in a suitable manner, the point can be imaged with very high resolution. Theory shows that the longer a point remains in the beam, the better it can be resolved in the resulting image; thus, theoretically, the wider the along-track antenna beam-width, the higher the image resolution. Furthermore, theory predicts that this high resolution is independent of target distance, so that equally high resolution is obtainable across the swath width of the image. Despite the theory, however, although very high resolution is indeed achieved with synthetic aperture SLR systems, this resolution is limited in practice by unavoidable perturbations in the flight path of the aircraft which restrict the time over which the signal phase history can be processed to produce the image.

In Figure 8 is shown a synthetic aperture SLR image produced by the Goodyear GEM-system employing the AN/APD-102 radar, when it was flown on 10 March 1974 over the Northumberland Strait between Pictou, Nova Scotia and Murray Harbour, Prince Edward Island. The resolution of this image is about 15 m and is constant across-track. All ice in this image is first-year sea ice having a concentration of 9/10 as determined from visual observations. The ice floes and the brash, including refrozen floes contained within the larger floes, are clearly visible. In this case, the dark edges along the shores represent shore leads. In the upper left-hand corner some shore fast ice is visible with an open-water lead intruding into the bay. Obvious advantages of this imagery over real-aperture imagery are the higher resolution and the uniformity of resolution and contrast across-track.

Fig. 8.

AN/APD-₁₀₂ B Goodyear X-hand synthetic aperture SLR image taken over Northumberland Strait on 10 March 1974 at an altitude of 11 300 m. Coverage is 36 km across-track arid 45 km along-track, (photograph courtesy of Aero Service Corp.).

As an example of the application of SLR imaging systems to ice dynamics studies, we should note that the Arkticheskiy i Antarkticheskiy Nauchno-Issledovatel’skiy Institut in Leningrad has used a synthetic-aperture SLR, the RLS BO “Toros” system, to distinguish between different ice types and to determine ice drift (Gorbunov and Losev, 1974). The resolution of this radar is approximately 15 m. The ice types they can identify comprise newly-formed ice, grey ice, first year, and multi-year ice. The concentration of ice on the sea surface can be determined from the imagery, including a qualitative description of the ice surface (degree of hummocking). The amount, dimensions, distributions and orientation of channels and cracks can also be obtained. For the ice drift study which was carried out in the northern sea routes, the side-looking radar was mounted in an Antanov (AN-24) aircraft. Ice drifts for a two-day period were found to be between 25 to 50 km away from the shore and between 2.5 and 7 km near the shore. Results for the larger drift velocities were in error by only 1 to 2%, as compared to 10 to 20% for the lower velocities.

Snow detection with SLR

From the various examples of imagery presented above, it is clear that SLR images tend not to show reflections from snow surfaces, showing instead the stronger reflections from the surface underneath the snow. However, it has been observed by Waite and MacDonald that old, perennial snow is sometimes visible in SLR imagery. This is presumed to be due to volume scattering from the inhomogeneities which exist in old snow.

Future SLR applications

High-resolution SLR imagery is now becoming increasingly available from aircraft-borne coherent systems, and there is no doubt that these systems will contribute a great deal more to the glaciologist’s knowledge through such experiments as AIDJEX, POLEX, etc. It is also probable that the most significant glaciological applications of coherent SLR imaging systems are yet to come—with satellite-borne systems providing continuous, high-resolution, synoptic coverage of large geographic areas. The imagery from these satellite systems will be comparable in resolution quality to that from existing airborne SLR systems; it will also compare favourably with the photographic imagery available from existing satellite systems, with the added advantages of independence of cloud cover and light conditions.

Calibration of SLR imagery—associated sensors

Although it has been shown that imaging systems are useful to the glaciologist, at least for qualitative purposes, it is also clear that definitive interpretation of this imagery will not be possible until more work has been done to relate the imagery to the physical properties of the ice. In this regard it must be noted that only a very small proportion of the ice imagery being collected is associated with extensive ground-truth information about the ice being imaged—a situation which is likely to continue, considering the vast geographical areas involved.

In the absence of ground truth, SLR imagery may be “calibrated” for more quantitative interpretation by the use of one or more associated sensors gathering information about specific ice properties. A property which could be used in this way is thickness, and radar sensors for remotely measuring ice thickness will be discussed in a separate section of this paper. Cither associated sensors which can be used are optical and infra-red cameras and line-scan devices, laser profilometers, microwave radiometers and microwave scatterometers. Of these, the microwave radiometer and the microwave scatterometer offer some of the same operational advantages as radar (all-weather capability and independence of light level), and at the same time measure properties which complement the SLR information (Reference Gloersen, Gloersen, Nordberg, Schmugge, Wilheit and CampbellGloersen and others, 1973; Reference Gloersen and GloersenCampbell and others, 1974; Meeks and others, in press). We will not deal here with the microwave radiometer, although it should be noted that extensive co-operative experiments have been carried out by the U.S.A. and (he U.S.S.R. to relate and compare ice data from both radiometers and radars (Ramseier and others, 1974; Gloersen and others, 1974[a], [b]).

Introduction

That microwave radar can penetrate ice and snow is a fact that has long been suspected by aircraft pilots who have found their radar altimeters registering erroneously large altitude readings when flying over large glaciers. The radar altimeter is, in fact, a prototype of the probing type of radar sensor used to study the internal structure of ice and snow. Here, unlike the SLR imaging system, it is not the coverage area which is important, but the ability to measure range, i.e. depth, accurately.

A basic requirement for this type of radar sensor is that it be designed for good penetration of the ice and/or snow. Some knowledge of the electrical properties of these materials is essential for the users of these sensors, and of particular interest are the velocity and attenuation of radio waves propagating through them.

Electrical properties of ice and snow

The most important electrical property influencing the propagation characteristics of a medium is the relative permittivity:

where ε_r is commonly known as the dielectric constant of the medium.

The velocity of propagation is given by

where c = 3 x 10⁸ m/s is the velocity of electromagnetic waves in vacuum.

Thus, the time taken for propagation through depth h is given by

For our purposes, ε_r’ and ε_r" are variables which are dependent on the type of ice or snow, on temperature, and on the radio-frequency used. Attenuation of radio waves, usually expressed in dB loss per meter, is related directly to the loss tangent which is given by

Radar reflections occur whenever there is an abrupt change in the value of, ε_r, i.e. where significant changes in ε_r take place within the distance of a wavelength of the radio wave. It is when such changes in ε_r take place at interfaces between air, snow, ice, and water that radar reflections can be expected from such interfaces.

Measurements of the permittivity of fresh-water and glacier ice have been made over wide ranges of temperature at frequencies extending upwards from very low frequencies to as high as 96.5 GHz, (Reference Lamb and TurneyLamb and Turncy, 1949; Reference CummingCumming, 1952; Von Reference Von HippelHippel, 1954; Reference YoshinoYoshino, 1961; Perry and Straiton, 1973; Vant and others, 1974; private communication from W. B. Westphal). The loss tangent is very small at microwave frequencies, with values less than 2 X 10^-3 at all temperatures lower than 0°C and all frequencies greater than 1 GHz, but rising substantially at frequencies less than 200 MHz. The real component of permittivity, ε_r is found to be virtually independent of temperature and of frequency at frequencies above 100 MHz, with measured values ranging between 2.9 and 3.2, depending on the author.

Using a value for real relative permittivity of 3.14 (a good statistical average), we arrive at a velocity of propagation in fresh-water and glacier ice of 1.69 X 10⁸ m/s, a value which can be considered to be independent of frequency and temperature. From the loss-tangent data, it is possible to predict that over the frequency and temperature ranges of interest here, attenuation in fresh-water and glacier ice will be less than 0.05 dB/m at 1 GHz and —10°C, increasing approximately proportional to the square of the frequency, and increasing by about a factor of 2 with increasing temperature over the range —40°C to 0°C .

The situation with snow and sea ice is more complex, reflecting the fact that these are complex materials occurring naturally in a number of types and in an endless variety of composites of these types. Permittivity measurements on snow have been reported by Gumming (1952) and Reference EvansEvans (1965), and have been reviewed by Reference Vickers, Vickers, Heighway, Gedney, Santeford and SmithVickers and Rose (1972) who have reduced the data to obtain approximations for attenuation and propagation velocity. For dry snow, these properties are found to be relatively independent of temperature for temperatures below 0ºC. But they are very dependent on snow density, with freshly fallen snow giving an attenuation per unit length which is an order of magnitude less than that for fresh-water and glacier ice, and with a similar frequency dependence. This attenuation increases with increasing snow density, until in very hard-packed snow it approximates that for fresh-water or glacier ice. Similarly, velocity of propagation in dry snow decreases from very close to the free-space value for freshly-fallen, light snow to approximately the value for fresh-water or glacier ice when the snow is very hard-packed. Clearly, therefore, any radar sensor system for snow depth studies will require some associated means of establishing the snow density if the radar data are to be correctly interpreted.

The presence of water on or in the snow or ice, even in small quantities, seriously complicates the situation. An estimate by Reference RoyerRoyer (1973) gives an attenuation rate in water of 70 dB per meter for a frequency of 1 GHz and temperature of 0°C. It varies rapidly with frequency, and is greater by two orders of magnitude at 10 GHz. Clearly, the attenuation through wet snow or ice at microwave frequencies can be expected to be much higher than when water is not present.

Sea ice is the most difficult of all lo characterize electrically. The presence of brine and salts in the ice causes relatively high attenuation at radio-frequencies, and problems arise from the fact that there are many different kinds of sea ice with an endless variety of concentrations and distributions of brine and salts. Nevertheless, a number of authors have attempted to measure losses in sea ice at various frequencies, and a summary of their findings is given in Figure 9. It should be noted that the points in Figure 9 represent measurements with both naturally occurring ice and laboratory ice; however, all the measurements were made using samples within the narrow ranges of temperatures and salinity shown.

Fig. 9.

Summary of loss data for ice of a standard salinity and at a standard temperature (provided by M. Vant).

A recent study by Vant and others (1974) has dealt in more detail with the electrical properties of different categories and types of sea ice. In their measurements, they have separated sea ice into the two categories: first-year ice and multi-year ice. They have further separated first-year ice into two types: frazil and columnar. Multi-year ice in their study consisted of recrystallized columnar ice. At a fixed frequency of 10 GHz the permittivity of these ice types was explored as a function of temperature. The dielectric constant ε_r’ of both first-year types was found to approximate that for fresh-water ice at temperatures below —30º C, but to rise dramatically as the temperature increases towards 0º C. For multi-year ice, a similar temperature dependence was observed for ε_r’, but the value at low temperatures was found to be substantially less than that for other ice types. Dielectric loss ε_r" was found to vary considerably between sea-ice types, being very high for first-year types and substantially less for multi-year types.

The above observations of electrical properties for first-year sea ice are generally consistent with the fact that first-year ice contains brine pockets which freeze only at very low temperatures. The observations for multi-year ice are consistent with the known fact that these brine pockets drain out gradually and are replaced by air, with the result that multi-year ice is a complex and time-varying combination of ice, air, and brine.

Radar probes for floating ice

There is a need for ice depth sounders which will quickly and effectively provide ground truth for SLR imagery used for large-area ice reconnaissance. Such sounders are also expected to be invaluable for ice growth studies. In addition, there are major operational requirements for such sounders to make possible the safe and effective movement of vehicles on ice and of ships through ice. These requirements exist for both fresh-water ice and sea ice.

The first challenge in designing radars for depth sounding of thin floating ice is to obtain the very high resolution required—typically a few centimeters. In fact, there have been significant advances in radar electronics over the past decade which make possible the attainment of very high resolution for accurate depth-sensing systems. These advances have been in the development of wide-band techniques for modulating radar transmissions, and we will discuss the two basic approaches which are being used:

• (a) wide-band frequency modulation of the carrier wave amplitude (FM/CW systems), and

• (b) very short time-duration modulation of the radio energy (short-pulse and impulse systems).

Early experiments with a simple FM/CW radar at 10 GHz were reported by Christoffersen (1970), in which the radar was both surface-borne and airborne. These experiments confirmed that X-band is a poor choice of wavelength for sea ice. Another much lower frequency FM/CW system adapted directly from an aircraft altimeter system in the 420-470 MHz band is being used commercially for sea-ice sounding by Geophysical Service incorporated; it is understood (personal communication) to be capable of measuring thicknesses greater than about 1 m and is intended for mounting on surface vehicles. Using information obtained from bore-hole samples, the system is calibrated for the wide range of propagation velocities to be expected in sea ice. An FM/CW radar at X-band, designed specifically for thin freshwater ice and snow has been described by Page and others (1973) and in an improved version suitable for airborne use by Chudobiak and others (1974). Typical results obtained with this system mounted on a sled moving across the Ottawa River are shown in Figure 10a. These plots were produced by the digital computer processing of a magnetic tape record (Venier and others, in press). The same radar was used to monitor ice growth in a laboratory cold room as a function of time, and the results were compared with thickness-gauge measurements. As shown in Figure 10b, these results indicate that the radar measured ice sheets as thin as 15 cm to an accuracy of the order of ± 1 cm.

Fig. 10.

(a) Profile of ice thickness obtained with FM/CW radar mounted on a moving sled, {b) Results of FM/CW radar ice thickness growth study.

A pulsed radar system using very short pulse (1.3 ns) modulation of a carrier wave at 2.7 GHz has been reported by Vickers and others (1974) for fresh-water floating-ice studies from a low-flying airborne platform. In trials over Lake St. Clair and Lake Superior, supported by auger teams on the ice, thicknesses of single, unbroken ice layers greater than 10 cm were measured to an accuracy of 2 cm.

Experiments with a much lower-frequency pulsed radar have been reported by Bogorod-skiy and TripoPnikov (1974) for airborne sounding of sea-ice thickness from an altitude of 100 m. A pulse length of 50 ns was used with a carrier frequency of 100 MHz. Good upper-and bottom-side returns were obtained with young-ice thicknesses up to about 1.5 m and old-ice thicknesses up to several meters. Despite the relatively long pulse length which undoubtedly limited the resolution of the thickness measurements, the feasibility of using an airborne radar to obtain both upper- and lower-surface returns from sea ice was demonstrated. The authors also carefully point out the problem of interpreting such sea-ice thickness measurements without accurate knowledge of the temperature and classification of the ice being sounded.

In the impulse system, a relatively new type of time-domain radar, a narrow burst of energy at baseband (d.c.) is used to produce very wide-band transmissions. Such a system is being used commercially by Geophysical Survey Systems, Incorporated, and its use for ice sounding from surface vehicles has been described by Reference BertramBertram and others (1972) and Reference GoodmanGoodman (1974). With a 5 ns pulse, a wide range of sea-ice depths have been sounded using bore holes to calibrate the system for temperature and ice type.

A very narrow sub-nanosecond base-band pulse is the basis of an impulse radar system developed at the Communications Research Centre in Ottawa and tested in a co-operative ice studies program with the Canadian Department of the Environment. Designed for fresh-water ice studies, it selects and uses a frequency band of approximately 1 GHz of the impulse energy at X-band. A block diagram of this very simple system is shown in Figure 11a.

This radar was tested in the field during February and March 1974. The output display for this system mounted on a sled on the Ottawa River is shown in Figure 11b and 11c. The smaller reflection in Figure 11b was obtained from a snow-free, ice-air interface while the larger reflection comes from the ice-water interface; this is the normal situation with bare, dry ice. The reflections shown in Figure 11c were obtained in an area covered by snow containing a thin ice crust sandwiched between two layers of snow, as illustrated by the schematic vertical snow and ice profile.

In Figure 1 id is shown a comparison of ice thickness as determined by bore-hole drilling and by the impulse radar mounted on a surface vehicle during trials on the St. Lawrence River near Wolfe Island. The continuous strip recording shown in Figure lie was obtained during these trials on the St. Lawrence River. Below this record is plotted the ice thickness as read from this strip recording; also plotted are the measured ice thickness at five bore holes. The above results have shown the system to be capable of measuring fresh-water ice thicknesses as small as 15 cm with an accuracy of better than ±0.5 cm, when mounted on surface vehicles.

Fig. 11.

(a) Diagram of X-band nanosecond impulse radar, (b) Impulse radar return for 30 cm of bare river ice (1 ns/division horizontally), (c) Impulse radar return for river ice with snow cover as follows: A-B soft snow, B thin ice crust, B-C coarse granular snow, C-D ice ice (5 ns/division horizontally), (d) Comparison of ice thickness as measured using impulse radar and bore-hole drilling, (e) (i) Continuous strip profile of St. Lawrence River ice over a distance of 2 000 m. Vertical scale: ice thickness (0.16 m/division); horizontal scale: distance (100 m/division), (ii) Thickness interpreted from (i) and plotted, with five bore-hole measurements plotted for comparison.

Fig. 12.

Impulse radar mounted on (a) truck, (b) hovercraft, (c) helicopter, (d) DG-3 aircraft.

During these trials, prototypes of this radar were mounted on various vehicles, as shown in Figure 12. The radar returns shown in Figure 13a were obtained with the system mounted in a helicopter (lying at an altitude of 30 m and at a velocity of 80 knots (40 m s^-1) over the St. Lawrence test area near Wolfe Island. Ice thickness of approximately 20 cm was measured to within ±10% of the value obtained by ground-truth measurements. Due to the fact that in this case the temperature was near the melting point and the surface of the ice was moist, the top-side ice return is much greater than the bottom-side return. The return shown in Figure 13b was obtained from a DG-₃ aircraft at an altitude of 60 m and velocity of 120 knots (60 m s^-1) over Lake Nipissing. Again, the upper surface return is high because of surface moisture; in fact, this flight was carried out after a prolonged warm period during which daytime temperatures were above freezing.

Fig. 13.

Returns obtained with impulse radar mounted in (a) helicopter at altitude of 30 m, (b) DC-3 aircraft at altitude of 60 m.

Radar sounding of snow

The possibility of applying microwave radar designed with high range resolution to the sounding of snow depth is indicated by the results from X-band impulse radar shown in Figure lie. This possibility was studied and reported by Vickers and Rose (1972) who used a short-pulse radar at 2.7 GHz to explore snow-pack stratigraphy with high resolution. They found that good radar returns could he obtained from internal structure in a snow pack when it contained no free water. They indicated the need to obtain snow density information to calibrate the radar. Their conclusion was that such a radar, with sufficiently high transmitter power, could be used operationally for monitoring snow-pack hydrology from either a surface or an airborne platform.