- Email: [email protected]
Remote sensing of Mars by a flexible orbiting radar
Pergamon www.elsevier.nl/locatelasr
Adv. Space Res. Vol. 23, No. 11, pp. 1929-1932.1999 Q 1999 COSPAR. Published by Elsevier Science Ltd. All tights reserved
Printed in Great Britain 0273-l 177i99$20.00+ 0.00 PII: SO273-1177(99)00274-4
REMOTE SENSING OF MARS BY A FLEXIBLE ORBITING RADAR S. P. Kingsley(‘), H. St.C. K. Alleyne(*), M. A. Balikhin’2’, G. Junkin(“, S. D. Green(‘), S. N. Walker”‘, and K. H. Yearby’*’
“’ Department of Electronic and Electrical Engineering, University ofshefield, UK. ‘2’Department of Automatic Control and @sterns Engineering, University of Sheffield, UK.
ABSTRACT
A satellite based radar system to investigate the planet Mars can be expected to carry out several different roles, acting as an altimeter, a ground probing radar, and an ionospheric topside sounder. These roles place conflicting requirements on the radar design, in addition to the limitations of mass, power, telemetry and antenna deployment. These conflicting needs can be resolved through the use of a single, fully programmable radar instrument. ‘Ibis paper discusses the concept of such an instrument. 0 1999 COSPAR. Published by Elsevier Science Ltd. INTRODUCTION To date, all investigations atmosphere, the planetary allow, for the first time, a study will be to determine
of the planet Mars has restricted themselves to studying the planetary plasma environment, the surface or global interior of the planet. The use of an orbiting ground penetrating radar (GPR) will study of the geological layering on the planet to a depth of a few kilometres. The main goal of this if layers of liquid water still exist on Mars below the planetary surface.
There is much evidence on the surface of Mars that in its geologic past the climate was wetter. During its history, the climate changed and the water seeped underground. As the planet cooled, the water near the surface froze, forming a thick cryosphere. Occasionally, this icy layer was breached and catastrophic flooding occurred (Carr, 1996; Masursky et al., 1977). Models of the Martian subsurface show that the frozen cryosphere varies in depth between 2-6km (Clifford, 1993) and below it a possibility of finding liquid water. The actual depth at which liquid water may be found depends upon the volume of water Mars possessed, the amount subsequently lost into space and the depth of the crust. The Martian ionosphere has been studied using either radio occultation methods or from the Viking landers. Its density and height vary on a diurnal cycle with the dayside density peak of -2 x 105cme3at 125km (Hanson et al., 1977) and nightside peak densities of -5 x 103cm” at 150km (Zhang et al., 1990). By using the radar as a top side sounder, a full profile of the top side if the ionosphere may be measured. Radar signals that penetrate the ionosphere will be subject to delays and phase distortions. A knowledge of these distortions is required in order to interpret the GPR and altimeter signals correctly. The design, a response to the Mars Express A0 issued by ESA in December 1997 is based upon the radar instrument specified as part of the straw man payload. The main constraints/objectives are a power of 6OW, telemetry rate 285Mbit/day, and a ground penetration depth of 2km with a resolution of Sm. Although not selected, the design contains some novel concepts and is therefore worth discussing. SYSTEM DESIGN Radar design has to be a balance between several opposing factors. The main constraint is that between the range resolution and ground penetration. Since the range resolution of a radar is inversely proportional to the bandwidth of the radiated signal, the larger the bandwidth, the greater the range resolution. However, use of high bandwidth waveforms implies high frequency carrier waves that do not provide adequate ground penetration. For example, with a frequency sweep of 515MHz (bandwidth 1929
1930
S. P. Kingsley er al.
1OMHz) the expected ground penetration would be around 2km with a resolution of 5m whilst a 1MHz signal of bandwidth 1OOkHzwould penetrate to =4km but only have a range resolution of 500m. Waveform Modulation The choice of radar waveform places constraints on the power that the transmitter will radiate, the spatial resolution and the number of signals that may be integrated to achieve a better signal to noise ratio. Table 1. compares the mean power based upon pulse duration and wave repetition frequency (WRF) for various pulse types (assuming a peak power of 60W).
Modulation Pulse Carrier-free Chirp Freq. Sweep
Table 1: A Comparison of Four Types of Waveform Modulation Duration Mean Power(W) WRF (Hz) O.lus 2000 0.012 O.lps 2000 0.012 0.12ms 2000 14.4 2ms 192 23
4
Each waveform has its pros and cons. The first three allow a high WRF to be used. Around 30 return signals may be integrated before the echoes begin to de-correlate, improving in the signal to noise ratio. However, the mean transmitted power for pulse and carrier-free systems is exceptionally low. The frequency sweeping modulation provides a higher transmitted power at the expense of a low WRF. For a GPR, surface penetration is of prime importance. Both pulse and chirp modulation require a high carrier frequency with respect to the signal bandwidth. A 5m range resolution requires a 1OMHz bandwidth, implying a carrier of 100MHz. At such frequencies, propagation losses are prohibitive (Ori and Ogliani, 1996). To achieve good penetration, frequencies <2OMHz should be used implying a lower bandwidth and reduced range resolution. The use of high frequencies can be avoided by using carrier free (impulsive) waveforms. However, this is not an option on mean power considerations unless extremely high voltage pulses are used. This is not consistent with high reliability and low EM1 effects. Ionospheric distortions require some correction for proper interpretation of the data. For chirp systems, compensation requires the computation of a real time, full band replica pulse. Pulse and frequency sweep waveforms may be corrected either by predistorting the transmitted signal or by the processing of the received signal. Pulse systems require (pre)processing of only one frequency. Frequency sweep systems require each frequency to be given its own predistortion. The best waveform appears to be a frequency sweeping system, transmitting a carrier wave over a IOMHz range from 51SMHz providing the low WRF does not pose a problem. This type of radar involves difficulties in antenna design as it has to operate over a large bandwidth. This waveform, generated by a direct digital synthesiser (DDS), consists of a staircase of 10,000 steps spaced 1kHz apart. The use of a dual DDS provides a useful way to drive two mixers with a 90” phase shift between them for complex de-ramping. The waveform can be predistorted in both phase and sweep rate for each frequency across the swept bandwidth, so compensating for phase adjustments and relative changes in the doppler shift across the band. The waveform at each frequency may be further predistorted to take into account the effects of the ionosphere. Antenna The antenna poses one of the largest design problems for the whole radar. It has to operate over a wide bandwidth, typically SISMHz. Multiple deployments may be necessary if the satellite has to be manoeuvred between differing orbits or to allow reorientation of the high gain antenna for data transmission. Retraction before such manoeuvres reduces the probability of setting unwanted vibrations in the satellite due to the inertia of the elements. The best orientation of the antenna is perpendicular to the flight path as this configuration may be used to reduce the clutter. The antenna should also be perpendicular to the solar panels to minimise any interference between the two electrical systems. The resulting antenna design consists of two tubular 15m elements. The metallic conductor is embedded in a carbon reinforced polyester support structure. The support is manufactured such that it may be flattened into a ribbon for storage but when released the inbuilt mechanical stresses cause the antenna to take on its tubular cross-section, forming a rigid structure. Modelling of such an antenna shows a fairly constant impedance match of 300 Q across the proposed bandwidth.
A Flexible Orbiting Radar
1931
The proposed radar transmits a series of frequency sweeps from 5- 15MHz and compresses them in the receiver using analogue de-ramping. The resulting beat-frequency signal is then analysed using FFT techniques (Skolnik, 1990). As signal frequency rises with range, low frequency components correspond to close range targets. The signal processing chain consists of a double FFT on the de-ramped signal. The first is a 1024 point complex FFT (192 per second), the second analysing 192 points (1024 per second). The sampling bandwidth of 512kHz allows the surface to be tracked to within -15km. These rates of computation assume that the position of the surface is known. If the surface position is lost, the radar would have to determine its range automatically. The surface could lie anywhere within the 2ms sweep. This is an extremely large processing task as the surface could lie in any one of 20,000 range gates. Depending upon the computational efficiency, it may be possible to increase the sampling bandwidth by a factor 2 or 4. This would then ease the task of surface tracking considerably as the surface topology does not vary on scale sizes of 30-60km. By employing a system for data selection, only the most significant range gates need be returned to Earth. This should make better use of the low telemetry allocation. If, however, the onboard processing cannot be performed correctly, whole sweeps will have to be returned in the telemetry for use in an Earth based processing system. This would result in a reduction in the temporal resolution of the data. and Operations Overall control of the radar is carried out by a series of commands‘fiom Earth. These commands instruct the control system to either configure the radar and run a series of predefined instructions (a macro) or uplink changes or new instruction macros into memory. If the commands received are corrupted, a default macro designed to work with minimum system requirements may be invoked in order to obtain some scientific return. The macros enable the radar to operate autonomously, having the ability to estimate and update values for certain operational parameters depending upon the chosen mode of operation but flexible enough to respond to a request from Earth. There are four basic operational modes foreseen for the radar:- altimeter/GPR, GPR/altimeter, surface detection and ionospheric sounder. Altimeter/GPR Mode. Using its full lOMHz bandwidth, the radar will operate primarily as an altimeter, but will also have full GPR functionality. The basic range resolution would be around 15m. The exact range may be determined by interpolation between range gates, with an uncertainty of 0.8m (assuming an SNR of 22 dB and a smooth surface). The GPR system would have a range resolution of 5m and a penetration depth of around 2km. GPR/Altimeter Mode. The primary target of this mode is for the radar to function as a GPR. Greater surface penetration is achieved using a lower bandwidth. Initially a frequency range 5- 1OMHz would be employed, allowing a depth penetration of around 2.2km and range resolution of 10m. If necessary, the DDS may be reprogrammed to use a lower frequency range if this penetration depth proves inadequate. The altimeter function still provides information but at a reduced resolution (-30m). Surface Detection Mode. Under normal conditions, the distance to the planet surface will be tracked automatically. Its approximate distance may be determined from the orbital dynamics of the satellite and this may be used as an estimate to set the WRF. If, however, the range is unknown or lock is lost the radar may automatically switch itself into a low WRF mode to get an unambiguous surface range before returning to its previous operational mode. Ionospheric Sounder Mode. During the daytime, the satellite will be orbiting above a dense ionosphere. Under such conditions, the operation frequency range of the radar would be changed to 0. I-5MHz to allow the topside of the ionosphere to be probed. Operational Modes. Actual operations will most likely involve combinations of the basic operational modes and depend upon whether the instrument is investigating the dayside or nightside of the planet. The dayside ionospheric plasma frequency (5MHz) restricts the radar to operating as a top side sounder. On the night side the plasma frequency is -0.3MHz and so both altimeter and GPR modes are equally usable. An example operational mode for the night side would involve using the radar as a GPR or altimeter with the occasional switching to the ionospheric sounder mode to probe the ionosphere. This would then allow corrections to the degree of phase distortion to be carried out. CLUTTER REDUCTION Ground clutter poses a severe problem with the radar. At any given instant the voltage and phase of the clutter signal appearing at the terminals of the antenna are the result of the contributions of a very large number of scattering elements - mainly
1932
S. P. Kingsley et al.
boulders on the surface and discontinuities in the topography. Because of this summation over a large number of signal contributions, surface clutter is inherently noise-like in its characteristics. The following methods will be used to reduce the clutter content of the return signal. As the antenna is deployed perpendicular to the direction of flight, any clutter arising from sources along the track will contain a doppler shift in addition to that from the eccentric orbit of the spacecraft. As the spacecraft moves with respect to the clutter sources, the doppler shift will change. This will affect the rate of change of frequency of the return signal with respect to the transmitted waveform. By processing from the mixer using a double FFT any doppler shifted signals such as those generated by clutter will be smeared out in frequency space whilst non doppler shifted signals should produce strong peaks. For clutter arising from cross track sources, transmitted and return signals propagate in the direction of the antenna’s null cone. The resulting antenna gain will be extremely low and signals will be lost in the noise of the system. The antenna is likely to be angled down towards the surface to enhance this effect. The maximum correlation time for received signal is proportional to lidoppler bandwidth. For the proposed radar design, this period is of the order of l4ms. If integration times greater than these are used, the effect is to lose SNR at the edges of the resolution cells and reduce the size of the clutter. Clutter only poses a problem from sources either very close-in or at very long range. Most short and medium range clutter does not enter the receiver because it is switched off for the 2ms (+300km of range) of the next transmission cycle. Clutter falling outside the transmission cycle is severely attenuated by range eclipsing losses (i.e. only part of the signal from a particular range enters the receiver because of transmit-receive switching). The lack of a complete FM sweep means the clutter signal fails to process properly and is attenuated. Close-in clutter signals signals from the horizon echoes, i.e. at the same attenuated as described very oblique angles.
are minimised of Mars come time as echoes above, but also
as described above and come from relatively small annuli. Very long range clutter from much larger areas. They can avoid eclipsing by arriving as ‘second time around’ from the next transmit-receive cycle. However, these long range echoes are not only by increased propagation losses and the inefficiency of the backscattering process at
Once a number of scans of the same region exist, it should be possible to use along track Synthetic Aperture Radar processing techniques, improving the signals and minimising the clutter content. The efficiency of each of these various methods needs to be assessed both via simulations their performances.
and instrument testing to evaluate
SUMMARY A design for an orbiting ground penetratin, 0 radar has been presented. This radar overcomes the conflicting need of ground penetration, high spatial resolution and peak power by the use of a frequency sweeping system with flexible programming and an optimised antenna design. Such a device should be capable of detecting liquid water at a depth of several kilometres. REFERENCES Carr. Michael H., Channels and valleys on Mars: cold climate features formed as a result of a thickening cryosphere, Planet. sp.‘sci. 44, I41 I, 1996. Clifford, Stephen M., A Model for the Hydrodynamic and Climactic Behavior of Water on Mars, J. Geophys. Res E 98, 10973, 1993. Hanson, W. B., S. Sanatani, D. R. Zuccaro, The Martian ionosphere as observed by the Viking retarding potential analysers, J. Geophys. Res A 82,435 1, 1977. Masursky, Harold, .I. M. Boyce, A. L. Dial, G. G. Scaber, and M. E. Strobell, Classification and time formation of Martian Channels based on Viking data, J. Geophys. Res A 82,4016, 1977. Ori, G. G., and F. Ogliani, Potentiality of the ground penetrating radar for the analysis of the stratigraphy and sedimentology of Mars, Planet. SpSci 44, 1303, 1996. Skolnik, M. I., Radar Handbook, Pub. McGrawHill, 1990. Zhang, M. H. G.. J. G. Luhmann, and A. J. Kilore, An observational study of the night-side ionospheres of Venus and Mars with radio occultation methods, J. G’eophys. Res A 95, 17095, 1990.
Adv. Space Res. Vol. 23, No. 11, pp. 1929-1932.1999 Q 1999 COSPAR. Published by Elsevier Science Ltd. All tights reserved
Printed in Great Britain 0273-l 177i99$20.00+ 0.00 PII: SO273-1177(99)00274-4
REMOTE SENSING OF MARS BY A FLEXIBLE ORBITING RADAR S. P. Kingsley(‘), H. St.C. K. Alleyne(*), M. A. Balikhin’2’, G. Junkin(“, S. D. Green(‘), S. N. Walker”‘, and K. H. Yearby’*’
“’ Department of Electronic and Electrical Engineering, University ofshefield, UK. ‘2’Department of Automatic Control and @sterns Engineering, University of Sheffield, UK.
ABSTRACT
A satellite based radar system to investigate the planet Mars can be expected to carry out several different roles, acting as an altimeter, a ground probing radar, and an ionospheric topside sounder. These roles place conflicting requirements on the radar design, in addition to the limitations of mass, power, telemetry and antenna deployment. These conflicting needs can be resolved through the use of a single, fully programmable radar instrument. ‘Ibis paper discusses the concept of such an instrument. 0 1999 COSPAR. Published by Elsevier Science Ltd. INTRODUCTION To date, all investigations atmosphere, the planetary allow, for the first time, a study will be to determine
of the planet Mars has restricted themselves to studying the planetary plasma environment, the surface or global interior of the planet. The use of an orbiting ground penetrating radar (GPR) will study of the geological layering on the planet to a depth of a few kilometres. The main goal of this if layers of liquid water still exist on Mars below the planetary surface.
There is much evidence on the surface of Mars that in its geologic past the climate was wetter. During its history, the climate changed and the water seeped underground. As the planet cooled, the water near the surface froze, forming a thick cryosphere. Occasionally, this icy layer was breached and catastrophic flooding occurred (Carr, 1996; Masursky et al., 1977). Models of the Martian subsurface show that the frozen cryosphere varies in depth between 2-6km (Clifford, 1993) and below it a possibility of finding liquid water. The actual depth at which liquid water may be found depends upon the volume of water Mars possessed, the amount subsequently lost into space and the depth of the crust. The Martian ionosphere has been studied using either radio occultation methods or from the Viking landers. Its density and height vary on a diurnal cycle with the dayside density peak of -2 x 105cme3at 125km (Hanson et al., 1977) and nightside peak densities of -5 x 103cm” at 150km (Zhang et al., 1990). By using the radar as a top side sounder, a full profile of the top side if the ionosphere may be measured. Radar signals that penetrate the ionosphere will be subject to delays and phase distortions. A knowledge of these distortions is required in order to interpret the GPR and altimeter signals correctly. The design, a response to the Mars Express A0 issued by ESA in December 1997 is based upon the radar instrument specified as part of the straw man payload. The main constraints/objectives are a power of 6OW, telemetry rate 285Mbit/day, and a ground penetration depth of 2km with a resolution of Sm. Although not selected, the design contains some novel concepts and is therefore worth discussing. SYSTEM DESIGN Radar design has to be a balance between several opposing factors. The main constraint is that between the range resolution and ground penetration. Since the range resolution of a radar is inversely proportional to the bandwidth of the radiated signal, the larger the bandwidth, the greater the range resolution. However, use of high bandwidth waveforms implies high frequency carrier waves that do not provide adequate ground penetration. For example, with a frequency sweep of 515MHz (bandwidth 1929
1930
S. P. Kingsley er al.
1OMHz) the expected ground penetration would be around 2km with a resolution of 5m whilst a 1MHz signal of bandwidth 1OOkHzwould penetrate to =4km but only have a range resolution of 500m. Waveform Modulation The choice of radar waveform places constraints on the power that the transmitter will radiate, the spatial resolution and the number of signals that may be integrated to achieve a better signal to noise ratio. Table 1. compares the mean power based upon pulse duration and wave repetition frequency (WRF) for various pulse types (assuming a peak power of 60W).
Modulation Pulse Carrier-free Chirp Freq. Sweep
Table 1: A Comparison of Four Types of Waveform Modulation Duration Mean Power(W) WRF (Hz) O.lus 2000 0.012 O.lps 2000 0.012 0.12ms 2000 14.4 2ms 192 23
4
Each waveform has its pros and cons. The first three allow a high WRF to be used. Around 30 return signals may be integrated before the echoes begin to de-correlate, improving in the signal to noise ratio. However, the mean transmitted power for pulse and carrier-free systems is exceptionally low. The frequency sweeping modulation provides a higher transmitted power at the expense of a low WRF. For a GPR, surface penetration is of prime importance. Both pulse and chirp modulation require a high carrier frequency with respect to the signal bandwidth. A 5m range resolution requires a 1OMHz bandwidth, implying a carrier of 100MHz. At such frequencies, propagation losses are prohibitive (Ori and Ogliani, 1996). To achieve good penetration, frequencies <2OMHz should be used implying a lower bandwidth and reduced range resolution. The use of high frequencies can be avoided by using carrier free (impulsive) waveforms. However, this is not an option on mean power considerations unless extremely high voltage pulses are used. This is not consistent with high reliability and low EM1 effects. Ionospheric distortions require some correction for proper interpretation of the data. For chirp systems, compensation requires the computation of a real time, full band replica pulse. Pulse and frequency sweep waveforms may be corrected either by predistorting the transmitted signal or by the processing of the received signal. Pulse systems require (pre)processing of only one frequency. Frequency sweep systems require each frequency to be given its own predistortion. The best waveform appears to be a frequency sweeping system, transmitting a carrier wave over a IOMHz range from 51SMHz providing the low WRF does not pose a problem. This type of radar involves difficulties in antenna design as it has to operate over a large bandwidth. This waveform, generated by a direct digital synthesiser (DDS), consists of a staircase of 10,000 steps spaced 1kHz apart. The use of a dual DDS provides a useful way to drive two mixers with a 90” phase shift between them for complex de-ramping. The waveform can be predistorted in both phase and sweep rate for each frequency across the swept bandwidth, so compensating for phase adjustments and relative changes in the doppler shift across the band. The waveform at each frequency may be further predistorted to take into account the effects of the ionosphere. Antenna The antenna poses one of the largest design problems for the whole radar. It has to operate over a wide bandwidth, typically SISMHz. Multiple deployments may be necessary if the satellite has to be manoeuvred between differing orbits or to allow reorientation of the high gain antenna for data transmission. Retraction before such manoeuvres reduces the probability of setting unwanted vibrations in the satellite due to the inertia of the elements. The best orientation of the antenna is perpendicular to the flight path as this configuration may be used to reduce the clutter. The antenna should also be perpendicular to the solar panels to minimise any interference between the two electrical systems. The resulting antenna design consists of two tubular 15m elements. The metallic conductor is embedded in a carbon reinforced polyester support structure. The support is manufactured such that it may be flattened into a ribbon for storage but when released the inbuilt mechanical stresses cause the antenna to take on its tubular cross-section, forming a rigid structure. Modelling of such an antenna shows a fairly constant impedance match of 300 Q across the proposed bandwidth.
A Flexible Orbiting Radar
1931
The proposed radar transmits a series of frequency sweeps from 5- 15MHz and compresses them in the receiver using analogue de-ramping. The resulting beat-frequency signal is then analysed using FFT techniques (Skolnik, 1990). As signal frequency rises with range, low frequency components correspond to close range targets. The signal processing chain consists of a double FFT on the de-ramped signal. The first is a 1024 point complex FFT (192 per second), the second analysing 192 points (1024 per second). The sampling bandwidth of 512kHz allows the surface to be tracked to within -15km. These rates of computation assume that the position of the surface is known. If the surface position is lost, the radar would have to determine its range automatically. The surface could lie anywhere within the 2ms sweep. This is an extremely large processing task as the surface could lie in any one of 20,000 range gates. Depending upon the computational efficiency, it may be possible to increase the sampling bandwidth by a factor 2 or 4. This would then ease the task of surface tracking considerably as the surface topology does not vary on scale sizes of 30-60km. By employing a system for data selection, only the most significant range gates need be returned to Earth. This should make better use of the low telemetry allocation. If, however, the onboard processing cannot be performed correctly, whole sweeps will have to be returned in the telemetry for use in an Earth based processing system. This would result in a reduction in the temporal resolution of the data. and Operations Overall control of the radar is carried out by a series of commands‘fiom Earth. These commands instruct the control system to either configure the radar and run a series of predefined instructions (a macro) or uplink changes or new instruction macros into memory. If the commands received are corrupted, a default macro designed to work with minimum system requirements may be invoked in order to obtain some scientific return. The macros enable the radar to operate autonomously, having the ability to estimate and update values for certain operational parameters depending upon the chosen mode of operation but flexible enough to respond to a request from Earth. There are four basic operational modes foreseen for the radar:- altimeter/GPR, GPR/altimeter, surface detection and ionospheric sounder. Altimeter/GPR Mode. Using its full lOMHz bandwidth, the radar will operate primarily as an altimeter, but will also have full GPR functionality. The basic range resolution would be around 15m. The exact range may be determined by interpolation between range gates, with an uncertainty of 0.8m (assuming an SNR of 22 dB and a smooth surface). The GPR system would have a range resolution of 5m and a penetration depth of around 2km. GPR/Altimeter Mode. The primary target of this mode is for the radar to function as a GPR. Greater surface penetration is achieved using a lower bandwidth. Initially a frequency range 5- 1OMHz would be employed, allowing a depth penetration of around 2.2km and range resolution of 10m. If necessary, the DDS may be reprogrammed to use a lower frequency range if this penetration depth proves inadequate. The altimeter function still provides information but at a reduced resolution (-30m). Surface Detection Mode. Under normal conditions, the distance to the planet surface will be tracked automatically. Its approximate distance may be determined from the orbital dynamics of the satellite and this may be used as an estimate to set the WRF. If, however, the range is unknown or lock is lost the radar may automatically switch itself into a low WRF mode to get an unambiguous surface range before returning to its previous operational mode. Ionospheric Sounder Mode. During the daytime, the satellite will be orbiting above a dense ionosphere. Under such conditions, the operation frequency range of the radar would be changed to 0. I-5MHz to allow the topside of the ionosphere to be probed. Operational Modes. Actual operations will most likely involve combinations of the basic operational modes and depend upon whether the instrument is investigating the dayside or nightside of the planet. The dayside ionospheric plasma frequency (5MHz) restricts the radar to operating as a top side sounder. On the night side the plasma frequency is -0.3MHz and so both altimeter and GPR modes are equally usable. An example operational mode for the night side would involve using the radar as a GPR or altimeter with the occasional switching to the ionospheric sounder mode to probe the ionosphere. This would then allow corrections to the degree of phase distortion to be carried out. CLUTTER REDUCTION Ground clutter poses a severe problem with the radar. At any given instant the voltage and phase of the clutter signal appearing at the terminals of the antenna are the result of the contributions of a very large number of scattering elements - mainly
1932
S. P. Kingsley et al.
boulders on the surface and discontinuities in the topography. Because of this summation over a large number of signal contributions, surface clutter is inherently noise-like in its characteristics. The following methods will be used to reduce the clutter content of the return signal. As the antenna is deployed perpendicular to the direction of flight, any clutter arising from sources along the track will contain a doppler shift in addition to that from the eccentric orbit of the spacecraft. As the spacecraft moves with respect to the clutter sources, the doppler shift will change. This will affect the rate of change of frequency of the return signal with respect to the transmitted waveform. By processing from the mixer using a double FFT any doppler shifted signals such as those generated by clutter will be smeared out in frequency space whilst non doppler shifted signals should produce strong peaks. For clutter arising from cross track sources, transmitted and return signals propagate in the direction of the antenna’s null cone. The resulting antenna gain will be extremely low and signals will be lost in the noise of the system. The antenna is likely to be angled down towards the surface to enhance this effect. The maximum correlation time for received signal is proportional to lidoppler bandwidth. For the proposed radar design, this period is of the order of l4ms. If integration times greater than these are used, the effect is to lose SNR at the edges of the resolution cells and reduce the size of the clutter. Clutter only poses a problem from sources either very close-in or at very long range. Most short and medium range clutter does not enter the receiver because it is switched off for the 2ms (+300km of range) of the next transmission cycle. Clutter falling outside the transmission cycle is severely attenuated by range eclipsing losses (i.e. only part of the signal from a particular range enters the receiver because of transmit-receive switching). The lack of a complete FM sweep means the clutter signal fails to process properly and is attenuated. Close-in clutter signals signals from the horizon echoes, i.e. at the same attenuated as described very oblique angles.
are minimised of Mars come time as echoes above, but also
as described above and come from relatively small annuli. Very long range clutter from much larger areas. They can avoid eclipsing by arriving as ‘second time around’ from the next transmit-receive cycle. However, these long range echoes are not only by increased propagation losses and the inefficiency of the backscattering process at
Once a number of scans of the same region exist, it should be possible to use along track Synthetic Aperture Radar processing techniques, improving the signals and minimising the clutter content. The efficiency of each of these various methods needs to be assessed both via simulations their performances.
and instrument testing to evaluate
SUMMARY A design for an orbiting ground penetratin, 0 radar has been presented. This radar overcomes the conflicting need of ground penetration, high spatial resolution and peak power by the use of a frequency sweeping system with flexible programming and an optimised antenna design. Such a device should be capable of detecting liquid water at a depth of several kilometres. REFERENCES Carr. Michael H., Channels and valleys on Mars: cold climate features formed as a result of a thickening cryosphere, Planet. sp.‘sci. 44, I41 I, 1996. Clifford, Stephen M., A Model for the Hydrodynamic and Climactic Behavior of Water on Mars, J. Geophys. Res E 98, 10973, 1993. Hanson, W. B., S. Sanatani, D. R. Zuccaro, The Martian ionosphere as observed by the Viking retarding potential analysers, J. Geophys. Res A 82,435 1, 1977. Masursky, Harold, .I. M. Boyce, A. L. Dial, G. G. Scaber, and M. E. Strobell, Classification and time formation of Martian Channels based on Viking data, J. Geophys. Res A 82,4016, 1977. Ori, G. G., and F. Ogliani, Potentiality of the ground penetrating radar for the analysis of the stratigraphy and sedimentology of Mars, Planet. SpSci 44, 1303, 1996. Skolnik, M. I., Radar Handbook, Pub. McGrawHill, 1990. Zhang, M. H. G.. J. G. Luhmann, and A. J. Kilore, An observational study of the night-side ionospheres of Venus and Mars with radio occultation methods, J. G’eophys. Res A 95, 17095, 1990.