Helicopter Hard-mounted GPR Snow and Ice Thickness Measurement Systems. Call-up Number: # F

Transcription

1 Helicopter Hard-mounted GPR Snow and Ice Thickness Measurement Systems Call-up Number: # F by Sensors by Design, Ltd. 217 Lorne Avenue Newmarket, Ontario L3Y 4K5 Prepared for: Dr. Simon Prinsenberg Fisheries and Oceans Coastal Oceanography Division Bedford Institute of Oceanography P.O. Box 1006 Dartmouth, Nova Scotia Canada, B2Y 4A2 October 8, 1999

2 Table of Contents Introduction... 1 Background... 2 Hardware Components... 2 Snow Thickness Processing... 3 Hard-mounted Snow/ice Thickness Sensor... 4 Locating Suitable Commercial Equipment... 4 GPR systems from Geophysical Survey Systems, Inc... 5 GPR systems from MALÅ GeoScience... 6 GPR systems from Sensors & Software Inc... 7 Recommendations GPR Ice Thickness Measurements In More Complex Environments Rubble fields in Fresh Water Ice Fresh water Brash Ice Sea Ice Field Trial - GPR Vs. EM GPR System Hardware Electronics Improvements GPR System Improvements through software and implementation Summary References Appendix A - Ground Penetrating Radar Manufacturers...19 Appendix B - Brochures of commercial GPR systems...20 Appendix C - Case Studies provided by Sensors and Software, Inc...32 Appendix D - Paper on the snow thickness sensor presented at GPR ii

3 INTRODUCTION The purpose of this report is to provide information for the development of a second ground penetrating radar sensor for hard mounting to a Coast Guard BO- 105 helicopter to measure snow depths. Also, the report explores other possible radar system requirements for the measurement of fresh water ice thickness and sea ice thickness. Ideally, the radar equipment will fit within the pod developed by Transport Canada as shown in Figure 1. As the ground penetrating radar hardware used within the existing snow thickness sensor is no longer manufactured, new commercially available hardware must be reviewed for its suitability for assembly into a real-time snow thickness sensor. Developments in commercially available ground penetrating radar technology and signal processing methods may permit the use of radar for measuring sea ice thickness and more complex fresh water ice environments in addition to measuring snow depths. A brief look at the more complex ice measurement problems is provided. Figure 1 Photograph of pod mounted to skid-gear of a BO-105 helicopter. 1

4 BACKGROUND The current snow thickness radar system was developed for installation within the Ice Probe sea ice thickness measurement system (Aerodat, 1995). Ice Probe measures the combined thickness of snow and sea ice. The radar component was to measure the snow thickness separately so that the Ice Probe system could report snow thickness and ice thickness separately. To date, the snow thickness sensor has not been used operationally with Ice Probe. Snow thickness survey flights have been performed using a small helicopter towed bird made for the snow sensor. The snow thickness sensor has minimum snow thickness measurement of approximately 25 cm with a resolution of 5 cm (Prinsenberg et al., 1993, Lalumiere 1992). The radar range window has been configured for a maximum flying altitude of approximately 30 m. The minimum flying height is approximately 10 m. Radar signal reverberations picked up by the radar receiver create noise which attenuates sufficiently after approximately 60 to 80 ns. Figure 4 in the paper attached (Lalumiere, 1998) in Appendix D shows a plot of raw radar data with the antenna ringing and the radar response from a fresh-water ice surface. HARDWARE COMPONENTS The ground penetrating radar system was originally purchased from a commercial supplier in The radar system was flight tested for use as a snow thickness sensor in 1991 at Tuktoyaktuk, N.W.T. as part of the Transport Development Corporation s (TDC) electro-magnetic (EM) ice thickness measurement system. For the 1992 trial near St. Anthony, NFLD., new electronics were developed so that all the radar electronics could be contained in a small package mounted in the TDC EM bird. Each of these systems required data to be logged on analog tape or on a laptop computer with an analog-todigital converter board. All processing of data were performed after the flight. Results of the 1991 Tuktoyaktuk field trial can be found in Lalumiere, Development work began in 1997 to incorporate new low-cost DSP-based microcontrollers and stereo analog-to-digital and digital-to-analog converters for the implementation of real-time signal processing for the measurement of snow thickness and flying height. The snow thickness radar system is packaged in a small 30 cm by 35 cm by 15 cm package, with flying height and snow thickness measurement results outputted over an RS-232 serial link. Test flying of the new snow thickness radar sensor was performed during March 1997 and March During the 1998 field trial, the real-time flying height and fresh water ice thickness data were logged using the BIO Video Sensor System (Sensors by Design, 1999) which also logged GPS positioning information, laser 2

5 and radar altimeter data and downward looking digital video images. The paper included in Appendix D shows some of the results from the 1998 field trial. SNOW THICKNESS PROCESSING As part of the 1991 and 1992 field trials, a model was established as a basis for classifying radar echoes as ice echoes or snow echoes. A software algorithm was implemented following the model to automatically estimate snow thickness. The model used the following assumptions: the radar footprint diameter is approximately equal to antenna height (about 15 m); over a smooth flat reflector most of the energy is returned from a region with a radius of less than one tenth antenna height (first Fresnel zone); small radar targets in a rubble field return echoes with much smaller amplitudes than large flat targets (flat ice); the echo from the ice surface (whether covered by snow or not) has the largest amplitude in the trace; and the echo from the air/snow interface is the largest signal greater than random noise levels that arrives before the ice echo. Before peak location begins, the raw radar data is filtered to remove high frequency random noise, low frequency noise and background system noise. For every trace the maximum peak value is found and its value and location are stored. This peak might correspond to an echo from the top of the ice. The data is searched for the first peak that has a value greater than a given threshold. The threshold is chosen to be well above the RMS noise level. If a peak is found, its location and value are also recorded. This peak might correspond to an echo from the air/snow interface. The peak locations are processed for snow thickness by subtracting the snow peak position from the ice peak position, dividing by the sampling frequency and multiplying by the radar velocity in the snow (0.15 m/ns). No snow thickness determination is made if: the ice echo is too small; no snow echo is found; or the ice peak amplitude is smaller than the snow peak amplitude. A more detailed description of the snow thickness processing can be found in the report on the 1991 Tuktoyaktuk field trial (Lalumiere, 1992). In preparation for the 1997 field trial, snow thickness processing algorithm was coded to run in real-time on the DSP board. The DSP board provides flying 3

6 height and snow thickness over an RS-232 serial interface to a logging computer (such as the BIO Video Sensor System). HARD-MOUNTED SNOW/ICE THICKNESS SENSOR There are a number of options for development of a new ground penetrating radar sensor for hard-mounting to a Coast Guard BO-105 helicopter. As the ground penetrating radar hardware used within the present snow thickness sensor is no longer manufactured, new commercially available hardware must be reviewed for its suitability for assembly into a real-time snow thickness sensor. Another option, would be to source the same hardware components used in the current snow thickness sensor as used equipment. Requirements for the hard-mounted system: overall low cost; ease of interfacing to real-time digital signal processing (DSP) board; small packaging for mounting within current BO-105 pod (seen in Figure 1); data logged using the BIO Video Sensor System; operate at increased flying height from current bird mounted system. Any new GPR system with digital output is going to require a new processing computer to process the raw radar data into flying height and snow thickness. The current embedded DSP board used in the snow thickness sensor has only one serial link so a new real-time processing system will likely be required. A hard-mounted GPR system has two operational implications that are not part of a bird mounted system. The first is possible reverberation noise of the radar signal bouncing off the helicopter. All of the GPR antenna systems reviewed below use shielded antennas which should minimize the reverberation effect. Secondly, a bird mounted radar system operates at a survey altitude lower than typical (or desired) helicopter survey altitude. The increased flying height requires a longer radar range window, increased radar performance and there is a chance that more atmospheric noise will be picked up by the radar receiver. As the increase in system performance in recent commercial GPR systems comes from reduced receiver noise levels, a test flight of a new system would show no apparent improvement in performance if atmospheric noise levels become larger than receiver noise levels. LOCATING SUITABLE COMMERCIAL EQUIPMENT The real-time snow thickness system requires a commercial ground penetrating radar (GPR) system that will provide raw data to a computer where the real-time processing will be performed. The DSP board in the current snow thickness 4

7 system outputs the processing results (flying height and snow thickness) at approximately 16 samples per second over a RS-232 serial link. There are three suppliers of commercial GPR equipment, Geophysical Survey Systems, Inc, of New Hampshire, USA, MALÅ GeoScience of Sweden (they have a US office and a Canadian distributor) and Sensors & Software Inc. of Ontario, Canada. Address and contact information for the commercial suppliers are provided in Appendix A. Brochures from the different manufacturers are provided in Appendix B. GPR systems from Geophysical Survey Systems, Inc Geophysical Survey Systems, Inc (GSSI) manufactures GPR systems with analog signals outputted from the antenna units. The antenna units contain the radiating elements and the transmitter and receiver electronics. The analog data has a bandwidth which is approximately 50 khz. Current production control units provide control signals for the transmitter and receiver electronics and digitize the analog radar signals coming from the antenna unit. Though no longer available, GSSI s model 3102DP antenna unit is a component in the current snow thickness sensor. The model 3102DP has a second transmitter which provides a transmitted pulse with double the standard voltage (300 V versus 150 V). The high voltage transmitter and the receiver electronics are available as spare parts. GSSI s model 5103 antenna unit is most similar to the unit in the present snow thickness sensor. It has a centre frequency of 400 MHz and its dimensions are 30 cm by 30 cm by 17 cm. This antenna is too large for use in the BO-105 s pod. It would need to have the transmitter and receiver antennas removed from its factory packaging for mounting inside the pod. The 5103 antenna unit does not have the option for the higher voltage transmitter. There is a good chance that the high voltage transmitter (available as a spare part for the older 3102DP antenna) could be mounted in the 5103 unit. The high voltage transmitter is important for good radar return signal levels when flying at higher altitudes (20 or 30 metres or perhaps greater). GSSI control units (System 2 or System 10) do not provide real-time digital output which is necessary for snow thickness processing purposes. Using a GSSI antenna unit requires buying and building control electronics to replicate the current functions in the snow thickness sensor. One of the boards in the control electronics of the snow thickness sensor is from an obsolete GSSI System 7 control unit. Older control units are available as used equipment and are not too expensive. Table 1 lists the parts and labour necessary to build a 5

8 GSSI-based hard-mounted snow thickness sensor with the same measurement capability of the current snow thickness sensor. Table 1 - Cost estimate for replicating the bird-mounted snow thickness sensor as a hard-mounted helicopter installation. Item Labour/Equipment Cost Estimate Source radar control Used equipment $7,000 boards GSSI HV transmitter $US 656 $1,000 GSSI receiver $US 1,176 $1,750 electronics for cal. unit. Purchase GSSI 5103 $US 4,900 $7,500 antenna unit Power supply, DSP $1,500 board and misc. parts Build radar control 10 days $8,000 electronics (Copy current snow thickness radar system) Repackage GPR 3 days $2,400 antenna for mounting in helicopter pod Total: $29, GPR systems from MALÅ GeoScience MALÅ GeoScience offers their GPR systems with 500 MHz and 800 MHz antennas. The dimensions of the 500 MHz antenna are 50 cm by 33 cm by 16 cm. The dimensions of the 800 MHz antenna are 38 cm by 20 cm by 12 cm. The factory packaging of the 500 MHz is too large to fit inside the BO-105 s pod, but the 800 MHz may fit inside the pod without much alteration of the antenna or the pod. The 800 MHz antenna would allow the measurement of thinner layers, but the trade off would be to limit flying height to lower altitudes. The MALÅ control unit has a parallel interface for fast output of raw radar data providing at least double the current snow thickness sensors scan rate of 16 Hz. The MALÅ GPR control electronics provide the fastest transmitter repetition rate of the GPR manufacturers. The fast transmitter repetition rate permits fast radar scan rates with long radar range windows. MALÅ GeoScience antennas electronics are powered by batteries and are linked by fibre-optic cable to the control unit. Adding a metal power cable to the system 6

9 may add noise to the radar signal, which may require some engineering time to eliminate. MALÅ GeoScience does not provide specifications for their receiver noise levels or transmitter voltages, but they do report that the overall system performance is 150 decibels (db) (unfortunately the system performance specifications do not include antenna efficiency which varies from one manufacturer to the next). This system performance is approximately 30 db greater than the performance of the current snow thickness sensor, increasing the operational flying height by perhaps 10 to 20 m. The upper limit will be dictated by the increased atmospheric noise levels found at higher survey altitudes. A test flight will be needed to measure the atmospheric noise levels at altitudes ranging from 30 to 50 m. The use of the MALÅ GPR system for real-time snow thickness processing will require the development of a new processing computer to convert the raw radar data to flying height and snow thickness. In typical GPR applications, the MALÅ GeoScience control unit connects to a computer running MALÅ GeoScience user interface software. For the snow thickness sensor, the user interface software is not used and the control unit would connect directly to the real-time processing computer. Discussions with MALÅ GeoScience indicate that they would be willing to provide the communications protocols necessary for the real-time processing computer to control the GPR control unit and receive the digital data. Table 2 lists the parts and labour necessary to build a MALÅ GeoScience-based fixed mount snow thickness sensor with the same measurement capability of the current snow thickness sensor. GPR systems from Sensors & Software Inc. Sensors & Software Inc has two GPR systems that are relevant for possible use as a snow thickness sensor, the self-contained Noggin unit and the Pulse-Ekko 1000 model which has separate boxes for the transmitter antenna, receiver antenna and control unit. The Noggin 500 is small (39 cm by 22 cm by 16 cm) and relatively inexpensive (approximately $21,000). It might fit within the BO-105 s pod without any modifications (although an internal pod bracket would need to be changed). The main disadvantage of the Noggin is limited output data rate. The data link is RS-232 at 115 kbaud. Airborne applications require a long time window. Using the parameters of the current snow thickness sensor, which has a window length of 250 ns (which limits flying height to 30 m) and the Noggin sample rate of 200 7

10 ps, the maximum scan rate is 8 scans per second (which is half the rate of the current snow thickness sensor). Table 2 - Cost estimate for building a snow thickness sensor for hardmounting to a helicopter based on MALÅ GeoScience GPR equipment. Item Labour/Equipment Approximate Cost GPR control unit $US 13,325 $20,000 Transmit/receive $US 3,650 $5,500 electronics 500 MHz antenna $US 3,125 $4,700 Embedded PC $3,000 hardware Software tools $3,000 Port snow thickness 6 days $4,800 processing software to new computer hardware Testing radar and 6 days $4,800 packaging processing computer and GPR control unit Subtotal: $45, Mount in helicopter pod Possible pod modifications Total: Transport Canada airworthiness approvals TBD TBD TBD The PulseEkko 1000 unit is more expensive and larger than the Noggin, but it is more flexible in configuration and has a parallel data transfer option for faster radar scan rates. The transmit and receive antennas are in a separate package. The size of each antenna (with the electronics) is 23 cm by 16 cm by 11 cm. This small size may fit inside the BO-105 s pod without any modifications. Also available with the PulseEkko 1000 unit are 900 MHz antennas. If these units have sufficient performance, the measurement of thinner snow layers may be possible. Sensors & Software reports that their receiver noise level specifications are much lower than GSSI systems. The transmitter voltage is approximately 200 V. The system performance (transmitter voltage divided by receiver sensitivity) of the system is 160 decibels (db) (unfortunately the system performance specifications do not include antenna efficiency which varies from one manufacturer to the next). This system performance is approximately 40 db greater than the performance of the current snow thickness sensor, increasing the operational 8

11 flying height by perhaps 10 to 20 m. The upper limit will be dictated by the increased atmospheric noise levels found at higher survey altitudes. The use of either of the Sensors & Software units for real-time snow thickness processing will require the development of a new processing computer to convert the raw radar data to flying height and snow thickness. As with the MALÅ GeoScience unit described above, Sensors & Software GPR units connect to a computer running Sensors & Software user interface software. For the snow thickness sensor, the user interface software is not used and the control unit would connect directly to the real-time processing computer. Discussions with Sensors & Software indicate that they would be willing to provide the communications protocols necessary for the real-time processing computer to provide limited control of the GPR control unit and to receive the digital data. Table 3 lists the parts and labour necessary to build a Sensors & Software-based fixed mount snow thickness sensor with the same measurement capability of the current snow thickness sensor. Table 3 Cost estimate for building a snow thickness sensor for hardmounting to a helicopter based on Sensors & Software GPR equipment. Item Labour/Equipment Approximate Cost PE1000 GPR system $33,000 (or $21,000) with 450 MHz antennas (or Noggin 500 System) Embedded PC $3,000 hardware Software tools $3,000 Port snow thickness 6 days $4,800 processing software to new computer hardware Testing radar and 6 days $4,800 packaging processing computer and GPR control unit Subtotal: $48,600 (or $36,600) Mount in helicopter pod? days TBD Possible pod Transport Canada airworthiness TBD modifications approvals Total: TBD 9

12 RECOMMENDATIONS As seen from the above documentation, several unknowns in GPR hardware performance prevent us from stating a clear path for the development of the next fixed-mount radar sensor for both measurements of snow and ice properties. Building a system as outlined in Table 1, will replicate the snow thickness sensor functionality in a hard-mounted configuration, but helicopter survey altitudes will be limited to traditional radar-bird survey altitudes. However, to increase the operating altitude and to provide a future radar development path using current commercial equipment the following test plan is proposed. The first thing (phase 1) that needs to be determined are the possible system performance improvements provided by GPR systems from MALÅ GeoScience or Sensors & Software. One day rental of GPR systems from each manufacturer would enable the collection of test data to verify system performance specifications. The results from this test would be reported with recommendations for the next phase. Table 4 lists the costs associated with the phase 1 test. The second test (phase 2) would select several GPR configurations for helicopter reverberation noise testing. Airborne tests with the rental units mounted inside the small radar test bird could be performed. Airborne tests would provide information on noise pickup at higher altitudes. System performance checks could be made flying over a know surface layer such as a sandy beach at a nearby lake. This testing is based on helicopter availability on the ground at the Coast Guard hanger in Charlottetown and possible helicopter flying time. Table 5 lists the costs associated with the phase 2 test. After the second test the selection of a commercial GPR system for the hardmounted application can be made and the development work can commence if it is deemed appropriate to do so. The first and second trials could be performed before the year end of 1999, allowing for development of a prototype hard-mounted system prior to March 2000 field testing. 10

13 Table 4- Phase 1 Testing Cost Estimate Item Labour Estimate Cost Estimate Pickup the rental units 1 day $800 Setup systems and collect test data 3 days (1 day for each $2,400 system rented) Summarize results 1 day $800 Rental Items Equipment GSSI antenna and control unit rental 400 MHz antenna and $500 analog control unit MALÅ GeoScience GPR rental Control unit electronics $800 with 500 and 800 MHz antennas Sensors & Software GPR rental Noggin 500 unit and $1,100 PE1000 unit with 450 and 900 MHz antennas Misc. parts and consumables $200 Total Cost $6,600 Table 5 - Phase 2 Testing Cost Estimate Item Labour Estimate Cost Estimate Travel time to Halifax and return 1 day $800 Setup systems and collect test data 3 days $2,400 Summarize results 3 day $2,400 Direct Cost Items Equipment GPR equipment rental for 1 week MALÅ and S&S $3,000 Airfare to Halifax $1,600 Living and per diem $600 Shipping Equipment $500 Misc. parts and consumables $200 Total Cost $11,500 11

14 GPR ICE THICKNESS MEASUREMENTS IN MORE COMPLEX ENVIRONMENTS The current snow thickness sensor has been developed for operation over flat snow and fresh water ice surfaces. In the past GPR systems have been used with limited or no success in more complex environments such as sea ice. New GPR systems with increased system performance and new equipment implementations and signal processing techniques provide opportunities for GPR-based ice measurements that have not been successful in the past. Hardware improvements in commercial GPR systems in the last few years have increased performance of ground penetrating radar equipment. The primary improvements are in reduced receiver noise levels, lower system noise (ringing) levels and 16 bit digital data output. Signal processing to combine signal from multiple radar units for improved performance have been applied in other GPR applications and is still to be proven for snow and ice measurement applications. RUBBLE FIELDS IN FRESH WATER ICE Rubble fields provide many small targets for the GPR system. A radar profile over a rubble field is difficult to interpret as the small targets appear as many hyperbolic patterns. The tails of the hyperbolic patterns can be seen in the radar profile at a return time that would indicate that the ice is deeper than it actually is. Processing techniques that remove or reduce the hyperbolic patterns would permit the use of GPR systems for measuring ice thickness in the presence of a radar scattering rubble field. Radar returns from point targets on the surface (rayleigh point targets.2 m in diameter - from the model developed in Annan and Davis, 1977) are roughly the same as the echo from the bottom of the fresh water ice layer. Array processing techniques (such as data migration) which reduce the amplitude of the rubble field scatterers would provide a clearer radar profile of the bottom of the ice. Appendix C has a data migration case study provided by Sensors & Software, showing the reduction of hyperbolic patterns to point targets. After processing, the point targets are positioned in the data profile at the location directly under the radar system and returns from ahead or behind of the radar are reduced significantly. Targets from far off to the side of the radar antenna may still cause interference. The hyperbolic pattern from a point target off to the side will appear broader than the hyperbolic pattern from a point target directly under the travel path of the radar system, reducing the effectiveness of the data migration processing. Also, 12

15 for the case of a 15 m survey flying height over 0.5 m thick ice, migrated radar returns from point targets beyond 20 degrees to the side will appear at a return time greater than the time of the radar return from the bottom of the ice. To counter the off-to-the-side effects an array of transversely-mounted radar antennas under the helicopter will be required. The number of antenna elements in the array will depend on the required reduction in amplitude of the undesired scattering targets (or the required amount of clutter removal). The physical size of the array will depend on the number of antenna elements needed and the frequency of the radar antennas used, as lower frequency antennas are larger than high frequency antennas. The very rough ice surface of a rubble field can reduce the strength of the transmitted radar pulse entering the ice and consequently reduce the echo strength from the ice/water interface. An array GPR antennas offers the ability to combine the radar returns from each antenna pair, increasing the signal strength of the ice/water interface. FRESH WATER BRASH ICE Brash ice introduces the problem of having a significant amount of water within the ice layer which greatly increases the variability of the radar signal s velocity in the ice layer. Brash ice also provides many scattering targets that will make data interpretation difficult. A solution to the brash ice thickness measurement problem may be obtained with the use of an array of GPR antennas placed on the ice surface. Appendix C has a velocity measurement case study provided by Sensors & Software. A common mid-point (CMP) survey shows the use of multiple antenna separations to measure radar velocity in the surface layer. A simpler configuration using a single transmitter antenna and multiple receiver antennas can be used. This configuration is known as a Wide Angle Reflection and Refraction (WARR) sounding technique (Davis et al., 1987). With the WARR configuration, a GPR array can be made with fixed-mounted antenna elements. For testing purposes the array could be towed over the ice surface if it is strong enough to hold people. Also, the array could be mounted from a bridge or other structure over the ice surface if the ice is moving underneath. For a hardmounted helicopter-based system, the helicopter would hover close to the surface of the brash ice for short periods of data collection. Multiple travel paths of the radar signal from the transmitter through the brash ice to multiple receiver antennas would permit the measurement of the velocity of the radar signal in the brash ice and a likely ice thickness. 13

16 SEA ICE Surface-based ground penetrating radar technology has been used to measure sea ice thickness since the early 1970 s. Helicopter-borne sea ice measurements were made by 1975 (Morey, 1975) and perhaps earlier. By the late 1970 s, the Centre for Cold Ocean Resources Engineering (C-CORE) had designed a ground penetrating radar specifically for airborne sea ice thickness measurements (Butt and Gamberg, 1979). The C-CORE system was based on GSSI GPR electronics and it had an antenna designed for cargo hook mounting under a helicopter. Figure 2 shows a photograph taken in 1987 of the C-CORE system mounted under a Bell 206L helicopter at Resolute Bay, N.W.T. Figure 2 Photograph of cargo hook mounted to C-CORE ice thickness radar system Field Trial - GPR Vs. EM In 1987 Canpolar Consultants Limited was contracted by Transport Development Centre of Transport Canada to perform field testing of several ice thickness sensors. A field trial held at Resolute Bay, N.W.T. compared the C-CORE GPR ice thickness system with an experimental electromagnetic (EM) and laser system (Rossiter and Lalumiere, 1988). This report recommended the development of an operational EM-based ice thickness sensor, which lead to the development of the TDC EM ice thickness 14

17 sensor and following that system, the development of the Ice Probe system for the Coast Guard. The conclusions of the report states Although VHF radar systems can sound cold, undeformed sea ice, they fail totally for warm, saline or deformed ice (such as ridges), and for ice containing layers or brine inclusions. The summary of the report states The impulse radar performed as expected over cold, undeformed, relatively non-saline ice - when echoes are received they can be interpreted to give ice thickness to +/- 20 cm or better. As ice salinity or deformation increases, ice bottom echoes become very weak. The discussion section of the report states It is fairly clear from the previous section that both the impulse radar and EM induction systems were able to effectively measure the thickness of ice encountered during the field trials. Figure 3 shows a GPR profile over multi-year ice (from Rossiter and Lalumiere, 1988). GPR System Hardware Electronics Improvements Addressing the conductive sea ice problem, overall system performance improvements in GPR systems since the 1987 trial provides increased sensitivity for the detection of the ice/water interface echo. Attenuation of the radar signal in sea ice can vary from 1 to 10 db/metre (Morey, 1975). If 10 db/m is the worst case attenuation in warm first year sea ice, then a 40 db improvement in system performance would permit the sounding of 2 m thick sea ice which could not have been sounded in the past. System noise problems (ringing) shown in the radar figures in the 1987 field trial report (as shown in figure 3), significantly reduce the sensitivity of the GPR system. Current GPR systems have reduced system noise compared with systems available 10 years ago. Also, recent improvements in antenna shielding should minimize the reverberations encountered in a helicopter hard-mounted installation. GPR System Improvements through software and implementation Traditionally, the problem with measuring the thickness of multi-year sea ice is the scattering of the radar signal from targets on surface, within the ice and at the ice/water interface. Interpretation of a radar profile is difficult as the scattering targets appear as many hyperbolic patterns. Late arriving radar returns could be from the bottom of the ice or they could be from scattering targets within the multi-year ice or on the surface. 15

18 Figure 3 - GPR profile over multi-year ice from the Rossiter and Lalumiere, 1988 report. 16

19 As discussed in the above section on rubble fields, applying data migration processing on a profile of radar data will reduce the hyperbolic patterns to point targets. As a result, the latest arriving radar returns can be assumed to be from the bottom of the ice and the ice thickness interpretation can be made with greater confidence. SUMMARY The GPR hardware, processing and implementation improvements have not yet been demonstrated for ice thickness measurement. A field trial will be required to demonstrate some of the improvements. GPR data logging with accurate GPS positions will be required for the data migration processing. GPR hardware which supports a multi-antenna array configuration will need to be sourced or built. 17

20 REFERENCES Aerodat Inc. (1995) Design, Construction And Testing Of A Production Prototype Real-Time Airborne Electromagnetic Ice Measurement Sensor - Phase III Final Report, submitted to The Canadian Coast Guard, 44 pp. Annan, A.P., Davis,J.L. (1977) Radar Range Analysis for Geological Materials, Report of Activities, Part B; Geol. Surv. Can., Paper 77-1B, pp Butt, K.A., Gamberg, J.B. Technology of an Airborne Impulse Radar for Sounding Sea Ice, Preceedings of the International Workshop on the Remote Estimation of Sea Ice Thickness, St. John s, Newfoundland, September 1979, p Davis, J.L., Annan, A.P. and Vaughan, C. (1987) Placer Exploration Using Radar and Seismic Methods, CIM Bulletin Volume 80, No. 898 February 1987, p Lalumiere, L.A. (1998) Snow and Ice Thickness Radar System, GPR 98, Lawrence, Kansas, p Lalumiere, L.A. (1992) "Analysis of Snow Thickness Data Collected by Impulse Radar over the Beaufort Sea Shelf in 1991", Can. Contractor Rep. of Hydrogr. Ocean Sci. No. 43; viii + 73 pp. Morey, B.M. (1975) Airborne Sea Ice Thickness Profiling Using an Impulse Radar, U.S. Coast Guard Technical Report No. CG-D Prinsenberg, S.J., Holladay, J.S. and Lalumiere, L.A. (1993) " Electromagnetic/Radar Ice and Snow Sounding Project over the Newfoundland Shelf in 1992", Can. Tech. Rep. Hydrogr. Ocean Sci. No. 144; vii + 57 pp. Rossiter, J.R. and Lalumiere, L.A. (1988) "Evaluation of Sea Ice Thickness Sensors", submitted to Transportation Development Centre, Report No. TP9169E, 58 pp. Sensors by Design, Ltd. (1999) Video Sensor System Analysis Report, Can. Tech. Rep. Hydrogr. Ocean Sci., in press. 18