UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Digital signal processing of UHF radio echo sounding data from northern Ellesmere Island Prager, Bradley Thomas 1983

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
831-UBC_1983_A6_7 P72.pdf [ 15.15MB ]
Metadata
JSON: 831-1.0052939.json
JSON-LD: 831-1.0052939-ld.json
RDF/XML (Pretty): 831-1.0052939-rdf.xml
RDF/JSON: 831-1.0052939-rdf.json
Turtle: 831-1.0052939-turtle.txt
N-Triples: 831-1.0052939-rdf-ntriples.txt
Original Record: 831-1.0052939-source.json
Full Text
831-1.0052939-fulltext.txt
Citation
831-1.0052939.ris

Full Text

DIGITAL SIGNAL PROCESSING OF UHF RADIO ECHO SOUNDING DATA FROM NORTHERN ELLESMERE ISLAND by BRADLEY THOMAS PRAGER B.Sc. (Honours Geophysics), University of Alberta, 1980 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE in THE FACULTY OF GRADUATE STUDIES (Department of Geophysics and Astronomy) We accept t h i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1983 © Bradley Thomas Prager, 1983 In presenting t h i s thesis i n p a r t i a l f u l f i l m e n t of the requirements for an advanced degree at the University of B r i t i s h Columbia, I agree that the Library s h a l l make i t f r e e l y available for reference and study. I further agree that permission for extensive copying of t h i s thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It i s understood that copying or publication of t h i s thesis for f i n a n c i a l gain s h a l l not be allowed without my written permission. Department of Geophysics and Astronomy The University of B r i t i s h Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date October 4, 1983 DE-6 (3/81) ABSTRACT This thesis is a preliminary attempt to apply d i g i t a l signal processing techniques to ice radar data. Seismic processing inspired many of the techniques: linear f i l t e r i n g , d i f f e r e n t i a t i n g , automatic gain control, and stacking. The processing is designed to enhance the radar section by reducing noise and ' increasing the amplitude of small r e f l e c t i o n s , making r e f l e c t i o n s easier to locate. Once r e f l e c t i o n s have been located, ice thickness and re f l e c t o r properties can be interpreted from the data. Ice thickness and power r e f l e c t i o n c o e f f i c i e n t (PRC) can always be obtained; i f the depth to the r e f l e c t o r undergoes a s u f f i c i e n t l y large change, the propagation loss rate of the ice can also be calculated. The processing and interpretation techniques are applied to ice radar data from a 1981 survey on northern Ellesmere Island. Ice thickness, PRC, and propagation loss rate for the following areas are calculated: Milne and D i s r a e l i Glaciers, Milne and Ward Hunt Ice Shelves, and a small ice cap surrounding Mt. Oxford. The glaciers have a maximum thickness exceeding 700 m, a basal PRC of about -30 dB, and a t y p i c a l propagation loss rate of 0.025 dB/m (at 840 MHz). The thinner ice areas of the Milne and Ward Hunt Ice Shelves produce either no basal r e f l e c t i o n or only a faint r e f l e c t i o n ; t h i s i s taken to indicate basal accretion of brackish ice or brine soaking, produced by meltwater flowing out from the fiords underneath the ice shelves. i i TABLE OF CONTENTS Abstract i i L i s t of Tables v L i s t of Figures vi Acknowledgements v i i i 1. Introduction 1 1.1 Data Processing and Radio Echo Sounding 4 1.2 G l a c i o l o g i c a l Studies on Northern Ellesmere Island 6 2. Processing D i g i t a l Radio Echo Sounding Data 9 2.1 Introduction 9 2.2 Recording the Data 9 2.3 Processing to Produce Interpretable Sections 12 2.3.1 Stacking 15 2.3.2 High-Pass F i l t e r i n g ...18 2.3.3 D i f f e r e n t i a t i n g 18 2.3.4 Automatic Gain Control 20 2.3.5 Low-Pass F i l t e r i n g 23 2.3.6 Display 23 3. Interpreting a D i g i t a l Radar Section 31 3.1 Introduction 31 3.2 Locating Reflections 31 3.2.1 Effects Which Obscure Reflections 32 3.3 Ice Thickness and Layer Separation 37 3.4 Reflection and Propagation Losses 39 3.4.1 Power Reflection Coefficient 40 3.4.2 D i e l e c t r i c Loss and the Loss Tangent 43 4. Radio Echo Sounding on Northern Ellesmere Island 47 i i i 4.1 Introduction 47 4.1.1 Areas Flown and Navigation 48 4.1.2 Accuracy of Results 50 4.2 Mt. Oxford Ice Cap and the Milne and D i s r a e l i Glaciers . , 51 4.2.1 Ice Thickness 53 4.2.2 Propagation Loss Rate and the Basal PRC 60 4.3 Milne Ice Shelf 68 4.3.1 Ice Thickness 70 4.3.2 Basal Power Reflection Coefficient 70 4.4 Ward Hunt Ice Shelf 72 4.4.1 Ice Thickness 77 4.4.2 D i e l e c t r i c Loss and Basal PRC 77 4.4.3 The PRC of the Lake Ice/Sea Ice Interface ...79 5. Concluding Remarks ....82 5.1 Data Processing and Radio Echo Sounding 82 5.2 Radio Echo Sounding on Northern Ellesmere Island ..83 References 86 iv LIST OF TABLES 2.1 System parameters for the UBC 840 MHz radar 10 4.1 Propagation loss rate and basal PRC 64 v LIST OF FIGURES 1.1 The study area: Ellesmere Island, N.W.T., Canada 2 1.2 Areas sounded in 1966 and 1981 3 2.1 Stacking 17 2.2 High-pass f i l t e r i n g ...19 2.3 D i f f e r e n t i a t i n g 20 2.4 Automatic gain control 22 2.5 Low-pass f i l t e r i n g 24 2.6 A-scope, variable area, and threshold displays 26 2.7 Grey-scale display 28 2.8 Colour r e f l e c t i v i t y display 30 3.1 D i f f r a c t i o n patterns 36 3.2 Multiple r e f l e c t i o n s 36 4.1 Overland f l i g h t l i n e s 52 4;2 Radar section from the Mt. Oxford ice cap 54 4.3 Radar section from the D i s r a e l i Glacier 55 4.4 Radar section from the Milne Glacier 56 4.5 Depth p r o f i l e for the Mt. Oxford ice cap 57 4.6 Depth p r o f i l e for the D i s r a e l i Glacier 58 4.7 Depth p r o f i l e for the Milne Glacier 59 4.8 Echo strength versus depth for the Mt. Oxford ice cap 61 4.9 Echo strength versus depth for the D i s r a e l i Glacier ..62 4.10 Echo strength versus depth for the Milne Glacier 63 4.11 D i s t r i b u t i o n of b a s a l l y - r e f l e c t e d power 66 4.12 F l i g h t l i n e map for the Milne Ice Shelf 67 4.13 Radar section from the Milne Ice Shelf 69 4.14 Ice thickness map for the Milne Ice Shelf 71 v i 4.15 Basal PRC of the Milne Ice Shelf 73 4.16 Fl i g h t l i n e map for the Ward Hunt Ice Shelf 74 4.17 Radar section from the Ward Hunt Ice Shelf 76 4.18 Ice-layer thickness of the Ward Hunt Ice Shelf 78 v i i ACKNOWLEDGEMENTS I thank my supervisor Dr. G.K.C. Clarke for his suggestions and advice over the long course of t h i s thesis. I especially thank him for the opportunity to work on this thesis; most aspects of i t have proven to be both interesting and challanging. I also thank my "surrogate supervisor", Dr. B.B. Narod, for his assistance during Dr. Clarke's one year sabbatical absence. I owe much of what I know about d i g i t a l signal processing and especially seismic processing to S. Levy and M. Lane, both of the University of B r i t i s h Columbia. Their p r a c t i c a l experience and detailed knowledge of seismic processing aided me greatly. The data I present in t h i s thesis were acquired in 1981. The f i e l d party consisted of M. Black and B. Haagensen of the Defence Research Board P a c i f i c , and Dr. Clarke, Dr. Narod and myself of UBC. I am grateful for the excellent f l i g h t support provided the project by the aircrew of Canadian Armed Forces a i r c r a f t 13803: Capt. P. Schoneberg, Capt. D. McGill, and M. Cpl. K. H i l l . The accuracy of the Milne Ice Shelf navigation owes much to the suggestions of M. J e f f r i e s of the University of Calgary. Dr. D. Crabtree generated the colour section. Needed proofreading and editing were supplied by M. Maxwell and Dr. Clarke. Dome Petroleum and a UBC graduate fellowship supported me over the last two years of t h i s thesis. Computing funds were provided by NSERC grant A4327, and l o g i s t i c a l support by a D.S.S. contract. v i i i 1. INTRODUCTION Magnetically-recorded radio echo sounding data can be d i g i t a l l y processed and enhanced. The University of B r i t i s h Columbia (UBC) 840 MHz airborne ice radar records the data i t acquires on analogue magnetic tape rather than on photographic fil m (as i s t y p i c a l ) ; the analogue tapes can be replayed through an analogue-to-digital converter to produce f u l l y d i g i t a l data. Signal processing and image enhancement techniques can then be applied to the data to make them easier to interpret. In the f i r s t part of t h i s thesis I discuss the signal processing techniques I apply to the UBC 840 MHz radar data. D i g i t a l processing to enhance radio echo sounding data i s not common because most radio echo sounders record the data they acquire on photographic f i l m . The fi l m records can be immediately interpreted for ice thickness, but cannot be enhanced to make r e f l e c t i o n s easier to locate. In the second part of t h i s thesis I discuss interpretation techniques for ca l c u l a t i n g ice thicknesses, power r e f l e c t i o n c o e f f i c i e n t s , and propagation loss rates from the UBC 840 MHz radar data. These techniques are now well established for photographically-recorded data (see for example Robin and others 1969; Neal 1976), and can be automated and applied to di g i t a l l y - r e c o r d e d data. In the last part of this thesis I discuss the results of applying the processing and interpretation techniques to radio echo sounding data from Ellesmere Island, N.W.T., Canada 1 2 ( F i g u r e 1 . 1 ) . A r c t i c C i r c l e 7 5 ° 3 5 ° 85° 7 5 ° A r c t i c C i r c l e F i g u r e 1 . 1 The study a r e a : E l l e s m e r e I s l a n d , N.W.T., Canada. I include ice thickness, power r e f l e c t i o n c o e f f i c i e n t , and propagation loss rate results from various areas on the northern coast of Ellesmere Island (Figure 1.2): the Ward Hunt and Milne Ice Shelves, the Milne and D i s r a e l i Glaciers, and a small ice cap surrounding Mt. Oxford (which I s h a l l refer to as the Mt. Oxford ice cap for s i m p l i c i t y ) . This thesis is a preliminary attempt at applying d i g i t a l signal processing techniques to radio echo sounding data. The techniques I use are simple because the large quantity of recorded data demands computationally-efficient processing. 3 F i g u r e 1.2 Areas sounded i n 1966 and 1981. In t h i s t h e s i s I d i s c u s s the r e s u l t s from the 1981 soundings ( t h e sounded a r e a s a r e shaded); H a t t e r s 1 e y - S m i t h and o t h e r s (1969) d i s c u s s i c e t h i c k n e s s e s from the 1966 soundings (denoted by s o l i d l i n e s ) . This s i m p l i c i t y disguises the sophisticated data-handling methods required to make even the most elementary processing f e a s i b l e . Because these data-handling methods depend upon the computer hardware and operating system, in t h i s thesis I do not discuss their implementation (see Prager 1982a and 1982b for d e t a i l s of how these methods are implemented at UBC). Advanced processing and interpretation techniques, and a closer examination of the northern Ellesmere Island data, would produce results more, accurate and detailed than the 4 preliminary ones presented in this thesis. 1.1 DATA PROCESSING AND RADIO ECHO SOUNDING L i t t l e processing is applied to analogue radio echo sounding data. The received sounding data are often d i f f e r e n t i a t e d to allow the higher frequency r e f l e c t i o n onsets (steep leading edges) to stand out from the lower frequency energy backscattered by r e f l e c t o r roughness and ice inclusions. The data are then recorded photographically using one of three formats: 1. A-scope. The returned power i s displayed on an oscilloscope and photographed. The A-scope photographs can later be hand d i g i t i z e d on an x-y d i g i t i z e r , but t h i s i s extremely slow so only a few traces can be d i g i t i z e d . 2. Z-scope. The trace i s swept down the oscilloscope screen with the returning power modulating the beam int e n s i t y . Photographic film i s exposed and moved across the oscilloscope screen forming a continuous, intensity modulated section. Large amounts of data can be recorded using the Z-scope format, and the sections produced can be immediately interpreted for ice thickness. Z-scope recording i s the most common format used with radio echo sounding because sections consisting of many traces can be plotted, and no further processing is necessary. 3. Echo strength measurement (ESM). Neal (1976) developed this technique to record the maximum returned power from an individual r e f l e c t o r . The oscilloscope time base i s 5 switched off and the returning power in a window around the r e f l e c t i o n of interest i s plotted v e r t i c a l l y . Photographic fi l m i s exposed and moved across the oscilloscope screen; the envelope of the image w i l l be the maximum returned power in the window. ESM can be used to calculate the power r e f l e c t i o n c o e f f i c i e n t of the re f l e c t o r i f Z-scope or A-scope recordings are made simultaneously to allow range determinations. Watts and Wright (1981) used a sampling oscilloscope to record radio echo sounding data (from the Columbia Glacier, Alaska) on magnetic tape. Before sampling and recording their data they bandpass f i l t e r e d them and applied a time-varying gain to enhance l a t e - a r r i v i n g low-amplitude r e f l e c t i o n s . Watts and Wright did not d i g i t i z e and further process their data, but replayed the magnetic tapes through an oscilloscope to generate standard Z-scope sections. Processing hand-digitized A-scope records has been discussed by Neal (1977) and M i l l a r (1981). They did not process their data to make r e f l e c t i o n s clearer, but analyzed r e f l e c t o r and ice c h a r a c t e r i s t i c s : power r e f l e c t i o n c o e f f i c i e n t , r e f l e c t o r roughness, d i e l e c t r i c p e r m i t t i v i t y and conductivity. One cannot e a s i l y obtain this type of information from processed data because the processing af f e c t s the r e f l e c t i o n s . I therefore use processed data to locate r e f l e c t i o n s , and then use the located r e f l e c t i o n s to re-examine the o r i g i n a l data. 6 The UBC 840 MHz ice radar uses a sampling time base, analogous to that used by sampling oscilloscopes, to "strobe" the incoming pulses at 10 kHz. Only one sample per strobe is taken; hence the data can be recorded on audio magnetic tape. These tapes can be replayed through an analogue-to-digital converter to produce f u l l y d i g i t a l radar data. D i g i t a l processing similar to seismic processing used by companies exploring for o i l (for example see Kanasewich 1975) can then be applied to enhance the d i g i t a l radar data. Seismic processing reduces noise, increases resolution, and migrates r e f l e c t i o n s so that sections indicate the true s p a t i a l relationships of the r e f l e c t o r s . Ultimately, a l l these seismic processing techniques should be t r i e d on radio echo sounding data, but only noise reduction and simple resolution-increasing techniques are discussed in t h i s thesis. 1.2 GLACIOLOGICAL STUDIES ON NORTHERN ELLESMERE ISLAND Pe r i o d i c a l l y , huge "ice islands" calve from the ice shelves which fringe northern Ellesmere Island (the Ward Hunt Ice Shelf lost almost 600 km2 of i t s area during the winter of 1962; Hattersley-Smith 1963). These calvings are of considerable p r a c t i c a l interest because the ice islands they produce can floa t southwards through the Beaufort Sea, posing a hazard to offshore d r i l l i n g r i g s . The Ward Hunt Ice Shelf, the largest shelf, was extensively studied during the late 1950's and early 1960's. Leveling and seismic surveys show that the average thickness 7 is 43 m (see for example Crary 1958 and 1960). The shelf is composed of an upper layer consisting of iced f i r n and lake ice which overlies a basement layer consisting of i n t e r d i g i t a t i n g brackish and sea ice (see for example Marshall 1955; Lyons and Ragle 1962; Lyons and others 1971 and 1972). The lake ice layer unconformably overlies the basement ice in a broad syncline. In some areas a heavy concentration of accumulated dust and debris marks the interface. Ragle and others (1964) measured the d i e l e c t r i c properties of ice cores taken from the Ward Hunt Ice Shelf. They found that the basement ice layer has a high conductivity (compared to the lake ice layer or a glacier) because i t contains s a l t . A basal layer of saline or brackish ice would detrimentally a f f e c t radio echo sounding for two reasons: a radio wave propagating through the high-conductivity ice would be severely attenuated; and the r e f l e c t i v i t y at the ice/sea water interface would be low because the salt would decrease the d i e l e c t r i c contrast. Ice on northern Ellesmere Island was f i r s t radio echo sounded in 1966 (Hattersley-Smith and others 1969). In this experiment, the Ward Hunt Ice Shelf, the Milne and D i s r a e l i Glaciers and several g l a c i e r s and ice caps inland from the northern coast were sounded. The 1966 soundings show that the Milne and D i s r a e l i Glaciers vary between 0 - 700 m thick, and that the Ward Hunt Ice Shelf varies between 20 - 90 m thick. The power r e f l e c t e d from the ice bottom increases by 10 dB for the tongues of the Milne and D i s r a e l i Glaciers, a fact -which 8 suggests that they are f l o a t i n g . 2. PROCESSING DIGITAL RADIO ECHO SOUNDING DATA 2.1 INTRODUCTION In th i s chapter I discuss processing techniques which can be applied to d i g i t a l radar data to make r e f l e c t i o n s more easi l y i d e n t i f i e d . Processing objectives can be broadly c l a s s i f i e d into three categories: (i) reducing noise to make re f l e c t i o n s clearer; ( i i ) making the re f l e c t i o n s sharper (increasing resolution); ( i i i ) moving r e f l e c t i o n s so that sections indicate the true s p a t i a l relationships of the r e f l e c t o r s . In this thesis I l i m i t discussion to noise reduction and simple resolution-increasing techniques. 2.2 RECORDING THE DATA The transmitted power returned to the receiver i s passed through a logarithmic amplifier with an 80 dB dynamic range, then sampled and stored on four-track analogue magnetic tape (see Table 2.1 for system parameters; Narod 1979, and Narod and Clarke 1983 describe the UBC 840 MHz radar f u l l y ) . Various data are stored on the four tracks: 1. Channel 1. IRIG time code marking hours, minutes, seconds, and decimal seconds. This time is displayed on the radar system clock and can be concurrently imprinted on a e r i a l photographs to allow the photographs and radar data to be accurately correlated. 2. Channel 2. The actual radar data. 3. Channel 3. Synchronization code indicating when the radar 9 Table 2.1 System parameters for the UBC 840 MHz radar 10 System Performance Transmi t t e r : Power output Frequency Repetition rate Pulse length Pulse r i s e time Pulse f a l l time Rece iver: Bandwidth Dynamic range Recovery from saturat ion Antenna: Forward gain E-plane beam width (3 dB) H-plane beam width (3 dB) 1 26 4. 1 840 10 50 18 28 40 80 100 15.5 18 44 dB kW MHz kHz ns ns ns MHz dB ns dB deg deg Narod and Clarke (1983) present a f u l l l i s t of system parameters. data should be d i g i t i z e d . 4. Channel 4. Voice track for navigational and other comments. To sample the 40 MHz receiver bandwidth without a l i a s i n g requires an 80 MHz sampling frequency. Low power, portable, 11 inexpensive f i e l d equipment cannot sample at t h i s rate. Even i f the data could be sampled at 80 MHz, i t would be extremely d i f f i c u l t to record them completely without pre-processing. To avoid this problem, the data are sampled by strobing the incoming pulses in a manner analogous to that used by sampling oscilloscopes. The complete sounding record (trace) is formed by taking only one sample per radar pulse from many pulses. The f i r s t sample is taken at 10 ns on the f i r s t pulse, the second at 20 ns on the second pulse, and so on. Traces of either 512 points (5.12 MS) or 1024 points (10.24 <us) can be generated. The radar pulse repetition rate i s 10 kHz; therefore at a 10 ns sample separation, the apparent sampling rate i s 100 MHz while the actual sampling rate i s 10 kHz. To sample the data by strobing requires the input waveform to be r e l a t i v e l y constant over the time needed to acquire one trace (50 or 100 ms). If the a i r c r a f t airspeed i s 60 m/s and 1024-point traces are being taken, the l a s t point of a trace is taken 6 m farther along track than the f i r s t . Strobing the data does not cause d i f f i c u l t y because over the 6 m required to obtain one trace the ice does not change much, and even i f i t did, the radar illumination pattern on the r e f l e c t o r (footprint) would s p a t i a l l y f i l t e r the data. The fading pattern (variation in r e f l e c t i o n strength resulting from interference caused by r e f l e c t o r roughness and ice inclusions) w i l l be undersampled, but the s t a t i s t i c a l d i s t r i b u t i o n of r e f l e c t i o n strengths in the fading pattern 12 w i l l be correct (Narod and Clarke 1983). Hence the strobing does not af f e c t either ice thickness or r e f l e c t i v i t y computat ion. The analogue magnetic tapes are returned to the laboratory where they are d i g i t i z e d and the data stored on nine-track computer tape. I add a 52-byte header to each data trace to store various trace parameters: the trace number; the number of points contained in the trace; the IRIG time code; flags for i n c o r r e c t l y d i g i t i z e d traces and c a l i b r a t i o n traces; the geographical location; the point-number of the transmit and surface pulses; and the point-number of any v i s i b l e r e f l e c t i o n s . These headers are updated as the processing and interpretation proceeds. The radar system allows the operator to put a stable 1 MHz sawtooth calibration-pulse into the sampler and hence onto the magnetic tape, allowing accurate determination of the sampling time base. Subsequent processing skips flagged c a l i b r a t i o n and incorrectly d i g i t i z e d traces. 2.3 PROCESSING TO PRODUCE INTERPRETABLE SECTIONS Radar sections are images containing both signal ( r e f l e c t i o n s ) and noise (any unwanted signal which degrades the image). The processing i s adjusted to reduce noise as much as possible while maintaining good resolution. The noise has various sources: tape hiss; low frequency backscattered energy; receiver saturation; multiple r e f l e c t i o n s from energy which reverberates several times between r e f l e c t o r s and obscures deeper r e f l e c t i o n s . By reducing t h i s noise I can see 13 the r e f l e c t i o n s in the data more e a s i l y . Noise and signal cannot be t o t a l l y separated; any f i l t e r i n g applied to remove noise also f i l t e r s the signal, reducing the image resolution. For example, i f I reduce high-frequency tape hiss by low-pass f i l t e r i n g , high-frequency content is removed from r e f l e c t i o n s thus broadening them. Two c l o s e l y spaced, yet resolvable, r e f l e c t i o n s could merge into one broad r e f l e c t i o n . A good processing scheme increases the signal-to-noise r a t i o (SNR) and maintains good image resolut ion. I generally use linear high-pass and low-pass f i l t e r s to reduce noise. These f i l t e r s must be simple and computationally inexpensive because of the large •number of points to be processed. If one had a large computing budget or a small data set, more sophisticated and expensive f i l t e r s could be applied. Occasionally, noise reduction alone does not make re f l e c t i o n s easier to i d e n t i f y . For example, i f the power of low-amplitude r e f l e c t i o n s needs to be increased r e l a t i v e to the power of high-amplitude r e f l e c t i o n s (dynamic range reduction), I use nonlinear automatic gain control (AGC). With reduced dynamic range, low-amplitude r e f l e c t i o n s are easier to i d e n t i f y . Careful linear f i l t e r i n g i s necessary i f AGC is used because AGC degrades the SNR. Throughout this section I discuss in a q u a l i t a t i v e manner the frequency content of radar traces. A clear r e f l e c t i o n consists of roughly 20 data points: I c a l l signals which contain more than 20 data points low frequency, and signals which contain less than 20 data points high frequency. This terminology is convenient because i t allows one to more eas i l y v i s u a l i z e the effects of the f i l t e r s . S t r i c t l y speaking, frequencies cannot be this e a s i l y separated because r e f l e c t i o n s contain energy at a l l frequencies between the DC and Nyquist frequencies. The preliminary processing steps I can apply to enhance a radar section are as follows: 1. Stacking. This is a trace-to-trace moving average applied across the section. Reflections are coherent across at least several traces and are therefore emphasized. 2. High-pass f i l t e r i n g . This i s used to remove the low-frequency noise caused by r e f l e c t o r roughness and ice inclusions. 3. D i f f e r e n t i a t i n g . This is a vigourous high-pass f i l t e r used to convert leading edges of r e f l e c t i o n s (onsets) into spikes which precisely locate the onsets. 4. Automatic gain control. This i s used i f r e f l e c t i o n s are small and barely v i s i b l e . AGC reduces trace dynamic range; small r e f l e c t i o n s are made v i s i b l e because they are amplified r e l a t i v e l y more than large r e f l e c t i o n s . 5. Low-pass f i l t e r i n g . This i s used to reduce the high-frequency content which both AGC and d i f f e r e n t i a t i n g emphasize. Unless AGC is used, the processing sequence i s ir r e l e v a n t . -Because AGC i s nonlinear, di f f e r e n t results are 1 5 obtained i f i t s position in the processing sequence is changed; the other f i l t e r s are linear and can be applied in any order i f AGC is not used. In the section on AGC I discuss where t h i s operation should be applied. After processing, the data are ready to be plotted. I have five display options: (i) A-scope for examination of r e f l e c t i o n shape or quality-control monitor sections; ( i i ) variable area (A-scope with positive peaks f i l l e d in) to locate r e f l e c t i o n s ; ( i i i ) grey-scale as the general purpose interpretation section; (iv) threshold to show large r e f l e c t i o n s in good data; (v) colour l e v e l to show the strength of r e f l e c t i o n s in a section. 2.3.1 STACKING Because of the trace overlap introduced by the radar footprint and slowly varying ice changes, two adjacent traces should be similar except for the added noise. Applying a moving average (stacking) across the section reduces this noise and emphasizes r e f l e c t i o n s . Stacking does not reduce noise which i s coherent across the section (for example, multiple r e f l e c t i o n s ) , because stacking i s a low-pass f i l t e r applied across the section. Stacking i s e f f e c t i v e because f l a t - l y i n g r e f l e c t i o n s have large low-frequency content l a t e r a l l y . Dipping r e f l e c t i o n s do not stack well. The width of the moving average (stack fold) can be adjusted to either maintain good l a t e r a l resolution (small stack fold) or greatly reduce noise (large stack f o l d ) . 16 Analogue Z-scope recordings are often printed with traces overlapped, resulting in an integration on the recording fi l m (Evans and Smith 1969). I c a l l my moving average a stack rather than an integration because of i t s s i m i l a r i t y to the seismic stack where traces with a common true signal but diff e r e n t added noise are averaged together. Adjacent radar traces can be considered to contain a common signal because they are s p a t i a l l y f i l t e r e d by the radar footprint. The radar footprint on a r e f l e c t o r is an 18° x 44° e l l i p s e . At a 200 m a i r c r a f t t e r r a i n clearance, a 10 trace/s acqu i s i t i o n rate, and a 60 m/s a i r c r a f t speed, one radar footprint overlaps 60 m along track, comparable to s p a t i a l l y f i l t e r i n g over 11 traces. Increasing the SNR by widening the stack beyond about 15-fold i s often unproductive because the section becomes excessively f i l t e r e d . For example, an n-fold stack w i l l reduce the amplitude of Gaussian noise by /n", but also reduce the amplitude of any re f l e c t i o n s coherent over fewer than n traces. Stacks wider than 15-fold increase the SNR by removing dipping and l a t e r a l l y - s h o r t r e f l e c t i o n s . To locate deep re f l e c t o r s the stack f o l d can be increased because geometrical spreading increases the radar footprint. Selecting the stack fold depends upon what one wants to emphasize: unstacked data most c l e a r l y shows d i f f r a c t i o n s ; data stacked over less than one radar footprint reduces noise but does not broaden r e f l e c t i o n s ; heavily stacked data gives broad averages or emphasizes weak horizontal layers. I generally use 9-fold stacks because this width does not 1 7 s p a t i a l l y f i l t e r the section excessively, yet substantially reduces the noise. The low-pass f i l t e r i n g caused by stacking can be observed in Figure 2.1. unstacked 9-fold 25-fold 101-fold F i g u r e 2.1 S t a c k i n g . As t h e s t a c k f o l d i n c r e a s e s , b o t h the n o i s e and the r e s o l u t i o n d e c r e a s e . If the re f l e c t o r of interest i s steeply dipping or the a i r c r a f t range i s rapidly changing, aligning the r e f l e c t i o n of interest before stacking can be advantageous. Alignment removes l a t e r a l high-frequency content; thus the sp a t i a l f i l t e r i n g r e s u l t i n g from stacking does not severely af f e c t the r e f l e c t i o n . For my data this alignment i s generally not necessary i f the stack f o l d i s kept small (<15). 18 2.3.2 HIGH-PASS FILTERING The rate at which the power decays after the pulse no longer illuminates the r e f l e c t o r is known as the fading rate; this rate is a function of the r e f l e c t o r roughness. The fading introduces low-frequency content into the data which can be reduced by high-pass f i l t e r i n g . Removing low-frequency noise from a section s i m p l i f i e s r e f l e c t i o n i d e n t i f i c a t i o n , occasionally enhances small r e f l e c t i o n s obscured by backscattered energy, and allows the use of variable area and threshold p l o t t i n g formats. To high-pass f i l t e r the data I f i r s t smooth them with a moving average over a s p e c i f i e d time window, then subtract th i s smoothed version from the o r i g i n a l data. This f i l t e r i s computationally inexpensive and works moderately well. It causes negative lobes of one-half the averaging window width on both sides of large r e f l e c t i o n s . These lobes can obscure small r e f l e c t i o n s , but the averaging window width can be adjusted to minimize the e f f e c t . Small r e f l e c t i o n s superimposed on backscattered energy high-pass f i l t e r well (Figure 2.2). 2.3.3 DIFFERENTIATING D i f f e r e n t i a t i n g is a vigorous high-pass f i l t e r useful for locating r e f l e c t i o n onsets. In my data shallow r e f l e c t i o n s (<400 m) are generally characterized by sharp onsets which d i f f e r e n t i a t i n g turns into spikes. Deeper r e f l e c t i o n s (>400 m) are somewhat broadened by noise, making d i f f e r e n t i a t i n g less 19 I 1 I 1 l-° unfiltered 2.0 us 1.0 us 0.4 us F i g u r e 2.2 High-pass f i l t e r i n g . The u n f i l t e r e d t r a c e was f i r s t 9 - f o l d s t a c k e d . The time underneath the t r a c e i s the width of the moving average used to produce the smoothed t r a c e which i s s u b s e q u e n t l y s u b t r a c t e d from the o r i g i n a l t r a c e . e f f e c t i v e although i t s t i l l works. I d i f f e r e n t i a t e by convolving (1,0,-1) with the data. Scaling i s unimportant because the section i s being processed to show r e f l e c t i o n locations, and the derivative is not an indication of returned power. D i f f e r e n t i a t i n g increases the high-frequency content of a trace r e l a t i v e to the low-frequency content; t h i s e f f e c t can be seen by examining the f i l t e r transfer function. The transfer function in the Fourier domain i s a straight l i n e with unity slope passing through the o r i g i n (A(u) = -ico; A(CJ) i s ' the complex Fourier component of the transfer function at angular frequency CJ) . The DC component (w=0) is eliminated; the low-frequency 20 content (|w|<1)) i s reduced; and the high-frequency content (|GJ|>1) i s increased. The SNR at high frequencies is generally low, so d i f f e r e n t i a t i n g emphasizes high-frequency noise (Figure 2.3) which can subsequently be reduced by low-pass f i l t e r i n g . unfiltered differentiated F i g u r e 2.3 D i f f e r e n t i a t i n g . The u n f i l t e r e d t r a c e was f i r s t 9 - f o l d s t a c k e d . 2.3.4 AUTOMATIC GAIN CONTROL Set to FM recording at 15 ips (38.1 cm/s), the SNR of the Racal tape recorder is 48 dB (Racal Technical Handbook 1978). The maximum voltage that can be recorded (0 dB level) w i l l be 250 times above the -48 dB noise l e v e l . If I plot a 0 dB signal at a 1 cm deflec t i o n , a -48 dB signal w i l l produce an 21 unobservable 0.004 cm defl e c t i o n . Clearly the dynamic range must be reduced i f a l l r e f l e c t i o n s are to be seen on the same section. Automatic gain control (AGC) reduces dynamic range. I generate a trace containing either the root mean squared (RMS) or absolute value of the trace being processed, then smooth thi s positive-valued gain trace with one or two moving averages, and f i n a l l y divide the o r i g i n a l trace by the smoothed gain trace. The width of the moving-average window is adjusted to give the desired dynamic range reduction. If the smoothed average is small, the resulting normalization i s large; i f the smoothed average i s large, the resulting normalization is small. Consequently, large peaks are reduced, and small peaks are amplified (Figure 2.4). Watts and Wright (1981) applied time-varying gain (TVG) to their Columbia Glacier data to amplify small, late-occurring r e f l e c t i o n s . With their TVG small r e f l e c t i o n s which occur near the trace beginning are not amplified. Although t h e i r TVG does not suffer from some of the problems of my AGC, I use the AGC because i t is e f f e c t i v e over the entire trace. The major problem with AGC i s that a small r e f l e c t i o n occurring near a large one w i l l be decreased rather than amplified i f the smoothing window centred on the small r e f l e c t i o n overlaps the large. The large power causes "dead zones" of one-half window width on either side of large r e f l e c t i o n s . Noise trapped between two r e f l e c t i o n s separated 22 unfiltered 2.0 us 1.0 us 0.4 us F i g u r e 2.4 Automatic g a i n c o n t r o l . The u n f i l t e r e d t r a c e was f i r s t 9 - f o l d s t a c k e d . then h i g h - p a s s f i l t e r e d u s i n g a 0.10>s window. The time underneath the t r a c e i s the width of the moving average used to smooth the g a i n t r a c e . by s l i g h t l y more than one window width can often appear as re f l e c t i o n s because of the dead zones which bracket i t . On sections where no r e f l e c t i o n s or only faint r e f l e c t i o n s can be seen, the AGC should be t r i e d , but always cautiously. AGC-induced a r t i f a c t s can be i d e n t i f i e d because the a r t i f a c t follows, at one-half the window width, the r e f l e c t i o n which causes i t . The AGC is nonlinear and hence controls the processing flow. Low frequencies interfere with the action of the AGC so I generally high-pass f i l t e r the data beforehand. Because the AGC emphasizes high-frequency noise and re-enhances previously suppressed noise, I low-pass f i l t e r after AGC. Stacking i s a 23 form of low-pass f i l t e r i n g but must be performed before AGC because the assumption underlying stacking i s that the data contain a common signal plus added noise; AGC modifies a trace using only information contained in that individual trace, thus traces may no longer contain a common signa l . 2.3.5 LOW-PASS FILTERING Low-pass f i l t e r i n g reduces the high-frequency noise emphasized by the AGC and d i f f e r e n t i a t i o n . I low-pass f i l t e r by applying a moving average twice down the trace. This i s equivalent to convolving the data with a Ba r t l e t t window (a triangular waveform; see Kanasewich 1975, p. 109). I use a Bar t l e t t window because i t i s computationally inexpensive and does not remove a l l high-frequency content. Some high-frequency content i s needed to maintain good image resolut ion. Sections which have not been d i f f e r e n t i a t e d or automatic gain controlled generally do not need low-pass f i l t e r i n g . Excessive low-pass f i l t e r i n g decreases the image resolution by broadening r e f l e c t i o n s , or blending closely spaced r e f l e c t i o n s together (Figure 2.5). 2.3.6 DISPLAY Sections are plotted with the r e f l e c t i o n a r r i v a l time increasing v e r t i c a l l y downwards to a maximum of 10.24 jus, and the trace acq u i s i t i o n time increasing h o r i z o n t a l l y . I can plot a section fiv e ways (Figures 2.6, 2.7, and 2.8): 24 unfiltered 0.03 us 0 .09 us 0.25 us i g u r e 2.5 Low-pass f i l t e r i n g . The u n f i l t e r e d t r a c e i s the 1 .0„s AGC f i l t e r e d t r a c e from F i g u r e 2.4. The time underneath the t r a c e i s the width of the B a r t l e t t window used to f i l t e r the t r a c e . A-scope (Figure 2.6a). The returning power is plotted horizontally, the return time v e r t i c a l l y . A-scope sections are inexpensive to plot and work well on unprocessed data as a v i s u a l check on data q u a l i t y . Sparse A-scope sections often provide a means of i n i t i a l l y i d e n t i f y i n g r e f l e c t i o n s ; the gaps between traces can be f i l l e d by examining more densely or d i f f e r e n t l y plotted sections. Variable area (Figure 2.6b). A-scope traces are plotted with the p o s i t i v e peaks shaded i n . The variable area format emphasizes r e f l e c t i o n s and allows the interpreter to correlate pulse shape in addition to pulse amplitude. Variable area p l o t t i n g does not work well i f the data have 25 a DC bias or much low-frequency content; the data must be f i r s t d i f f e r e n t i a t e d or high-pass f i l t e r e d . If the dynamic range i s large I can AGC the data or plot them with large amplitude peaks overlapped, to observe the smaller amplitude peaks. This format would work well for data acquired by impulse radio echo sounders which record returned signal amplitude and phase (rather than only returned signal power) because signal phase could be correlated across the section in addition to signal strength. Variable area is my preferred format for plo t t i n g d i f f e r e n t i a t e d data. 3. Threshold (Figure 2.6c). Data values above a pre-selected threshold are connected by v e r t i c a l segments, values below the threshold remain unconnected. Threshold p l o t t i n g is economical on dot matrix or e l e c t r o s t a t i c p l o t t e r s , but unsuitable for pen pl o t t e r s because so many traces are plotted and the pen is l i f t e d and set down so often. On good data with clear r e f l e c t i o n s the p l o t t i n g threshold can be adjusted to make r e f l e c t i o n s stand out c l e a r l y . Threshold plots are v i s u a l l y s t r i k i n g ; unfortunately this format i s sensitive to threshold selection and does not F i g u r e 2.6 ( o v e r l e a f ) A-scope, v a r i a b l e a r e a , and t h r e s h o l d d i s p l a y s . The s e c t i o n i s from the M i l n e Ice S h e l f ( see F i g u r e s 4.12 and 4.13). The t r a n s m i t p u l s e a r r i v e s at 0.25,,s, the s u r f a c e r e f l e c t i o n at 1.0 | /S , and the bottom r e f l e c t i o n at 2.0»,s. R e f l e c t i o n s below 2.0„s a r e m u l t i p l e r e f l e c t i o n s ( s e e F i g u r e 3.2). Each s e c t i o n was p r o c e s s e d d i f f e r e n t l y : a) A-scope. 9 - f o l d s t a c k e d . Every 100th t r a c e p l o t t e d . b) V a r i a b l e a r e a . 9 - f o l d s t a c k e d , then h i g h - p a s s f i l t e r e d . Every 25th t r a c e p l o t t e d . c) T h r e s h o l d . 9 - f o l d s t a c k e d , h i g h - p a s s f i l t e r e d , then p l o t t e d w i t h a 5% t h r e s h o l d ( r e l a t i v e t o the maximum a m p l i t u d e ) . Every 5 t h t r a c e p l o t t e d . t w o - w a y t i m e (us ) t w o - w a y t i m e ( u s ) t w o - w a y t i m e ( u s ) 27 show any r e f l e c t i o n s below the threshold. 4. Grey scale. The returning power is assigned a grey l e v e l and plotted in a manner analogous to Z-scope sections. A grey-scale plot i s a threshold plot with multiple grey l e v e l s , but the grey-scale allows one to see small r e f l e c t i o n s which the threshold might cut o f f . I use a Versatec V-80 e l e c t r o s t a t i c p l o t t e r : t h i s p lotter has a 200 dot/inch resolution which allows me to plot up to 512 points/trace v e r t i c a l l y on 11 inch paper (4 dots/point are needed to generate the grey sc a l e ) . Grey-scale p l o t t i n g works best on sections which have been l i g h t l y high-pass f i l t e r e d to reduce any large DC bias. Automatic gain controlled sections do not grey-scale plot well because the reduced dynamic range causes the section to look l i k e a threshold p l o t . The grey levels can be selected to either emphasize r e f l e c t i o n locations (for example see Figure 4.13), or show the true returned power (Figure 2.7). Grey scale i s my preferred format for p l o t t i n g undifferentiated data. 5. Colour l e v e l (Figure 2.8). A colour i s assigned to each power l e v e l . Specialized image processing equipment i s F i g u r e 2.7 ( o v e r l e a f ) G r e y - s c a l e d i s p l a y . The s e c t i o n i s from the M i l n e Ice S h e l f (see F i g u r e s 4.12 and 4.13). A l l system, g e o m e t r i c a l s p r e a d i n g , and p r o p a g a t i o n l o s s e s have been removed assuming a p r o p a g a t i o n l o s s r a t e of 0.04 dE/m. Only the PRC remains and i s d i s p l a y e d as a g r e y l e v e l . The white p i x e l s at a p p r o x i m a t e l y 1.8j,s i d e n t i f y the bottom r e f l e c t i o n o n s e t . 28 I r e f l e c t i o n s t r e n g t h (dB) o CM I O C O I o I o 1 0 o l o I I I I I I I I I I I I I I I I I I I I T CM I I I I I I i—r co (Sn) dUJIl |©ABJi A B M - O M I 29 needed to produce these sections, and reproduction in journals, i f possible, i s generally c o s t l y . Colour-level sections are p a r t i c u l a r l y useful i f the data are processed to r e f l e c t some physical property of the ice (power r e f l e c t i o n c o e f f i c i e n t for example). It i s much easier to see the value of a point on a c o l o u r - l e v e l section than on a grey-scale section. Colour-level sections are no more useful than grey-scale sections for identifying r e f l e c t i o n s because r e f l e c t i o n s are i d e n t i f i e d by changes in returned power rather than magnitude. F i g u r e 2.8 ( o v e r l e a f ) C o l o u r r e f l e c t i v i t y d i s p l a y . The s e c t i o n i s from the M i l n e Ice S h e l f (see F i g u r e s 4.12 and 4.13). A l l system, g e o m e t r i c a l s p r e a d i n g , and p r o p a g a t i o n l o s s e s have been removed assuming a p r o p a g a t i o n l o s s r a t e of 0.04 dB/nt. Only the PRC remains and i s d i s p l a y e d as a c o l o u r l e v e l . The b l a c k p i x e l s at a p p r o x i m a t e l y 1 . 8,,s I d e n t i f y the bottom r e f l e c t i o n o n s e t . 3. INTERPRETING A DIGITAL RADAR SECTION 3.1 INTRODUCTION An interpreter stops reprocessing a radar section once clear r e f l e c t i o n s are obtained, or i t becomes clear that none w i l l ever be. By interpreting I mean locating r e f l e c t i o n s in the radar section and c a l c u l a t i n g their depth, then possibly c a l c u l a t i n g r e f l e c t o r and ice c h a r a c t e r i s t i c s from these located r e f l e c t i o n s : power r e f l e c t i o n c o e f f i c i e n t (PRC) and the ice propagation loss rate (LR). The "art" of locating r e f l e c t i o n s is beyond the scope of t h i s thesis. Locating r e f l e c t i o n s in good data i s straightforward; locating r e f l e c t i o n s in poor data takes practice. In this chapter I f i r s t discuss some problems which make r e f l e c t i o n s d i f f i c u l t to locate, then discuss how I calculate ice thickness, power r e f l e c t i o n c o e f f i c i e n t s , and propagation loss rates for the UBC 840 MHz radar data. Robin and others (1969) discuss the interpretation of radio echo sounding data in d e t a i l . 3.2 LOCATING REFLECTIONS Once the r e f l e c t i o n s have been located I colour their onset times with an erasable crayon. The locations of these li n e s are then d i g i t i z e d on an x-y d i g i t i z e r ; the program driving the d i g i t i z e r returns the x-y coordinates as trace number/point number pair s . Points between the d i g i t i z e d locations are l i n e a r l y interpolated to serve as a guide for a 31 32 computer program which automatically "fine tunes" these locations. The computer program fine tunes by searching the data in a window surrounding the interpolated point for the maximum amplitude or maximum slope. The location of the maximum i s the new location and i s stored in the data trace header. I use the maximum slope to locate r e f l e c t i o n onsets for ice thickness calculations, and the maximum amplitude to locate r e f l e c t i o n maxima for PRC and propagation loss c a l c u l a t i o n s . The transmit pulse and surface r e f l e c t i o n are always located; the bottom and internal r e f l e c t i o n s are located where v i s i b l e . I have developed algorithms which automatically track r e f l e c t i o n s across a section. These algorithms use the r e f l e c t i o n position from the previous trace as the centre of a search for the maximum amplitude or slope in the search window. Constraining the difference between loc a t e d - r e f l e c t i o n positions on adjacent traces to 1-3 points reduces fading-induced " j i t t e r " in the location. This simple algorithm works well on clear horizontal r e f l e c t i o n s ; more robust algorithms are needed to r e l i a b l y locate dipping or fai n t r e f l e c t ions. 3.2.1 EFFECTS WHICH OBSCURE REFLECTIONS Three types of eff e c t s can obscure r e f l e c t i o n s : 1. System effects caused by lim i t a t i o n s of the radar system. 2. A r t i f a c t s introduced by the processing. 3. Physical effects associated with the physics of 33 electromagnetic wave propagation through ice. The major system effect of the UBC 840 MHz radar is receiver dead time caused by c l i p p i n g . To obtain maximum penetration where the ice is either deep or lossy, a i r c r a f t t e r r a i n clearance i s kept low; t h i s proximity can cause the returning power to saturate the receiver. It takes roughly 230 ns for the system to recover from saturation and f a i t h f u l l y record the data (Narod and Clarke 1983). Once the data have been clipped and recorded there i s l i t t l e that subsequent processing can do to recover the lost information. Processing a section necessarily introduces a r t i f a c t s . A good processing scheme maximizes the ease with which the re f l e c t i o n locations can be located while minimizing the processing a r t i f a c t s . Sections with only faint r e f l e c t i o n s are heavily processed to make them v i s i b l e ; occasionally this causes processing a r t i f a c t s to appear as r e f l e c t i o n s because of the absence of true r e f l e c t i o n s . Both high-pass f i l t e r s and AGC cause "dead zones" surrounding large r e f l e c t i o n s . These dead zones can either obscure r e f l e c t i o n s , or cause noise trapped between two dead zones to appear as a r e f l e c t i o n . From sparsely-plotted unprocessed A-scope , sections the interpreter can determine which r e f l e c t i o n s are r e a l , then find these r e f l e c t i o n s in densely-plotted processed sections and continue the interpretation between the A-scope traces. Various physical effects make r e f l e c t i o n s d i f f i c u l t to locate: c r i t i c a l r e f raction at the i c e / a i r interface, dipping 34 r e f l e c t o r s , d i f f r a c t i o n s , focussing and defocussing by undulating r e f l e c t o r s , multiple r e f l e c t i o n s , and r e f l e c t i o n s from valley walls. These e f f e c t s are the same for d i g i t a l and analogue radio echo soundings because the physics and geometry of the two methods are i d e n t i c a l ; only data sampling and storage techniques are d i f f e r e n t . I w i l l give a brief outline of these physical effects (see for example Robin and others 1969, and Harrison 1970 for a more detailed discussion). C r i t i c a l r e f r a c t i o n , or t o t a l internal r e f l e c t i o n , occurs for rays t r a v e l l i n g upwards from the ice to the a i r at angles of incidence greater than 34°. No energy is transmitted through the interface, a l l the energy r e f l e c t s downward. Hence no r e f l e c t i o n s from slopes greater than 34° can be observed. For r e f l e c t o r s dipping less steeply than 34°, timing errors occur. For a coincident transmitter and receiver pair, received r e f l e c t i o n s must have been normally incident on the r e f l e c t o r (this i s only s t r i c t l y true for smooth r e f l e c t o r s ; backscattered energy can come from any point on rough r e f l e c t o r s ) . For a dipping r e f l e c t o r the normally incident point is not d i r e c t l y underneath the antenna as displayed on a time section (time sections display pulse return time rather than r e f l e c t o r depth), but actually a point up-dip from the nadi r. Removing this effect from an entire section to aid interpretation i s c a l l e d migration because r e f l e c t i o n s are "migrated" to their true s p a t i a l relationship in the section. As well, migration "collapses" d i f f r a c t i o n hyperbolae into the 35 scattering-points which produced them. Harrison (1970) presented a method to move located r e f l e c t i o n s to their true s p a t i a l relationship; the method moves only the p r o f i l e s of located r e f l e c t i o n s and does not migrate the entire radar section. Occasionally Harrison's method i s c a l l e d deconvolution which is somewhat unfortunate: in the seismic industry the technique would be considered a simple form of migration; the term deconvolution is used to denote the mathematical inverse of convolution. One type of deconvolution used by the seismic industry removes the effect of the sounding source from a seismic trace; the recorded trace is assumed to result from the convolution of the source wavelet with the r e f l e c t i v i t y structure of the earth. Migration also reduces d i f f r a c t i o n and focussing e f f e c t s caused by r e f l e c t o r d i s c o n t i n u i t i e s and undulations. D i f f r a c t i o n patterns are associated with d i s c o n t i n u i t i e s (Figure 3.1) and can aid in locating and i d e n t i f y i n g these d i s c o n t i n u i t i e s . Occasionally though, the d i f f r a c t i o n patterns can blend with true r e f l e c t i o n s making the r e f l e c t i o n s d i f f i c u l t to i d e n t i f y . The small radar footprint makes focussing e f f e c t s unimportant for the UBC 840 MHz radar. Two types of multiple r e f l e c t i o n s can confuse a section: a i r c r a f t multiples for which energy reverberates two or more times between a r e f l e c t o r and the a i r c r a f t , and internal multiples for which energy reverberates between two ice r e f l e c t o r s (Figure 3.2). These multiple r e f l e c t i o n s can obscure true r e f l e c t i o n s . Multiple r e f l e c t i o n s are easily 36 F i g u r e 3.1 D i f f r a c t i o n p a t t e r n s . The heavy l i n e s a r e e x p e c t e d d i f f r a c t i o n s from: a) bottom c r e v a s s e , b) s u r f a c e c r e v a s s e , c) unsoundable b l o c k , perhaps caused by b r i n e i n f i l t r a t i o n , d) s t e p d e c r e a s e i n i c e t h i c k n e s s . The shaded p o r t i o n i s i m p e n e t r a b l e Ice, rock, or water. aircraft Internal F i g u r e 3.2 M u l t i p l e r e f l e c t i o n s . T i s the t r a n s m i t p u l s e , S Is the s u r f a c e r e f l e c t i o n , B i s the bottom r e f l e c t i o n . M1, M2, and M3, a r e the f i r s t , second, and t h i r d m u l t i p l e r e f l e c t i o n s . Energy i s l o s t not o n l y through i n c r e a s e d g e o m e t r i c a l s p r e a d i n g and s c a t t e r i n g , but a l s o p e n e t r a t i o n through the r e f l e c t o r . i d e n t i f i e d because the time difference between the lower r e f l e c t i o n and f i r s t multiple (or between the n and n+1 order multiples), i s exactly the same as the time difference between 37 the two r e f l e c t o r s which produced the multiples. Reflections from valley walls occur infrequently with the UBC 840 MHz radar because of the small radar footprint. When valley wall r e f l e c t i o n s do occur they are usually e a s i l y i d e n t i f i e d because they change rapidly compared to r e f l e c t i o n s from the ice bottom. 3.3 ICE THICKNESS AND LAYER SEPARATION The ice depth to a r e f l e c t o r i s z = c i ( t r - t s ) / 2 (3.1) where z = depth in m c^ = ice radio wave ve l o c i t y in UI/MS t = time of surface r e f l e c t i o n in MS t = time of r e f l e c t i o n in MS "V* The accuracy of the thickness map produced w i l l depend upon the navigational accuracy (variable), the accuracy of the ice radio wave v e l o c i t y used (several percent e r r o r ) , and the accuracy of the i d e n t i f i e d r e f l e c t i o n locations (less than 1% error on A-scope traces). Timing accuracy (how accurately a r e f l e c t i o n i s located in the time section) depends upon the r i s e time of the sounding pulse. The UBC 840 MHz radar uses a pulse with an 18 ns r i s e time. This pulse is sampled every 10 ns; one sample 38 (approximately one-half the r i s e time) is a reasonable estimate of the timing accuracy. At C^=170 m/jus (two-way velocity is c^/2), this corresponds to a 0.85 m range accuracy in locating the r e f l e c t o r . The radio wave vel o c i t y in ice i s not known thi s accurately, and the true depth to the r e f l e c t o r (absolute range accuracy) also depends upon the f i r n - l a y e r correction used (this correction compensates for the higher propagation velocity through the overlying snow and f i r n ) ; thus timing accuracy is greater than the absolute range accuracy. The absolute range accuracy i s about ±10 m. The r e l a t i v e range accuracy is t h e o r e t i c a l l y 0.85 m; this estimate is optimistic because noise decreases the accuracy with which a r e f l e c t i o n can be located. The a b i l i t y to resolve c l o s e l y spaced r e f l e c t o r s depends upon the sounding pulse length. Two r e f l e c t o r s separated by less than one-half pulse length w i l l return overlapped r e f l e c t i o n s . The UBC 840 MHz radar uses a 90 ns pulse length (this includes the r i s e and f a l l times; Narod and Clarke 1983); t h e o r e t i c a l l y the radar should resolve layers separated by only 7m. I have examined radar sections where the ice thins to 0 m; v i s i b l e layer separation i s lost at about 10 - 15m because backscattered energy broadens the returned pulse. Closely spaced layers can sometimes be resolved i f the overlapped r e f l e c t i o n appears unusually broad. This works best where the ice is thinning or thickening and the individual r e f l e c t i o n s can be seen to merge. 39 3.4 REFLECTION AND PROPAGATION LOSSES The returned power measured at the receiver output i s the transmitted power less a l l losses: system losses from the radar and sampler, loss through radio wave geometrical spreading, loss by r e f l e c t i o n , loss from scattering by ice inclusions and roughness, and loss from d i e l e c t r i c attenuation during propagation through the ice. The system loss is constant and geometrical spreading loss can always be determined; in some instances the r e f l e c t i o n and propagation losses can be measured. In a l l cases I make the reasonable assumption that ice has the magnetic permeability of free space ( i . e . , i s non-magnetic). A r e f l e c t i n g surface causes losses in two ways: f i r s t , some of the energy i s transmitted through the surface instead of r e f l e c t e d ; and second, r e f l e c t o r roughness can scatter the incident energy thus decreasing the returned power. The power r e f l e c t i o n c o e f f i c i e n t (PRC) i s the r a t i o of the r e f l e c t e d (and returned) power to the incident power (in dB). It includes components from the per m i t t i v i t y contrast at the r e f l e c t o r , and from roughness. The propagation loss also has two components: a d i e l e c t r i c loss through ohmic heating and absorption; and a loss from scattering by ice inclusions. The ohmic heating i s caused by ice conductivity and can be calculated from the ice loss tangent. 40 3.4.1 POWER REFLECTION COEFFICIENT An electromagnetic wave impinging on an interface separating media of different d i e l e c t r i c properties i s partly r e f l e c t e d and partly transmitted. If a smooth interface separates media with r e l a t i v e p e r m i t t i v i t i e s e, and e 2 r and the electromagnetic wave i s normally incident, then (Stratton 1941) R = (/77-/i~;)/(/77+/77) (3.2) where R = r e f l e c t i o n c o e f f i c i e n t e,= pe r m i t t i v i t y of the media containing the incident wave e2= pe r m i t t i v i t y of the media containing the transmitted wave The PRC i s defined as PRC = 10•log,o{|R|2} (3.3) In general the p e r m i t t i v i t i e s can be complex (conductivity causes the imaginary component), but for the PRC's that I calculate in this thesis, I assume that the PRC is p r i n c i p a l l y caused by the contrast between the real components of the p e r m i t t i v i t i e s . 41 Equation (3.2) is idealized because the pe r m i t t i v i t y change at an interface is not always sharp. For example, i f a quarter-wave thick layer of p e r m i t t i v i t y e2=v/e 1 e 3 i s sandwiched between media of p e r m i t t i v i t i e s e, and e 3, no energy i s r e f l e c t e d (PRC = -» dB; Stratton 1941, p. 514). Using the PRC to separate the constituent components of the d i e l e c t r i c contrast (from v e l o c i t y and conductivity contrasts) is d i f f i c u l t . The fading pattern can be used to determine the r e f l e c t o r roughness (Harrison 1972; Neal 1977; M i l l a r 1981). The system performance (ratio of the transmit-pulse power to the smallest detectable signal) of the UBC 840 MHz radar i s 126 dB and the dynamic range 80 dB (Narod and Clarke 1983). Hence, I scale the data which contain no signal to -126 dB, and the largest peaks (the clipped transmit pulse) to -46 dB. After scaling the data I remove the effect of radio wave geometrical spreading by using the "radar equation" (Skolnik 1962) : <P> = (G 2X 2R 2 )/( 647r 2r 2) (3.4) where <P> = r a t i o of the received to transmitted power G = one-way antenna gain X = wavelength in a i r R = r e f l e c t i o n c o e f f i c i e n t r = range 42 If the radio wave travels through ice, the range must be corrected for the increased gain from refraction at the ice surface: r' = r +r, /n (3.5) where r' i s the corrected range, r a the range in a i r , r^ the range in ice, and n the index of refraction of the ice. I confirmed the 126 dB system performance by examining radar data from Kluane Lake, N.W.T., Canada. The unfrozen lake water has a r e l a t i v e p e r m i t t i v i t y of approximately 80; hence by equations (3.2) and (3.3) the lake surface PRC is -2 dB. No propagation loss occurs during t r a n s i t because the path i s t o t a l l y through a i r . If I remove the geometrical spreading and scale the data as explained above, I obtain a system performance of 126±2 dB. The data were f i r s t 9-fold stacked to allow peaks to be more r e l i a b l y located; I roughly compensate for the s l i g h t l y increased SNR resulting from stacking by using a 0 dB surface PRC for Kluane Lake in the performance c a l c u l a t i o n . The PRC for a given r e f l e c t i n g horizon can be presented either by p l o t t i n g the value above a section or p r o f i l e , or as a power r e f l e c t i v i t y section. To produce a power r e f l e c t i v i t y section I calculate the PRC for every point in the time section (assuming some propagation loss rate), then plot the section using d i f f e r e n t grey or colour levels (see Figures 2.7 43 and 2.8) . 3.4.2 DIELECTRIC LOSS AND THE LOSS TANGENT For media with constant d i e l e c t r i c properties, I can calculate the power loss from conductivity. In general, the radio wave velocity in ice is constant below the f i r n layer, but the conductivity changes with depth because i t i s a function of temperature. In th i s thesis I assume a constant conductivity because the ice I examine is r e l a t i v e l y thin (<800 m). An electromagnetic plane wave propagating in the z-direction through a non-magnetic medium can be described by (Stratton 1941) E(z,t) = E 0exp[ i (cjt-kz) ] (3.6) where E(z,t) = e l e c t r i c f i e l d E 0 = amplitude CJ = angular frequency t = time k = wavenumber in z-direction z = distance The complex r e l a t i v e p e r m i t t i v i t y e' is related to the conductivity a and the real part of the pe r m i t t i v i t y e by 44 e' = e-ia/cj (3.7) The wavenumber i s k 2 = (cj/c) 2e' (3.8a) or by equation (3.7) k 2 = (w/c) 2(e-io/u) (3.8b) The loss tangent tan5 is defined as tan6 = a/coe (3.9) Thus the wavenumber i s k 2 = (w/c) 2e(1-itan5) (3.10) k = (w/c)/e(1-itan6) (3.11) For small tan6 the wavenumber can be approximated by k * (CJ/C)I/T( 1-itan6/2) 45 (3.12) Therefore, for the propagating wave E(z,t) = E 0exp[-z (OJ/C )/7tan5/2 ] • exp[ iwt-iz (co/c )/T] (3.13) The power at depth z i s equation (3.13) squared, then integrated over one period. The d i e l e c t r i c loss rate, LR (dB/m), i s related to the r e l a t i v e power which reaches depth z For the UBC radar, co = 2 T T - 8 4 0 - 1 0 6 HZ, C = 3 0 0 • 1 0 6 m/s, which gives Equation (3.13) i s only v a l i d for constant e and tan6. By solving the wave equation which produces equation (3.6) for a depth-varying p e r m i t t i v i t y function, one can calculate the variation of the loss tangent with depth. Using the methods of inversion theory, a calculated loss tangent function could (m) : LR'Z = 10«loglo{exp[z(cj/c)v/7tan5]} dB (3.14) LR = 76.4v/Ttan5 dB/m ( 3 . 1 5 ) possibly be used to estimate the temperature-depth p r o f i l e . 4. RADIO ECHO SOUNDING ON NORTHERN ELLESMERE ISLAND 4.1 INTRODUCTION During June 1981, more than 2000 line-kilometres of ice on the northern coast of Ellesmere Island were radio echo sounded using the UBC 840 MHz radar. The project was planned both to radio echo sound the area, and to test the UBC 840 MHz radar on diverse ice types. In t h i s chapter I discuss the results from the Ward Hunt and Milne Ice Shelves, Milne and D i s r a e l i Glaciers, and Mt. Oxford ice cap. These areas were previously radio echo sounded in 1966, by Hattersley-Smith and others (1969) using the SPRI Mark II echo sounder. They present depth p r o f i l e s for the Milne and D i s r a e l i Glaciers, and spot depths for the Ward Hunt Ice Shelf. My glacier depth results agree with t h e i r s , but the Ward Hunt Ice Shelf results are s l i g h t l y d i f f e r e n t . I have also calculated the basal power r e f l e c t i o n c o e f f i c i e n t (PRC) for these areas, and determined the propagation loss rate (LR) for the Milne and D i s r a e l i Glaciers, and the Mt. Oxford ice cap. The various ice types have di f f e r e n t measured basal PRC's: >-l0 dB for ice shelves; -5 to -15 dB for f l o a t i n g g l a c i e r s ; and -30 dB for grounded g l a c i e r s . Some areas of the Milne and Ward Hunt Ice Shelves have an abnormally low basal PRC (<-30 dB); the low PRC may indicate that brackish ice is freezing onto the shelf bottom, or the ice is saline or brine soaked in these areas (Neal 1977). 47 48 4.1.1 AREAS FLOWN AND NAVIGATION More than 1000 km of radio echo soundings were flown over the Ward Hunt and Milne Ice Shelves, and the D i s r a e l i and Milne Glaciers (see Figure 1.2). Four d i f f e r e n t methods were to provide navigation: Omega, dead reckoning, a e r i a l photography, and a microwave transponder system. The transponder system was to provide accurate navigation for the Ward Hunt Ice Shelf; unfortunately the system did not operate. Oblique a e r i a l photographs were taken from a side window using a 70 mm a e r i a l reconnaissance camera. Cloud cover prevented the taking of high a l t i t u d e v e r t i c a l photographs which would have greatly aided the interpretation of the oblique photographs. Navigation is therefore provided by the a i r c r a f t ' s Omega receiver and dead reckoning. The Omega system i s a worldwide network of eight very-low frequency (VLF) radio stations (Hobbs 1981, p. 258-271) which broadcast at 10.2 kHz. A computer onboard the a i r c r a f t , when i t receives broadcasts from three or more stations, calculates and continuously outputs a i r c r a f t position to a precision of one-tenth of a minute (about 200 m). Uncorrected absolute accuracy i s approximately 3 km; r e c a l i b r a t i n g the receiver using known geographic fixes improves accuracy. Dead reckoning over features which can be observed on the topographic maps and in the radar data allow me to recalibrate the Omega coordinates. The navigator read the Omega coordinates from the receiver display at approximately 20 s i n t e r v a l s . I determined 49 the radar trace corresponding to each navigation point which is recorded on the voice track, and l i n e a r l y interpolated coordinates for the traces between these known points. At least two navigation points bracket each f l i g h t l i n e ; points on the f l i g h t l i n e s are therefore interpolated accurately, points outside (for example during turns) are not. The voice commentary can be matched to the trace numbers within 2 s, introducing approximately 100 m navigational error. This is much smaller than the navigational uncertainty of the Omega system. Errors in Omega coordinates are caused by l o c a l perturbations in the broadcast pattern a r i s i n g from ionospheric and other disturbances. Over the 100 km x 100 km survey area, these perturbations introduce systematic errors which are e s s e n t i a l l y constant. Hence, i f one coordinate can be adjusted to i t s true location, the other coordinates can be displaced by the same amount. These adjustments were made using the dead reckoning data: for example, I ensured that the glacier f l i g h t s followed the valley centrelines and fixed the points where the a i r c r a f t flew over land and islands. Where features on topographic maps could be i d e n t i f i e d , the oblique a e r i a l photographs were used to aid in these adjustments. With these corrections navigational absolute accuracy i s t y p i c a l l y 1 km. Relative accuracy between traces along a given f l i g h t l i n e i s far better than t h i s . 50 4.1.2 ACCURACY OF RESULTS The absolute accuracy of a l l results in this chapter depend upon the accuracy of the radio wave vel o c i t y and the scaling (converting to true power values) used, and the accuracy with which I can locate the trace. I have previously suggested that the navigational accuracy i s t y p i c a l l y 1 km. In some calculations (for example PRC computation or depth measurements over thin i c e ) , the propagation velocity choice is r e l a t i v e l y unimportant. For 800 m thick g l a c i e r s , however, the choice of a propagation v e l o c i t y can change the calculated thickness by as much as 30 m. For example, from deep-glacier r e f l e c t i o n s 9 u.s below the ice surface, a propagation ve l o c i t y of 169 m/u.s (Robin and others 1969) gives a depth of 760 m, whereas a velocity of 176 m/jus (Clarke and Goodman 1975) gives a depth of 792 m. I use a propagation velocity of 176 m/jus for three reasons: t h i s i s the v e l o c i t y used by Clarke and Goodman (1975) for their UHF soundings; the r e l a t i v e p e r m i t t i v i t y found for the Ward Hunt Ice Shelf was 2.9, although this value is for s l i g h t l y less dense ice (Ragle and others 1964); and I have not included any f i r n - l a y e r corrections to compensate for the higher v e l o c i t y through the overlying snow and f i r n . The f i r n - l a y e r correction can increase the depth by several metres (Robin and others 1969) so I prefer to err by overestimating rather than underestimating the v e l o c i t y . Over thin ice, thickness measurements are accurate to within the f i r n - l a y e r and resolution errors (±3 m thickness uncertainty); over thick 51 ice, thickness measurements are subject to these errors plus the error in the radio wave vel o c i t y (±10 m thickness uncerta i n t y ) . The power levels are accurate to ±3 dB; this estimate i s from the ±2 dB error in the least-squares f i t to the Kluane Lake data, and because the algorithm I use to scale the data to their true power levels occasionally produces a surface PRC of -8 dB instead of the expected -11 dB (Robin and others 1969). Basal PRC's w i l l therefore be accurate to approximately ±3 dB plus the error introduced by an in c o r r e c t l y chosen propagation loss rate. I use a 0.03 dB/m propagation loss rate for the ice shelf data. The ice shelves are at the most 90 m thick so even a factor of two change in the propagation loss rate would only change the calculated PRC's by a maximum of 5 dB. However, i t i s unlikely that I have inc o r r e c t l y chosen the propagation loss rate by a factor of two because the indicated basal PRC does not vary with changing ice thickness; an in c o r r e c t l y chosen propagation loss rate would cause a PRC change with depth. 4.2 MT. OXFORD ICE CAP AND THE MILNE AND DISRAELI GLACIERS With the data from f l i g h t s over the Mt. Oxford ice cap and down the Milne and D i s r a e l i Glaciers (Figure 4.1), I was able to produce radar sections which c l e a r l y show the ice bottom. From these sections I calculated the ice thickness, basal PRC, and propagation loss rate. The D i s r a e l i and Milne Glacier sections unambiguously show the ice bottom; the 52 84°00' 72°00' F i g u r e 4.1 O v e r l a n d f l i g h t l i n e s . F l i g h t l i n e s over the Mt. O x f o r d i c e cap, and down the M i l n e and D i s r a e l i G l a c i e r s . section from the Mt. Oxford ice cap shows clear r e f l e c t i o n s but i t i s uncertain whether the r e f l e c t i o n s are from the ice bottom or are d i f f r a c t i o n hyperbolae. If I assume that the propagation loss i s caused solely by conductivity, I can calculate the loss tangent for the i c e . My results (tan6 * 0.0002 to 0.0003) are similar to the loss tangents measured by Ragle and others (1964) for the Ward Hunt Ice Shelf. It is interesting to note that the Milne Glacier tongue and a tributary glacier which flows into D i s r a e l i Fiord 53 have a -5 to -15 dB basal PRC whereas the rest of the Milne and D i s r a e l i Glaciers have a -30 dB basal PRC; this v a r i a t i o n suggests that the Milne Glacier tongue and the tributary glacier are f l o a t i n g . Hattersley-Smith and others (1969) note a 10 dB increase in the power ref l e c t e d from these areas. I calculated a larger power increase because Hattersley-Smith and others used a 35 MHz pulse; I used an 840 MHz pulse which is much more sensitive to r e f l e c t o r roughness because the wavelength i s more than twenty times smaller. 4.2.1 ICE THICKNESS Figures 4.2, 4.3, and 4.4 show the processed radar sections for the Mt. Oxford ice cap, D i s r a e l i Glacier, and Milne Glacier respectively. I f i r s t 9-fold stacked the data, then high-pass f i l t e r e d them with a 2.0 us window. From these sections I plotted depth p r o f i l e s (Figures 4.5, 4.6, and 4.7). The radar sections are plotted with the transmit pulse as the datum; the apparent "topography" of the ice surface i s a combination of the true topography and the a i r c r a f t a l t i t u d e . I cannot separate true surface topography from changes in a i r c r a f t a l t i t u d e so I assumed that the observed topography results from variations in a i r c r a f t a l t i t u d e , and plotted the p r o f i l e s with smooth surface topography. Only for the Mt. Oxford ice cap and the upper ends of the D i s r a e l i (<18 km along the f l i g h t line) and Milne (<24 km) Glaciers did I have d i f f i c u l t y in ide n t i f y i n g the bottom r e f l e c t i o n . The r e f l e c t i o n s from these areas often appear to 8 . 0 0 i 7 . 0 0 ... i,„,,,,„ o o co o c • " o o CD T J O T J I D Q X I — 1 1 -ID o —I o m o I - - H i—• " j o ID o :z o m ID oo-Q o Z CD r —-• — i • rr—I ^ z J=-o o 2 no -«jr o o CO • o o C O " o o CO C D " * o o THG-WRY TRAVEL TIME (uS) 6 . 0 0 5 . 0 0 H . 0 0 3 . 0 0 • 'JxM&c?* ^ ^ ^ ^ '-•:-.[,. " f o g ; ' 0 . 0 0 F i g u r e 4.2 R a d a r s e c t i o n f r o m t h e Mt. O x f o r d i c e c a p . S e e F i g u r e 4.1 f o r l i n e l o c a t i o n . The d a t a w e r e 9 - f o l d s t a c k e d a n d h i g h - p a s s f i l t e r e d . The c e n t r e s o f t h e b l a c k / w h i t e p i x e l s a r e t h e r e f l e c t i o n o n s e t s . The t r a n s m i t p u l s e a r r i v e s a t 0.3„s; t h e s u r f a c e r e f l e c t i o n a r r i v e s a t 1-4„s; a n d t h e b o t t o m r e f l e c t i o n a r r i v e s a t 4-7„s. The a p p a r e n t i c e s u r f a c e t o p o g r a p h y r e s u l t s f r o m a c o m b i n a t i o n o f a c t u a l s u r f a c e t o p o g r a p h y a n d c h a n g e s i n a i r c r a f t a l t i t u d e . ,.9.00 8 . 0 0 7 . 0 0 L___ THO-WflY TRAVEL TIME (wSl 6^ QQ 5 . 0 0 14.00 3 . 0 0 2 . 0 0 o o cn • ™ o o o o CO-• o o : D "TJ :^ro Q * **» o I-—I is : D — i • ° _,° co — i ID O . m o S I D Q CD» o " T | O r -O I T J= r ~ o i—i o m r - , 0 1 S j" ^— o CO O" • o o CO CO" o o • o o CO" o o :t.„» .j^y^,^ MW)>^  ^  ^ ^  ^  0 . 0 0 F i g u r e 4.3 Radar s e c t i o n from the D i s r a e l i G l a c i e r . See F i g u r e 4.1 f o r l i n e l o c a t i o n . Note the t r i b u t a r y g l a c i e r which f l o w s i n t o D i s r a e l i F i o r d between 64-76 km. The data were 9 - f o l d s t a c k e d and h i g h - p a s s f i l t e r e d . The c e n t r e s of the b l a c k / w h i t e p i x e l s are the r e f l e c t i o n o n s e t s . The t r a n s m i t p u l s e a r r i v e s at 0.3 ws; the s u r f a c e r e f l e c t i o n a r r i v e s at 1-3„s; and the bottom r e f l e c t i o n a r r i v e s at 2-8^s. The apparent i c e s u r f a c e topography r e s u l t s from a c o m b i n a t i o n of a c t u a l s u r f a c e topography and changes i n a i r c r a f t a l t i t u d e . cn cr. F i g u r e 4.4 Radar s e c t i o n from the M i l n e G l a c i e r . See F i g u r e 4.1 f o r l i n e l o c a t i o n . The M i l n e Ice S h e l f i s between 80-105 km. The d a t a were 9 - f o l d s t a c k e d and h i g h - p a s s f i l t e r e d . The c e n t r e s of the b l a c k / w h i t e p i x e l s are the r e f l e c t i o n o n s e t s . The t r a n s m i t p u l s e a r r i v e s at 0 . 3 / / S ; the s u r f a c e r e f l e c t i o n a r r i v e s at 1-2^s: and the bottom r e f l e c t i o n a r r i v e s at 2-9vs. The apparent i c e s u r f a c e topography r e s u l t s from a c o m b i n a t i o n of a c t u a l s u r f a c e topography and changes i n a i r c r a f t a l t i t u d e . cn CD 57 F i g u r e 4.5 Depth p r o f i l e f o r the Mt. O x f o r d i c e cap. See F i g u r e 4.1 f o r p r o f i l e l o c a t i o n . The peaks of the b u r i e d h i l l s a r e r e l i a b l y i d e n t i f i e d as bottom; the downgoing limbs may i n some i n s t a n c e s be d i f f r a c t i o n s . The upper and lower l i n e s a r e the i c e s u r f a c e and bottom r e s p e c t i v e l y ; the d o t t e d l i n e s denote a r e a s where the r e f l e c t i o n s may be d i f f r a c t i o n s . 58 a CD ' o U) ~o CO l _ CO G crone (T09! 0 - K l IOIX) (W) G'OB ' T S r 3AO0§ 1H9I3H OT*-F i g u r e 4.6 Depth p r o f i l e f o r the D i s r a e l i G l a c i e r . See F i g u r e 4.1 f o r p r o f i l e l o c a t i o n . Note the t r i b u t a r y g l a c i e r which f l o w s i n t o D i s r a e l i F i o r d between 64-76 km. The upper and lower l i n e s a r e the i c e s u r f a c e and bottom r e s p e c t i v e l y ; the d o t t e d l i n e s denote ar e a s where the r e f l e c t i o n s may be d i f f r a c t i o n s . 59 OTJSI rrozi OTJB trrra OTJE <ro OOE-( [OIX) (W) " T S 3AQ8.H 1H3I3H F i g u r e 4.7 Depth p r o f i l e f o r the M i l n e G l a c i e r . See F i g u r e 4.1 f o r p r o f i l e l o c a t i o n . The M i l n e Ice S h e l f i s between 80-105 km. The upper and lower l i n e s a r e the i c e s u r f a c e and bottom r e s p e c t i v e l y ; the d o t t e d l i n e s denote a r e a s where the r e f l e c t i o n s may be d i f f r a c t i o n s . 60 consist of downward opening hyperbolae; the tops of these hyperbolae are probably bottom, and the downward portions d i f f r a c t i o n patterns. Reflection-strength analysis of these r e f l e c t i o n s also suggests t h i s . Hattersley-Smith and others ( 1 9 6 9 ) note that their sections from Mt. Oxford ice cap indicate strongly undulating subglacial topography; my data also indicate these undulations which are apparent on Figure 4.7. 4.2.2 PROPAGATION LOSS RATE AND THE BASAL PRC The measured strength of a r e f l e c t i o n , once system and geometrical spreading losses have been removed, i s echo strength = -2«LR«z + PRC (4.1) i f z is the depth to the r e f l e c t o r , and the propagation loss rate LR i s constant. The factor of two in front of LR in equation (4.1) is needed because of the two-way propagation of the received wave. I have plotted ice-bottom r e f l e c t i o n strength versus depth for the ice cap and glacier data (Figures 4.8, 4.9, and 4.10). The system detection l i m i t (the minimum r e f l e c t i o n strength which can be resolved assuming some constant a i r c r a f t t errain clearance) is also shown on these figures. Reflections which l i e close to t h i s l i n e are barely above the background noise and are therefore less r e l i a b l e than r e f l e c t i o n s which 61 ECHO STRENGTH (DB) F i g u r e 4.8 Echo s t r e n g t h v e r s u s depth f o r the Mt. O x f o r d i c e cap. The system and g e o m e t r i c a l s p r e a d i n g l o s s e s were removed from the d a t a and 4% of the r e f l e c t i o n s p l o t t e d . The c u r v e d l i n e i s the system d e t e c t i o n l i m i t assuming a 150 m t e r r a i n c l e a r a n c e . R e f l e c t i o n s which l i e on the s t r a i g h t l i n e would have undergone a 0.032 dB/m p r o p a g a t i o n l o s s r a t e and a -35.0 dB b a s a l r e f l e c t i o n . The p o i n t s near the system d e t e c t i o n l i m i t may be d i f f r a c t i o n p a t t e r n s . The p o i n t s appear t o c u r v e t o the l e f t w i t h i n c r e a s i n g d e pth but may have been t r u n c a t e d by the d e t e c t i o n l i m i t . The d i s t a n c e s r e f e r to d i s t a n c e a l o n g f l i g h t l i n e . 62 F i g u r e 4.9 Echo s t r e n g t h v e r s u s depth f o r the D i s r a e l i G l a c i e r . The system and g e o m e t r i c a l s p r e a d i n g l o s s e s were removed from the d a t a and 2% of the r e f l e c t i o n s p l o t t e d . The c u r v e d l i n e i s the system d e t e c t i o n l i m i t assuming a 125 m t e r r a i n c l e a r a n c e . R e f l e c t i o n s which l i e on the s t r a i g h t l i n e would have undergone a 0.028 dB/m p r o p a g a t i o n l o s s r a t e and a -27.2 dB b a s a l r e f l e c t i o n . The p o i n t s from 63-78 km a r e from the small t r i b u t a r y g l a c i e r which f l o w s i n t o the D i s r a e l i F i o r d . The p o i n t s from 0-18 km may be d i f f r a c t i o n s . The d i s t a n c e s r e f e r t o d i s t a n c e a l o n g f l i g h t 1 1 ne. 63 ECHO STRENGTH (DB) 0 . 0 - J O . O - 2 0 . 0 - 3 0 . 0 - 4 0 . 0 - S 0 . 0 - 6 0 . 0 - 1 0 . 0 - 8 0 . 0 - 9 0 . 0 F i g u r e 4.10 Echo s t r e n g t h v e r s u s depth f o r the M i l n e G l a c i e r . The system and g e o m e t r i c a l s p r e a d i n g l o s s e s were removed from the d a t a and 2% of the r e f l e c t i o n s p l o t t e d . The c u r v e d l i n e Is the system d e t e c t i o n l i m i t assuming a 100 m t e r r a i n c l e a r a n c e . R e f l e c t i o n s which l i e on the s t r a i g h t l i n e would have undergone a 0.022 dB/m p r o p a g a t i o n l o s s r a t e and a -30.5 dB b a s a l r e f l e c t i o n . The p o i n t s from 78-108 km a r e from the M i l n e Ice S h e l f . The p o i n t s from 63-78 km a r e from the tongue of the M i l n e G l a c i e r . The p o i n t s from 0-24 km may be d i f f r a c t i o n s . The d i s t a n c e s r e f e r to d i s t a n c e a l o n g f l i g h t l i n e . 64 Table 4.1 Propagation Loss Rate and Basal PRC. Ward Hunt Ice Shelf g l a c i e r ice (-10°C) Ward Hunt Ice Shelf glacier ice (-40°C) Ward Hunt Ice Shelf sea ice (-10°C) Mt. Oxford ice cap (least squares) Mt. Oxford ice cap (visual f i t ) D i s r a e l i Glacier (least squares) Milne Glacier (least squares) Propagation Loss loss rate (dB/m) tangent 0.039 0.0003 0.013 0.0001 1.3 0.01 0.016 0.00012 0.032 0.00025 0.028 0.00022 0.022 0.00017 Basal PRC (dB) -42.8 -35.0 -27.2 -29.5 Propagation loss rates (at 840 MHz) for the Ward Hunt Ice Shelf have been calculated using data from Ragle and others (1964, Table II) and equation (3.15). The Mt. Oxford ice cap data were f i t v i s u a l l y because points near the system detection l i m i t are unreliable. 65 l i e well to the l e f t of the l i n e . I have calculated the least-squares f i t to equation (4.1) for the clear bottom r e f l e c t i o n s ; Table 4.1 summarizes these r e s u l t s . On the echo strength versus depth plots one can see that the r e f l e c t i o n s from the upper ends of the g l a c i e r s and the ice cap group separately (low echo strength) from the clear bottom r e f l e c t i o n s and may therefore be d i f f r a c t i o n s rather than bottom r e f l e c t i o n s . The figures also show areas of high basal PRC: the Milne Ice Shelf, the tongue of the Milne Glacier, and a small tributary g l a c i e r which flows into D i s r a e l i Fiord. The Milne Glacier tongue and the tributary glacier are probably f l o a t i n g ; the basal PRC can therefore be used to d i f f e r e n t i a t e between grounded (PRC -30 dB) and f l o a t i n g (PRC -5 to -15 dB) glacier ice. On Figure 4.10 there i s a group of r e f l e c t i o n s at about 70 m deep with a low PRC (-45 dB before correcting for the propagation loss) from the Milne Ice Shelf; the low PRC may indicate that the ice i s brine soaked or brackish ice i s freezing onto the shelf bottom (Neal 1977). To examine q u a l i t a t i v e l y the roughness of the glacier bed I analyzed the d i s t r i b u t i o n of the power reflected from the ice bottom. The variation in the returned power results from interference caused by scattering objects and r e f l e c t o r roughness. I plotted the normalized occurrence (number of r e f l e c t i o n s f a l l i n g within a 1 dB power range normalized by the maximum occurrence) against r e f l e c t i o n strength with a l l system, geometrical spreading, propagation, and r e f l e c t i o n 66 T -40.0 -20.0 0.0 20.0 CENTRED ECHO STRENGTH (DB) T -40.0 -20.0 0.0 20.0 CENTRED ECHO STRENGTH (DB) o o * T r z -40.0 -20.0 0.0 20.0 CENTRED ECHO STRENGTH (DB) 40.0 40.0 40.0 F i g u r e 4.11 D i s t r i b u t i o n of b a s a l r e f l e c t i o n r e t u r n e d power. A l l l o s s e s , i n c l u d i n g the PRC and p r o p a g a t i o n l o s s e s , a r e removed from the d a t a b e f o r e the d i s t r i b u t i o n of echoes i s p l o t t e d . The t h r e e histograms a re as f o l1 o w s : a) Mt. O x f o r d . i c e cap. Data a r e c e n t r e d around a l i n e w i t h a 0.032 dB/m p r o p a g a t i o n l o s s r a t e and a -35.0 dB PRC. R e f l e c t i o n s from 12-37 km a r e used. b) D i s r a e l i G l a c i e r . The small peak c e n t r e d at -5 dB i s p r o b a b l y caused by d i f f r a c t i o n s . Data a r e c e n t r e d around a l i n e w i t h a 0.028 dB/m p r o p a g a t i o n l o s s r a t e and a -27.2 dB PRC. R e f l e c t i o n s from 18-63 km a r e used. c ) M i l n e G l a c i e r . Data a r e c e n t r e d around a l i n e w i t h a 0.022 dB/m p r o p a g a t i o n l o s s r a t e and a -30.5 dB PRC. R e f l e c t i o n s from 24-63 km a r e used. 68 losses removed (Figure 4.11). The variation in returned r e f l e c t i o n strength for the clear r e f l e c t i o n s i s approximately ±5 dB; t h i s large variation implies that the r e f l e c t i n g surface i s rough r e l a t i v e to the sounding wavelength (Neal 1977; M i l l a r 1981). The UBC 840 MHz radar has a 21 cm wavelength in ice; only the smoothest surface would not appear rough at t h i s wavelength. 4.3 MILNE ICE SHELF Four f l i g h t s were flown over the Milne Ice Shelf (Figure 4.12). Simple processing produced clear sections (Figure 4.13 is an example): 9-fold stacking to reduce noise, then high-pass f i l t e r i n g with a 0.5 us window. I obtained esp e c i a l l y clear bottom r e f l e c t i o n s for the areas adjacent to the Cape Egerton ice r i s e and a small tributary glacier on the western shore half-way up the f i o r d . A band 4 km wide stretching across the f i o r d 4 km from the seaward shelf edge produced only a faint r e f l e c t i o n . I have lab e l l e d the r e f l e c t i o n as bottom although i t might be the a i r c r a f t multiple. I could obtain no r e f l e c t i o n s for the area adjacent to the terminus of the Milne Glacier, or for the northwest F i g u r e 4.13 ( o v e r l e a f ) Radar s e c t i o n from the M i l n e Ice S h e l f . See F i g u r e 4.12 f o r l o c a t i o n . Two l a r g e bottom c r e v a s s e s , i n d i c a t e d by the downward-opening h y p e r b o l a e , b r a c k e t a gap which has no c l e a r bottom r e f l e c t i o n . T h i s may imply t h a t the i c e between the c r e v a s s e s i s b r a c k i s h . The s e c t i o n was 9 - f o l d s t a c k e d and h i g h - p a s s f i l t e r e d . The c e n t r e s of the b l a c k / w h i t e p i x e l s a r e the r e f l e c t i o n o n s e t s . > (0 o 0 . 7 5 _ A 1 . 0 0 H 1.25— 1 . 5 0 H 1 . 7 5 2.00H 2.25—1 70 corner of the shelf. Large bottom crevasses bracket the l o w - r e f l e c t i v i t y band (see Figure 4.13); perhaps the band i s a huge refrozen lead. If the refrozen ice was brackish or brine soaked, the returned power would be decreased because the high-conductivity ice would attenuate the pulse, and the basal r e f l e c t i v i t y would be low. 4.3.1 ICE THICKNESS Figure 4.14 i s an ice thickness contour map for the Milne Ice Shelf. The navigation is accurate to approximately 1 km. The ice shelf averages about 70 m thick, reaching 90 m thick off Cape Egerton, and over 100 m thick mid-way up the f i o r d where a tributary glacier flows onto the ice shelf from the western shore. I have continued the thickness contours up onto the Cape Egerton ice r i s e , and the small tributary glaciers because the ice shelf is continuous with them. No f l i g h t s were flown over the g l a c i e r or the ice r i s e so I have no estimate of their thickness. The absence of r e f l e c t i o n s from the northwest corner of the ice shelf and the area adjacent to the Milne Glacier terminus indicates that the ice thickness i s less than the minimum that can be resolved by the system (10 m). 4.3.2 BASAL POWER REFLECTION COEFFICIENT Equation (4.1) i s used to calculate the basal PRC of the Milne Ice Shelf. I estimated the propagation loss rate to be 72 0.03 dB/m, based upon the results found in the previous section where the propagation loss rate for colder glacier ice is 0.02 to 0.03 dB/m, and upon loss tangents found for the Ward Hunt Ice Shelf (Ragle and others 1964). The choice of the propagation loss rate is not c r i t i c a l because the ice is so thin that even a loss tangent several hundred percent d i f f e r e n t would only change the results by several decibels. The basal PRC results are mapped in Figure 4.15. It i s interesting to note the correlation between the r e l a t i v e l y thin ice and low basal PRC. If meltwater which flows out of the f i o r d is freezing onto the shelf bottom as brackish ice, the basal PRC could be substantially reduced. Salt would spread the d i e l e c t r i c contrast from fresh ice to s a l t water over the thickness of the brackish layer reducing the PRC (see Section 3.4.1). As well, the brackish ice would cause a propagation loss rate greater than the 0.03 dB/m rate I used; thus the basal PRC would be underestimated. 4.4 WARD HUNT ICE SHELF Except for f l i g h t l i n e s (Figure 4.16) over the Cape Discovery ice r i s e and the eastern part of the Ward Hunt Ice Shelf, the soundings do not show clear bottom r e f l e c t i o n s . The absence of clear bottom r e f l e c t i o n s could be a result of high conductivity within the ice shelf and the correspondingly large d i e l e c t r i c attenuation (Ragle and others 1964), and by a low basal PRC. 74 F i g u r e 4 .16 F l i g h t l i n e map f o r the Ward Hunt Ice S h e l f The numbers refer to data tapes. The points labelled A and B bracket the radar section presented in Figure 4 . 1 7 . 75 The soundings from the western part of the Ward Hunt Ice Shelf show a faint r e f l e c t i o n at about 20 m deep; th i s r e f l e c t i o n i s probably caused by the lake ice/basement ice interface (for example see Marshall 1955; Lyons and Ragle 1962; Ragle and others 1964). Lyons and Ragle (1962) describe the lake ice as unconformably overlying the basement ice in a broad syncline. Leveling and seismic surveys, and boreholes show that the ice shelf i s t y p i c a l l y 43 m thick (Lyons and Ragle 1962; Lyons and others 1971); therefore the bottom r e f l e c t i o n should arrive approximately 0.5 jus l a t e r than the surface r e f l e c t i o n . I could not identify the bottom r e f l e c t i o n for most of the Ward Hunt Ice Shelf despite many reprocessing attempts. Figure 4.17 i s an example of the best results I could achieve. One can see the synclinal internal layer just below the surface r e f l e c t i o n , but no bottom r e f l e c t i o n i s v i s i b l e . The undulating r e f l e c t i o n is from Cape Discovery ice r i s e bottom. There often seems to be a faint r e f l e c t i o n at the expected location; unfortunately i t does not disappear while f l y i n g over ice-free land or sea ice and i s probably processing-enhanced noise. F i g u r e 4 .17 ( o v e r l e a f ) Radar s e c t i o n f rom the Ward Hunt Ice S h e l f . See F i g u r e 4 .16 f o r s e c t i o n l o c a t i o n . The s u r f a c e r e f l e c t i o n has been a l i g n e d to 1.0»,s. The e x p e c t e d bot tom r e f l e c t i o n t ime 1s 1 . 5 „ s ; no r e f l e c t i o n i s v i s i b l e a t t h i s t i m e . I n s t e a d , a f a i n t r e f l e c t i o n at 1.25,,s 1s v i s i b l e ; t h i s i s p r o b a b l y the l ake 1ce/sea 1ce i n t e r f a c e . The u n d u l a t i n g r e f l e c t i o n i s f rom the bot tom of the Cape D i s c o v e r y i c e r i s e . The f a i n t r e f l e c t i o n at 1.7^s i s the a i r c r a f t m u l t i p l e r e f l e c t i o n . The b l a c k / w h i t e p i x e l s a r e the l o c a t e d r e f l e c t i o n s . The d a t a were 9 - f o l d s t a c k e d and h i g h - p a s s f i l t e r e d . (sn) em;) | d A B J » A B M - O M J 77 4.4.1 ICE THICKNESS The areas where I could identi f y the lake ice/basement ice interface (15 - 25 m deep) and the ice bottom (45 - 50 m deep) are mapped in Figure 4.18. I have also included three spot-depths for the Cape Discovery ice r i s e . Hattersley-Smith and others (1969) obtained comparable results (about 100 m) for their f l i g h t over the Cape Discovery ice r i s e ; however, they obtained only 30 - 40 m ice thickness for the eastern part where I obtained 45 - 50 m. On Figure 4 . 1 8 , the eastern part of the Ward Hunt Ice Shelf (the area with a clear bottom r e f l e c t i o n ) i s discontinuous with mainland Ellesmere Island, and the shelf edge appears to extend too far into the Ar c t i c Ocean when compared to the map of J e f f r i e s (1982). This may indicate that the navigation is poor in t h i s area and the soundings should be displaced several kilometres south. 4.4.2 DIELECTRIC LOSS AND BASAL PRC The lake ice layer which overlies the basement ice is 15 - 25 m thick, and the basement-ice layer 20 m thick (Lyons and others 1971). The loss tangent of the sea ice in the basement ice layer at -10°C and 840 MHz is 0.008 (Ragle and others 1964); t h i s loss tangent implies a one-way propagation loss rate of 1.04 dB/m. A pulse re f l e c t e d from the shelf bottom could suffer the following two-way losses: 1. 2 dB propagation loss from t r a n s i t through 20 m of lake ice having a propagation loss rate of 0.05 dB/m. 78 F i g u r e 4.18 I c e - l a y e r t h i c k n e s s of the Ward Hunt Ice S h e l f . I n c l u d e d a r e t h r e e s p o t - d e p t h s over the Cape D i s c o v e r y i c e r i s e . The e a s t e r n p a r t where a c l e a r bottom echo at 45-50 m depth was o b s e r v e d may need t o be d i s p l a c e d towards Cape A l b e r t Edward. 79 2. 40 dB propagation loss from t r a n s i t through 20 m of basement ice having a propagation loss rate of 1.0 dB/m. 3. 40 dB r e f l e c t i o n loss r e f l e c t i n g from a brine soaked bottom; this value i s observed for the Milne Ice Shelf i f the areas of low basal PRC are indeed brine soaked. The t o t a l loss i s 82 dB, but the system detection l i m i t is approximately -78 dB. Thus the expected r e f l e c t i o n strength is below the detection l i m i t . The thickness change for the Cape Discovery ice r i s e is only 100 m; this variation i s i n s u f f i c i e n t to obtain a good estimate of the propagation loss rate. However, I was able to calculate the basal PRC as -15 dB by assuming a propagation loss rate of 0.03 dB/m. The eastern part of the Ward Hunt Ice Shelf has a basal PRC of approximately -5 to -10 dB, similar to the results from the much thicker Milne Ice Shelf. 4.4.3 THE PRC OF THE LAKE ICE/SEA ICE INTERFACE. The r e f l e c t i o n at the presumed lake ice/basement ice interface i s caused in part by the p e r m i t t i v i t y contrast between the lake ice and basement ice. If e,' and e 2' are the complex p e r m i t t i v i t i e s of the lake ice and basement ice respectively, then by equation (3.2) one can calculate the r e f l e c t i o n c o e f f i c i e n t R: (4.2) 80 I f both l a y e r s have the same p r o p a g a t i o n v e l o c i t y , and the a s s o c i a t e d l o s s tangents are tan5 , and t a n 5 2 , then by equat ions (3 .7) and (3.9) R = ( / l - i t a n S z V l - i t a n S , ) / ( / l - i t a n 6 2 + / l - i t a n 6 , ) (4 .3) For smal l tan6 one can approximate R by R [ ( 1 - i t a n 5 2 / 2 ) - ( 1 - i t a n 8 1 / 2 ) ] / [ ( 1 - i t a n 6 2 / 2 ) + ( 1 - i t a n 6 , / 2 ) ] (4 .4) By i g n o r i n g terms of the o r d e r t a n 5 2 R =* i ( t a n 5 , - t a n 5 2 ) / 4 (4 .5) By equat ion ( 3 . 4 ) , the PRC caused by t h i s l o s s tangent c o n t r a s t i s PRC = I 0 - l o g 1 0 ( ( t a n 5 1 - t a n 5 2 ) 2 / l 6 } (4 .6) Measur ing the PRC of the lake ice /basement i c e i n t e r f a c e i s d i f f i c u l t because the energy b a c k s c a t t e r e d from the s u r f a c e o f t e n obscures the i n t e r f a c e r e f l e c t i o n . From i d e n t i f i a b l e 81 r e f l e c t i o n s I find that the PRC increases from about -40 dB near Ward Hunt Island, to -20 dB mid-way between the island and the Cape Discovery ice r i s e , then decreases to -40 dB near the Cape Discovery ice r i s e . I do not yet know whether th i s PRC increase is real or results from energy backscattered by increasing surface roughness. To produce a -40 dB PRC the loss tangent contrast would need to be 0.04. This contrast is larger than any reported by Ragle and others (1964). Their largest contrast (at 840 MHz) is less than 0.01; a 0.01 contrast produces a -52 dB PRC. Marshall (1955) and Crary (1958) note that a heavy dust layer marks the lake ice/basement ice interface. The dust could increase the PRC because the e f f e c t i v e p e r m i t t i v i t y would be a function of the dust and ice p e r m i t t i v i t i e s , and the dust concentration (Smith and Evans 1972; Glen and Paren 1975). The PRC probably results from a combination of v e l o c i t y contrast, loss tangent contrast, and the dust concentration at the lake ice/basement ice interface. 5. CONCLUDING REMARKS 5.1 DATA PROCESSING AND RADIO ECHO SOUNDING My basic processing scheme consists of reducing the noise in a radar section, then p l o t t i n g the section. High-frequency noise (higher than the signals one i s looking for) i s reduced by low-pass f i l t e r i n g across the section (stacking); low-frequency noise caused by backscattered energy from r e f l e c t o r roughness and ice inclusions is reduced by high-pass f i l t e r i n g down the section. Nonlinear AGC can be used to enhance low-amplitude r e f l e c t i o n s . AGC must be used cautiously and should only be applied to c a r e f u l l y f i l t e r e d data. I processed the data on a large mainframe computer (an Amdahl 470 V/8) but minicomputers should be able to perform a l l the basic processing. I used a PDP-1l/34a minicomputer with an attached Versatec V-80 e l e c t r o s t a t i c p l o t t e r to produce the grey-scale sections, and an International Imaging Systems (I 2S Model 70) image processor to produce the colour r e f l e c t i v i t y section. The problems which other researchers might encounter involve d i g i t i z i n g the analogue tapes, then storing the tremendous quantity of d i g i t a l data. The data set which I analyzed for this thesis contains the equivalent of 300,000 512-point traces. Each point i s 14 b i t s : the entire data set is more than 300 megabytes. Nine-track, high-density computer tape is the only p r a c t i c a l storage medium. 82 83 The processing I report in this thesis reduces noise on a trace-to-trace basis. Noise reduction techniques which operate on the entire radar section ( s p a t i a l and temporal f i l t e r i n g ) could be developed. The plotted sections are time sections ( r e f l e c t i o n a r r i v a l time rather than r e f l e c t o r depth i s plotted); in future work I plan to convert these to depth sections. 5.2 RADIO ECHO SOUNDING ON NORTHERN ELLESMERE ISLAND I obtained good ice thickness results for the Milne Ice Shelf (70 - 100 m), the Milne and D i s r a e l i Glaciers (they thin from >700 m to 0 m), and for the eastern part of the Ward Hunt Ice Shelf (45 - 50 m). I could not obtain t o t a l ice thickness for the rest of the Ward Hunt Ice Shelf, but I did obtain the depth to an internal r e f l e c t o r (15 - 25 m). This internal r e f l e c t o r i s probably the lake ice/basement ice interface. Where I obtained good ice thickness results, I could calculate the basal PRC: almost 0 dB for ice shelves; -30 dB for grounded g l a c i e r s ; -20 to -40 dB for the Ward Hunt Ice Shelf internal interface; -15 dB for the Cape Discovery ice r i s e ; and -5 to -15 dB for fl o a t i n g g l a c i e r s . The measured propagation loss rate for the Milne and D i s r a e l i Glaciers i s approximately 0.03 dB/m (tanS 0.00025). The Milne Ice Shelf i s surprisingly thick. Hattersley-Smith and others (1969) attempted to sound this ice shelf but their equipment did not operate u n t i l they had overflown the shelf and were onto the Milne G l a c i e r . I could 84 find no reference to the Milne Ice Shelf thickness in the l i t e r a t u r e and consequently expected the thickness to be similar to that of the Ward Hunt Ice Shelf (about 40 m). The Milne Ice Shelf averages approximately 70 m thick and i s more than 90 m thick in places. The regions where the ice i s thickest are adjacent to the Cape Egerton ice r i s e , and to a tributary g l a c i e r mid-way up the f i o r d . The basal PRC is large (>~10 dB) under the thick ice. I also obtained a large basal PRC under thick ice for the eastern part of the Ward Hunt Ice Shelf, adjacent to the Cape Albert Edward ice r i s e . The low PRC associated with the thinner ice could be the result of brine soaking or basally-accreted brackish ice. Meltwater flowing from the fi o r d s would flow beneath the ice shelves, but over the sea water, following the thinnest ice contours. The flow can be e a s i l y v i s u a l i z e d by considering the bottom as an inverted topography; the water flows through "valleys" (thin ice) rather than over " h i l l s " (thick i c e ) . If the meltwater i s freezing onto the shelf bottom as brackish ice, the basal PRC could be greatly reduced (Neal 1977). It is interesting to note that Keys and others (1968) found that the Ward Hunt Ice Shelf dams meltwater in D i s r a e l i F i o r d to a 43-44 m depth; t h i s suggests that the ice shelf i s 44 m thick. I obtained clear bottom r e f l e c t i o n s from the shelf only for the eastern part where ice i s thicker than 45 m. This suggests that meltwater flows out from the f i o r d under the western part of the shelf, and that where the water flows, the basal PRC is reduced. REFERENCES Clarke, G.K.C. and Goodman, R.H. 1975. Radio echo soundings and ice-temperature measurements in a surge-type g l a c i e r . Journal of Glac iology, Vol. 14, No. 70, p. 71-78. Crary, A.P. 1958. Arct i c ice island and ice shelf studies, Part I. A r c t i c , Vol. 11, No. 1, p. 3-42. Crary, A.P. 1960. Arct i c ice islands and ice shelf studies, Part I I . A r c t i c , Vol. 13, No. 1, p. 32-50. Evans, S. and Smith, B.M.E. 1969. Radio equipment for depth sounding in polar ice sheets. Journal of S c i e n t i f i c  Instruments (Journal of Physics E), Series 2, Vol. 2, p. 131-136. Glen, J.W. and Paren, J.G. 1975. The e l e c t r i c a l properties of snow and ic e . Journal of Glaciology, Vol. 15, No. 73, p. 15-38. Harrison, C.H. 1970. Reconstruction of subglacial r e l i e f from radio echo sounding records. Geophysics, Vol. 35, No. 6, p. 1099-1115. Harrison, C.H. 1972. Radio propagation effects in g l a c i e r s . Unpublished Ph.D. thesis, Cambridge, U.K., 193 p. Hattersley-Smith, G. 1963. The Ward Hunt Ice Shelf: recent changes of the ice front. Journal of Glaciology, Vol. 4, No. 34, p. 415-424. Hattersley-Smith, G., and others. 1969. Glacier depths in northern Ellesmere Island: airborne radio echo sounding in 1966, by G. Hattersley-Smith, A. Fuzesy, and S. Evans. DREO Technical Note 69-6, Geophysics Hazen 36. 46 p. Hobbs, R.R. 1981. Marine Navigation 2: C e l e s t i a l and  Electronic , second edit i o n . United States Naval Institute, Maryland. 343 p. 86 87 J e f f r i e s , M. 1982. The Ward Hunt Ice Shelf, spring 1982. A r c t i c , Vol. 35, No. 4, p. 542-544. Kanasewich, E.R. 1975. Time Sequence Analysis in Geophysics, second e d i t i o n . The University of Alberta Press, Edmonton, Canada. 364 p. Keys, J., and others• 1968. On the oceanography of D i s r a e l i Fjord on northern Ellesmere Island, by J . Keys, O.M. Johannessen, and A. Long. Manuscript Report, No. 6, Marine Science Centre, McGill University, Montreal, Canada. 7 p. + figures. Lyons, J.B. and Ragle, R.H. 1962. Thermal history and growth of the Ward Hunt Ice Shelf. Union Geodesique et  Geophysique Internationale. Association Internationale  d'Hydrologie S c i e n t i f i q u e . Commission des Neiges et des  Glaces. Collogue d'Obergurgl 10-9-18-9 1962, p. 88-97. Lyons, J.B., and others. 1971. Basement ice, Ward Hunt Ice Shelf, Ellesmere Island, Canada, by J.B. Lyons, S.M. Savin, and A.J. Tamburi. Journal of Glaciology, Vol. 10, No. 58, p. 93-100. Lyons, J.B. and others. 1972. Growth and grounding of the Ellesmere Island ice r i s e s , by J.B. Lyons, R.H. Ragle and A.J. Tamburi. Journal of Glac iology, Vol. 11, No. 61, p. 43-52. Marshall, E.W. 1955. Structural and stratigraphic studies of the northern Ellesmere ice shelf. A r c t i c , Vol. 8, No. 2, p. 109-114. M i l l a r , D.H.M. 1981. Radio-echo layering in polar ice sheets. Unpublished Ph.D. thesis, Cambridge, U.K. 177 p. Narod, B.B. 1979. UHF radio echo sounding of Yukon glac i e r s . Unpublished Ph.D. thesis, University of B r i t i s h Columbia, Vancouver, Canada. 183 p. Narod, B.B and Clarke, G.K.C. 1983. UHF radar system for airborne surveys of ice thickness. Canadian Journal of  Earth Sciences, Vol. 20, No. 7, p. 1073-1086. 88 Neal, C S . 1976. Radio-echo power p r o f i l i n g . Journal of  Glaciology, Vol. 17, No. 77, p. 527-530. Neal, C S . 1977. Radio echo studies of the Ross Ice Shelf. Unpublished Ph.D. thesis, Cambridge, U.K. 87 p. Prager, B.T. 1982a. Radar data processing programs. Unpublished internal report, University of B r i t i s h Columbia, Vancouver, Canada. Prager, B.T. 1982b. PROCESS2: a data processing and p l o t t i n g  package• Unpublished internal report, University of B r i t i s h Columbia, Vancouver, Canada. Racal Technical Handbook. 1978. Racal Store 4D/7D Technical  Handbook, fourth e d i t i o n . Racal-Thermionic Ltd., Hythe, Southampton, England. 32 p. + c i r c u i t diagrams. Ragle, R.H., and others. 1964. Ice core studies of the Ward Hunt Ice Shelf, 1960, by R.H. Ragle, R.G. B l a i r , and L.E. Persson. Journal of Glaciology, Vol. 5, No. 37, p. 39-59. Robin, G. de Q. and others. 1969. Interpretation of radio echo sounding in polar ice sheets, by G. de Q. Robin, S. Evans, and J.T. Bailey. Philosophical Transactions of the Royal  Society of London, Series A. Vol. 265, No. 1166, p. 437-505. Skolnik, M.I. 1962. Introduction to Radar Systems. McGraw-Hill Kogakusha, Ltd., London. 648 p. Smith, B.M.E., and Evans, S. 1972. Radio-echo sounding; absorption and scattering by water inclusion and ice lenses. Journal of Glac iology, Vol. 11, No. 61, p. 133-146. Stratton, J.A. 1941. Electromagnetic Theory. McGraw-Hill Book Company, New York. 615 p. Watts, R.D. and Wright, D.L. 1981. System for measuring thickness of temperate and polar ice from the ground or from the a i r . Journal of Glac iology, Vol. 27, No. 97., p. 459-469. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0052939/manifest

Comment

Related Items