DESIGN of an IMAGE RADIATION MONITOR for ILS GLIDE SLOPE in the PRESENCE of SNOW a Dissertation Presented to the Faculty of The

DESIGN OF AN IMAGE RADIATION MONITOR FOR ILS GLIDE SLOPE IN THE PRESENCE OF SNOW A Dissertation Presented to The Faculty of the

Fritz J. and Dolores H. Russ College of Engineering and Technology Ohio University

In Partial Fulfillment of the Requirement for the Degree Doctor of Philosophy

by Frank Marcum

August, 1995 ACKNOWLEDGMENTS

The author wishes to thank Dr. Roger Radcliff for his support over the years and his help in organizing this material. Mr. Joe Shovlin is also recognized for his help in reviewing the technical papers associated with this material. TABLE OF CONTENTS Page

LIST OF FIGURES GLOSSARY vii I. INTRODUCTION 11. PREVIOUS RESEARCH 111. DESCRIPTION OF GLIDE SLOPE AND PARAMETERS IV. CALCULATION OF FIELD FROM ANTENNA OVER GROUND A. The Optical Approximation B. Validity of the Optical Approximation V. GLIDE SLOPE PERFORMANCE VS. REFLECTION COEFFICIENT A. Null Reference B. Sideband Reference C. Capture Effect VI. EFFECT OF SNOW OVER GROUND VII. ANALYSIS OF SNOW EFFECTS A. Effects on Null Reference Glide Slope B. Effects on Sideband Reference Glide Slope C. Effects on Capture Effect Glide Slope D. Probability of Snow Type E. Effects of Rough Snow Surfaces and Terrain VIII. MONITOR DESIGN CONCEPT A. Monitor Error Budgets and Calibration B. Monitor Siting Criteria IX. CONCLUSIONS BIBLIOGRAPHY A. Ohio University Documents B. FAA Literature APPENDIX A. Tolerance Limits ABSTRACT LIST OF FIGURES Page 1. Vertical Position of Aircraft Relative to Course Deviation Indicator. 18 2. Two-Dimensional Geometry of Problem. 21 3. Geometry of Optical Problem. 4. Minimum Distances for Surface Wave and Space Wave Equality. 26 5. Characteristic Glide Slope Radiation Pattern. 27 6. Antenna Configurations for Glide Slopes. (Ref: Wilcox Glide Slope Manual.) 29 7. Multiple Reflections from a Multi-layered Image Plane. 3 5

8. Example of Reflection Coefficient for Snow of Increasing Depth over Ground. 39 9. Null Reference Cat I Tolerance Limits. 42

10. 3:l Sideband Reference Cat I Tolerance Limits. 43 11. 2.5:l Sideband Reference Cat I Tolerance Limits. 44 12. 4:l Sideband Reference Cat I Tolerance Limits. 44 13. Capture Effect Cat I Tolerance Limits. 45 14. Capture Effect Cat I11 Tolerance Limits. 46 15. Critical Snow Parameters that Cause Out of Tolerance Performance on Null Reference. Dielectric Constant = 1.4. 48

16. Critical Snow Parameters that Cause Out of Tolerance Performance on Null Reference. Dielectric Constant = 2.0. 48 17. Critical Snow Parameters that Cause Out of Tolerance Performance on 3:l Sideband Reference. Dielectric Constant = 1.4. 49 18. Critical Snow Parameters that Cause Out of Tolerance Performance on 3:l Sideband Reference. Dielectric Constant = 2.0. 49 19. Critical Snow Parameters that Cause Out of Tolerance Performance on Capture Effect, Cat I. Dielectric Constant = 1.4. 50

20. Critical Snow Parameters that Cause Out of Tolerance Performance on Capture Effect, Cat 111. Dielectric Constant = 1.4. 50 21. Conductivity vs. Relative Dielectric Constant for Dry Snow. 52

22. Roughness Reduction Factor. 54 23. Monitor Block Diagram. 56 vii GLOSSARY Alignment Error - difference between actual mean path angle and commissioned glide path angle. ~ypicallycaused by transmitter misalignment, antenna height error, or failure to account for sloping ground plane. AM - Amplitude Modulation. The amplitude of a radio frequency carrier is varied, or modulated in such a way that a simple detector circuit can receive the information encoded. Below Path Clearance - guarantees that the CDI needle will always be at full scale deflection when the pilot is well below the glide path and above any obstacles. Bend - long-duration deviation of the course from the nominal path angle. Typically caused by scatterer very close to the glide slope, or slow change in ground plane with distance. CDI - Course Deviation Indicator. A cockpit device that displays ddm scaled by a microamp value. An ammeter is used to locate the aircraft position relative to the on- course location, where a zero ddm value is measured. For example, if the pilot is below the glide path, the needle is above its centered reading. CSB - Carrier Plus Sideband. That portion of the ILS signal that contains 90 Hz and 150 Hz AM sidebands that are modulated in phase and broadcast vith the RF carrier. ddm - Difference in Depth of Modulation. For ILS, two audio viii

tones (90 Hz and 150 Hz) are space modulated onto the radio frequency carrier. At different points in space, varying levels of each tone can be received. An ILS receiver measures the amplitude of each tone versus the carrier level; the depth of modulation. The difference in these modulation levels is converted to a bipolar voltage that drives a cockpit display for aircraft guidance. In the case of glide slope, a greater amount of 90 Hz (fly down) tone indicates that the pilot is too high; a greater amount of 150 Hz (fly up) tone indicates he is too low. FAA - Federal Aviation Administration. Glide Path Angle - the mean angular path along which a glide

path receiver measures equal amounts of 90 Hz and 150 Hz

tones or 0 ddm. ILS - Instrument an ding System, the current radio navigation landing aid. ILS consists of: 1. localizer, for horizontal alignment with a runway intercept point or runway centerline. It operates in the frequency range between 108 and 112 MHz;

2. glide slope, for vertical guidance for rate of descent to a point of decision or touchdown point on the runway. It operates in the frequency range between 329 and 336 MHz.

The two combined steer the aircraft to a decision point at which time the pilot should be able to see the runway and complete his approach. Roughness - very short-duration, seemingly random deviations ix of the course from the nominal path angle. ~ypically caused by scatterers distant to the glide slope. SBO - SideBand Only. That portion of the ILS signal that

contains only 90 Hz and 150 Hz AM sidebands. The 90 Hz tone is out of phase with that on the CSB. Space Modulation - a phased array of antennas can produce desired maxima and minima in space. The CSB and SBO signals on the glide slope are space modulated so that

their sum produces a predominance of 150 Hz tone below

the glide path and a predominance of 90 Hz tone above the path. Scalloping - medium-duration, cyclical deviations of the course from the nominal path angle. Typically caused by scatterer in the vicinity of the glide slope. Symmetry - a quality factor expressing the amount of equality of angular excursion above and below the path at the

width points. Equality is defined at 50%. Width - the angular excursion between specific ddm levels

equal to k0.08875 ddm for glide slope (equivalent to +75 uA) . 10 I. INTRODUCTION The problem of monitoring the performance of the Instru- ment Landing System (ILS) image-type glide slopes has been investigated for a number of years. Both experimental and theoretical studies have yielded information about system performance, but the problem was by no means completely solved. Major error sources contributing to glide slope performance were identified as electronic component drifts and/or failures, scattering from nearby reflective surfaces, and changes to the ground plane in the vicinity of the glide slope. Transmitter signal errors can affect the radiated antenna signals that form the glide path and course width. Radiated signal integrity is verified by integral monitoring [I]. An integral monitor samples the antenna currents, verifying that the transmitted signals are capable of generating the commissioned path angle and course width. Integral monitors are calibrated by flight measurements to determine what changes in transmitter signal cause the path and width to go out of tolerance. The integral monitor accurately senses transmitter and antenna changes that affect the far-field, but it cannot detect changes in the environment that affect the ground- reflected signal [2]. Reflective objects near the glide slope produce multipath errors, which cause roughness, course bends, and scalloping in the approach region. ILS critical areas [3] were estab- 11

lished to reduce multipath interference from objects such as structures, vehicles, and aircraft stopped on the ground. The ILS critical area is a region in front of each radio navigation antenna system where these objects are restricted. This procedure reduces certain types of errors to what might be expected from terrain irregularities. There are two sources of radiation necessary to form the glide slope signal. These are the signal radiated directly from the antenna and its ground-reflected image. Addition of standing water or snow cover to the path-forming region of the ground plane in front of a commissioned facility can change the image radiation characteristics. Changes to the ground plane are currently addressed by the ILS Maintenance Manual

[4]. This procedure calls for visual inspection of the critical area or use of a snow depth monitor [5]. The snow depth monitor is a sonar-like device that sends an alarm to the system when the ground plane rises due to snow. When the snow reaches a certain depth, a Notice to Airmen is published, advising pilots not to use the glide slope (forcing higher landing minima) until the snow can be removed from the path- forming region. The FAA does not measure the image radiation nor the electrical parameters of snow. One drawback to the snow depth monitor is that the measurement is highly localized. Only the snow depth near the monitor is measured. If snow drifts are present throughout the path-forming region, the effective depth in front of the glide 12 slope may be different than what is measured by the monitor. The now-defunct near-field monitor attemptedto determine the variance of path and width by signal measurements in the near-field of the antennas. This monitor received the glide slope signal at a point on the airport and extrapolated it to the far-field. The monitor was overly sensitive to standing water and snow and was discontinued because it caused unneces- sary outages during bad weather and low visibility conditions

[6]. The monitor went into alarm while the glide slope was still usable. There are no monitors in use in the United States that accurately monitor the effects of snow on the glide slope signal. This dissertation proposes to: 1. examine the effects of snow on the image-type glide slopes and to derive a concise description of the

conditions that cause the system to go outside FAA- designated tolerances; 2. design a simple monitoring scheme that will measure any change in image radiation from the glide slope. Knowledge

of the direct and ground-ref lected signals will allow one to quantify any change in system performance. To achieve these goals, a formulation using geometrical optics for the effects that the reflection coefficient has on image- type glide slope in the far-field will be developed. The optical assumptions will be validated by showing that any other components of the fields are negligible. The effects of 13 snow cover are then related to the formulation and analyzed. The theoretical snow types that cause anomalous performance will be compared against FAA criteria for snow. A novel monitor designed by the author for measuring the ground-reflected image radiation will then be discussed. The data relating reflection coefficient to glide slope performance can then be used as a chart for determining when the monitor should send an alarm. 14

11. PREVIOUS RESEARCH There are very few publications on the effects of snow in the open literature. Most of the work studied has come from research done by Ohio University for the FAA. The earliest research at Ohio University [7],[8],[9] involves the use of physical optics to solve for the currents induced on the ground plane. Signals were then computed by integration of these currents for the total electric field in the far-field and at the near-field monitor. The algorithm is computation- ally excessive for smooth surfaces because of the enormous amount of ground plane segmentation. While this technique is appropriate for computing roughness on the glide path due to irregular terrain, it is inefficient in computing critical glide slope parameters. It was later determined that the effects of snow are more easily computed using geometrical optics [10],[11]. Reasonable correlation was reported between measurements and predictions. The conclusion was that the near-field monitor does not accurately represent the far-field in the presence of water or snow. This was attributed to proximity effects [12],[13]. The ray paths for the direct and ground-reflected signals to the monitor are not parallel as they nearly are in the far-field. Also, the incident rays to the specular reflection point in front of the monitor have a higher incidence angle. Since the reflection coefficient is a function of angle, the amount of reflected energy is different 15 than in the far-field. One observation made in these reports is that wet snow tends to raise the path angle; dry snow might lower the path angle slightly. It should also be noted that a plane wave reflection coefficient was used to compute signals in the near-field. Better correlation with modeled results might have been obtained by inclusion of the surface wave. This would have more accurately accounted for the shape of the wave close to the monitor. Another method would be implementa- tion of a Fourier integral technique for the ground-reflected signal, such as the procedure outlined by Redlich [14]. The major drawback to the techniques researched is that they require precise knowledge of the electrical characteristics of the snow and soil. The snow may be dependent on temperature, time, and recent climatic changes. Since the nature of the snow is not measured, and is difficult to measure, the techniques developed can not lead to a determination of far-field conditions based on monitor indications. The remainder of the Ohio University literature matures into characterizing the effects of snow either by flight measurement or simple raised ground plane analysis. The consensus was that the effects of snow were negligible for depths less than 8-10"; the path angle raised except in very rare circumstances. As discussed previously, the near- field monitors were phased out in favor of integral and snow depth monitors. Some flight measurements indicated that the path angle 16 could lower on a sideband reference glide slope with a truncated ground plane in the presence of snow. A truncated ground plane is one that is level to at least 1000' in front of the array, then drops off. In response, Redlich [15] and Marcum [16] examined the effect of snow on the path angle of the null reference and sideband reference glide slopes with truncated ground plane. The Fourier integral technique outlined in [14] was used, modified by the multi- layer reflecting ground plane. Contrary to previous beliefs, it was found that the truncation of the ground plane was not a factor on the control of path angle for the size of the ground planes investigated. Multiple reflections between the air-snow-soil interfaces were the significant contributing factors.

Walton [17], Walton and Tolley [18], and Lopez [19] examined the effect of wet snow on the width of the null reference glide slope. Their primary interest was in course width with some indication of path angle shift. They did not explore realistic snow conditions or the other image- type glide slopes. Lopez went on to say that there is a need for a monitor that determines the effect snow cover has on the system. 17 111. DESCRIPTION OF GLIDE SLOPE AND PARAMETERS The Instrument Landing System (ILS) is an aircraft landing aid that has been in existence for nearly 50 years. ILS uses space modulated radio signals that are detected by an AM radio. Two tones are modulated onto the carrier. The receiver measures the modulation depths of each tone and sends the difference in depth of modulation (ddm) to the cockpit display; the course deviation indicator (CDI). The ILS is made up of the localizer, which provides lateral guidance, and the glide slope, which provides vertical guidance. Together, these systems define a corridor along which the pilot can fly to safely land an aircraft in low cloud ceilings and limited visibility conditions.

The glide slope is that part of the ILS that guides the pilot's rate of descent and determines the location along the runway of the landing point at the terminal (approach) end of his flight. The glide slope receiver drives a cockpit display that informs the pilot how much above or below the path the aircraft is. The pilot's objective is to fly the aircraft so that the CDI needle is always centered. If this is done and all systems are functioning properly, the pilot should travel very safely to a point near and above the landing strip at which time he can decide whether he can complete his landing.

This is illustrated in Figure 1. The image-type glide slope consists of a vertical array of phased antennas that utilizes the ground plane to form a 18 pattern null in a part of the amplitude modulated sideband signal.

Above Glide Paih On Glide Path Below Gl& Path

Figure 1. Vertical Position of Aircraft Relative to Course Deviation Indicator.

Under ideal conditions, the contours form hyperboloids whose axis of rotation coincides with the glide slope mast. The mast must be offset from runway centerline so that it is not an obstruction to aircraft. The shape of the ddm contour intersecting the vertical plane containing the runway is then an hyperbola. Typical assumptions made about the ground plane are that it is infinite in extent, homogeneous throughout or perfectly conducting, the surface is flat and level, and there are no protrusions through the surface. If any of these assumptions is not true, the glide path may have course bends, roughness, or scalloping at various points in space. If these parameters exceed specified limits, 19 the system is classified as unflyable and the pilot will not have the use of this landing aid. These parameters are discussed in the USFIM [20] and are defined for several categories of flight quality. The emphasis of this document will be on Category I (Cat I) tolerances with the understanding that the input parameters for higher categories may be extracted from Appendix A of this document. Since the capture effect glide slope is typically used at Cat I11 sites, Cat I11 data will be shown for this system only. Several of these signal parameters are affected by varying ground planes. Glide path angle and course width are set by antenna height and transmitter adjustments. Other signal quantities such as symmetry, below path clearance, roughness, bends, and scalloping are often dominated by ground plane irregularities or reflecting structures, and are essentially site dependent. Since planar reflecting surfaces have been assumed, these quantities will not be discussed. The analysis shall be limited to path angle and course width effects. The glide path is defined as the locus of points at which the measured CDI is 0 uA. At large distances, these points tend to form a straight line whose projection to the ground intersects at an angle. This glide path angle, 8,, is chosen by FAA Flight Procedures so that the aircraft flies safely above all obstacles it might otherwise encounter while on 20 approach, yet not so high that the aircraft will descend too steeply. According to the USFIM [20], the path angle is not allowed to drift more than +lo% or -7.5% from the desired path angle for Cat I tolerances. A path angle that is too low can cause a pilot to fly lower than he thinks he is, possibly close to an obstruction. A path angle that is too high causes the pilot to descend more rapidly, meaning a harder landing. The course width determines the sensitivity of the cockpit needle for times when the aircraft is above or below the intended path. Course width is defined as

where 8, is the angle above the path at which -75 uA is measured, and 8, is the angle below the path at which +75 uA is measured.

The standard course width for glide slope is 0.7" as per

USFIM [20] specifications. This is not allowed to vary by more than f 0.2O. A course width that is too narrow can cause a pilot to over-compensate when he drifts off course; it is rougher and harder to fly. A course width that is too broad allows the pilot to fly further off course than he thinks he is. 2 1 IV. CALCULATION OF FIELD FROM ANTENNA OVER GROUND The ground plane in front of the array plays an important part in forming the space modulated signals on a glide slope.

The total field Etr from an antenna over ground can be written as

the sum of E,, the source field component and E,, the image field component reflected from the image plane. In the above, k is the wave number in free space, and j indicates the argument of the exponential is a complex number. The other parameters are illustrated by the geometry in Figure 2.

Figure 2. Two-Dimensional Geometry of Problem. 22

The value for k in free space will be assumed to be 2.1208 m-' throughout this document.

A. The Optical Approximation If one is interested only in far-f ield effects, and it is assumed that the ground plane is flat, level, and homogeneous, the use of image theory and geometrical optics applies. The solution for the field, normalized by exp(-jkD)/D, can be simplified to

where the first term on the right is the normalized source field, and the second term is the normalized ground-reflected image field. R is the reflection coefficient (complex ratio of reflected to direct signal), h is the height of the antenna above the image plane, and the other quantities are defined as shown in Figure 3. Although the image plane is simply defined at the air- snow surface, it is desirable to translate the image plane to the coordinate origin. This allows one to compare the effects of varying depths of snow to the no-snow conditions. When the reflecting surface is raised above a fixed coordinate system by a height d, (3) becomes - jkdsine [ejk(h-d) sine + ~~-jk(h-d)sine Et - I - jkhsine + ~~~-jkhsine

&s pt. Souce

D. hage Plane

moge

Figure 3. Geometry of Optical Problem.

Since the antenna and observation point have not moved with respect to the coordinate origin, the first term on the right remains unchanged. The reflected field has been altered by

B. Validity of the O~ticalA~~roximation The charts presented in this document will be calculated using the Fresnel reflection coefficient, f,. Fresnel reflection coefficients determine exact scattered fields for plane 24 wave incidence or for perfectly conducting ground planes. When the ground plane is other than perfectly conducting and the source wave is spherical, more terms must be included in the solution for the total field. The field from the Fresnel reflection coefficient is known as the space wave and the extra terms are known as the surface wave.

Stratton [23] shows how Weyl represented a spherical wave using a Fourier-Bessel technique to obtain a sum of plane waves integrated over (A = k sin 6) . The reflection coeff i- cient, R, may then be calculated exactly as

where J, is the zeroth order Bessel function of the first kind. r is the horizontal distance between the source and receive antennas, z, is the height of the source antenna above the ground plane, z is the height of the receive antenna above the ground plane, and the distance between image and receive points is

The Fresnel reflection coefficient is part of the integrand because it is a function of A. One may write f, when the ground plane is an infinite half-space as

This equation has not been solved in closed form and is difficult to integrate numerically. Jordan and Balmain

[21] write the first order terms of the scattered horizontal field from an infinite half-space. One may then calculate the reflection coefficient as

where the first term on the right yields the space wave and the second term is the surface wave. G is

and

where 6 is the angle of incidence and u = k,/k,, the ratio of wave numbers for air and reflecting surface. It is clear that when the space wave is much greater than the surface wave, the reflection coefficient is equal to the Fresnel reflection coefficient. The above equations can be solved for the smallest distance, D,, where the space wave is greater than the surface wave. For an angle of 3O and k, real,

Figure 4 shows that the critical distances are very small.

Figure 4. Minimum Distances for Surface Wave and Space Wave Equality.

The optical method used in this document is therefore accurate for all practical glide slope applications. The only requirement that must be made is that the antennas must remain above the snow layer. If the snow should exceed the antenna height, the space wave would be quenched, and the surface wave would dominate. 27

V. GLIDE SLOPE PERFORMANCE VS. REFLECTION COEFFICIENT

Two signals combine in space to produce the AM signal received by the glide slope receiver, the CSB and SBO. The CSB

is that portion of the ILS signal that contains 90 Hz and 150

Hz AM sidebands that are amplitude modulated and broadcast with the RF carrier. The SBO is that portion of the ILS signal that contains only 90 Hz and 150 Hz AM sidebands. The 90 Hz SBO tone is modulated out of phase from that on the CSB tone. Radiation patterns characteristic of a glide slope are shown in Figure 5.

Figure 5. Characteristic Glide Slope Radiation Pattern.

These signals are combinations of total fields as calculated in (2) or (3). As the relative phase of these signals vary, the amplitudes of the 90 Hz or 150 Hz tones will vary in such a way that the total modulation will remain essentially a constant. The receiver detects the modulations, 28 then measures the signal strength of the tones and compares them against the signal strength of the carrier. The sum of these two quantities is the total modulation of the signal; the difference is the ddm. The ddm is displayed for the pilot on a needle indicator by converting it to a scaled microamp value known as CDI. When there is sufficient modulation, the CDI can be calculated as

SBO CDI = A-Real (-1 CSB

where A is the voltage scaling factor between SBO and CSB required to generate a certain course width about the path. The Real function must be included because only the vector sum

of the 90 Hz and 150 Hz sidebands with respect to the carrier contribute to the ddm measurement.

There are 3 types of image-type glide slopes in use in the United States. These are the null reference, sideband reference, and capture effect. Illustrations of these systems are given in Figure 6 [22]. In subsequent discussions, the reflection coefficient shall be written as a phasor

R = M& (13)

where M is the magnitude, and $I is the phase of the reflection coefficient.

el ,"*

- 3 SIDEB-ND REFERENCE URE ECFEC T NULL REFERELC ;--CONFGJQATION 1GbRATl0hl

Figure 6. Antenna Configurations or Glide Slopes. (Ref: Wilcox Manual [22].) 30 A. Null Reference The null reference glide slope consists of two antennas, each with its own signal. The SBO antenna is placed at a height above ground where it will produce a signal minimum on path. The CSB antenna is placed at a height above ground where it will generate a signal maximum on path. The expressions for

CSB and SBO are written as

CSB = f (h,,d), SBO = f (h2,0).

Combine (3), (12), and (14) with (13) to write the CDI as

(1+~~)COS (X2-X,) + ~MCOS(X2+X,-@) CDI = A (1+~~)+ ~MCOS (2X,-@)

where X, has been introduced for kh, sine. Under ideal ground plane conditions,

where is the wavelength of the carrier frequency, and 0, is the desired glide path angle. The magnitude of the reflection coefficient can be solved 31 in terms of its phase. The form of the equation is a root of the quadratic

where

CDI cos(2X1-r#l) - cos (3X1-r#l) n G, = CDI -- cos (X,) A when (16) is incorporated into (15). Since the image plane is a passive reflector, onlymagnitudes of reflection coefficient between 0 and 1 are realizable.

B. Sideband Reference The sideband reference glide slope also consists of two antennas. The SBO is generated from equal and oppositely phased signals on both antennas. The heights are chosen to produce a signal minimum on path. The CSB antenna is broadcast on the lower antenna. The expressions for CSB and SBO are written as

CSB = f (hl,6), SBO = f (h,,6) - f (hl,6).

Combine (3), (12), and (19) with (13) to write the CDI as (1+~~)[COS (X2-X1)- 11 + 2M [COS (X2+X,-@)- cos (2X1-@)] CDI = A ( 1+M~) + 2M cos ( 2X1-@) (20) where X, has been introduced for kh, sin8. One can rewrite (20) as

(1+~~)[COS( (p-l)Xl)-11 + 2M [COS( (p+l)Xl-@)-cos(2X,-@) 1 CDI = A (1+~~)+ ~MCOS (2X,-@) (21) where p is the height ratio between the upper and lower antenna. Typical values for p are between 2.5 and 4. Most of the analysis will be done assuming a height ratio of 3, since this system is the most widely used. Under ideal ground plane conditions,

where X is the wavelength of the carrier frequency, and 8, is the desired glide path angle. Again, the magnitude of the reflection coefficient can be solved in terms of its phase. The form of the equation is a root of the quadratic where

(T+1)cos(2X,-@)CDI -cos( (p+l)X,-@) n G, = (--el)CDI -cos( (p-l)X,) A when (22) is incorporated into (21). Again, since the image plane is a passive reflector, only magnitudes of reflection coefficient between 0 and 1 are realizable.

C. Capture Effect The capture effect glide slope is a two-frequency array that relies on a phenomenon known as capture effect to reduce roughness due to non-ideal ground plane conditions. The parameters of this system are often controlled by the terrain at a particular site. The capture effect glide slope will be examined with the understanding that the data only describe trends in the presence of snow. It will become evident later that although the capture effect glide slope is designed to cancel energy at low angles, it is not a cure-all for problems with snow cover. There are two groups of signals present; course and clearance. These signals are separated by several kHz, yet both fit within the passband of an ILS receiver. The receiver has a tendency to lock onto the stronger of the two signals 34 and extract its information over that of the weaker signal. Three antennas are required for this system. The course array is known as the M-array. Like the null reference, the

SBO signal is generated by forming a pattern null with its image using the middle antenna. The upper and lower antennas also form a null on path but both are out of phase with the middle to provide signal reduction at low angles. This pair is similar to the sideband reference except that while the antennas are in phase with each other, they have an extra 180 degree path-length difference to produce a null on path. The

CSB is generated as in the null reference by applying signal to the lower antenna, except a small amount of anti-phased CSB is also placed on the middle antenna for signal reduction below path. The area below path is dominated by the clearance signal.

It is typically displaced from the carrier frequency by 8 kHz and has only the 150 Hz tone modulated on it. Its signal is generated similar to the second half of the SBO. Wherever the clearance carrier signal is comparable to or greater than the course carrier signal, the CDI rolls off strongly into the 150 Hz. The clearance signal does not normally contribute to guidance when the pilot is in the course region and will be ignored for computational simplicity. The techniques developed for the previous glide slopes can be used, but the expressions are long. Data will be presented, but not the expressions. VI. EFFECT OF SNOW OVER GROUND The effect of snow over ground can be developed by considering the multiple reflection problem of a dielectric slab separating two media. The geometry of the problem is

shown in Figure 7.

1 Obr Pt. 1

Sol (medium #3)

Figure 7. Multiple Reflections from a Multi-layered Image Plane.

Stratton [23] solves for the bulk reflection coefficient for normal incidence by summing all rays reflected and transmitted from the surface of the slab (snow cover) in the direction of the observation point (the aircraft), then applying an identity for the infinite series. His method can be extrapolated for oblique incidence as 36 where k2 and d are the wave number and depth of the snow, and

0, represents the angle from the tangent of the surface that the wave refracts into the snow layer. r,,, and r, denote the reflection and transmission coefficients at the boundary separating the incident medium, m, from the reflecting medium, n. For smooth surfaces, r,, - -r, and r, = l+r, so that

The reflection coefficient at each boundary is calculated using the Fresnel reflection coefficient written as

sin~m-\l~2,,,,,-cos28, r,,,,,= sin^,+/-

where Z, is the ratio of characteristic impedances in medium m divided by that in medium n, and 8, is the angle that the wave propagates through medium m with respect to the tangent at the boundary surface. The expression for the reflection coefficient has the complication that it is defined at the air-snow interface and not at the coordinate origin. As was demonstrated previously, translating the image plane to the coordinate origin modifies where k, is the wave number in air, and 8, is the observation angle. k, and 8, are related to k, and 8, by maintaining continu- ity of field components at the interface

Two phenomena are at work as illustrated by the following special cases. Assume the snow is wet and a good conductor (reflective and highly attenuating). The exponential terms

involving k, tend to zero and (28) reduces to

Addition of wet snow advances the phase of the reflection

coefficient. Examination of equations (15) and (20) indicate that under these conditions, increasing d will raise the path angle and broaden the course width. This phenomenon will be referred to as the raised ground plane effect. 38 The second case occurs when reflections at the air-snow interface are negligible compared to those from the soil. These conditions occur when the snow is dry, or icy. rI2 is negligible compared to r2, and the expression reduces to

k, and 8, are greater than k, and O,, so the phase moves in the opposite direction for increasing d. Since propagation through the snow is slower than in air, the phase of the reflection coefficient is delayed, tending to lower the path angle and narrow the course width for small d. As d increases, the path angle increases to a maximum, then returns to normal before starting the cycle over. This shall be referred to as the path-length difference phenomenon. Also note that because the argument in the exponential of

(31) is typically greater than that in (30), path lowering occurs first. The path lowering is not as pronounced because its magnitude is typically less. The total reflection coefficient can be described as a phasor with origin at the center of a unit circle whose vector moves slowly counterclockwise while making fast small clockwise circles. Refer to Figure 8 for an example where r2, is assumed to be a better conductor than r,,. -1 Figure 8. Example of Reflection Coefficient for Snow of Increasing Depth over Ground.

Anomalous reflection coefficients should be expected when the phase of r, is inverted by the complex exponential in

(26). This occurs for snow depths of

where E, is the relative dielectric constant, and n is any non-negative integer. In the above, conductive currents are assumed negligible, the angle of incidence is assumed small, and the identity in (29) is applied. This conclusion concurs with that drawn by Redlich [15]. 40 VII. ANALYSIS OF SNOW EFFECTS Reflection coefficients that cause out of tolerance path angle and/or course width can be computed by application of

(17) and (18) for null reference, or (23) and (24) for sideband reference, or by deriving the expressions for capture effect. These results have been submitted for publication in the IEEE-AES Transactions [24]. Since the magnitude of the reflection coefficient is known in terms of its phase, one can concisely describe the locus of reflection coefficients that cause out of tolerance conditions by graphing the contours as a phasor on a polar plot.

Path angle criteria are determined by setting the CDI to 0 and computing X, at each tolerance angle. The X, take on the form

for path high (H) and path low (L), respectively under Cat I tolerances. Width criteria are computed by first selecting A for a system such that the 75 uA points are 0.7" apart under nominal conditions (such as R = -1). One must then find values for M and @ so that they simultaneously meet the conditions that xnH and X: have an angular difference equal to either the broad 4 1 or sharp course widths at their respective 75 uA points. For example, insert the value for A into the null reference expression and CDI = -75 uA and let

into (17). 6 is either the broad or sharp course width. Then, insert CDI = +75 uA and let

Solve (17) under the two sets of conditions simultaneously for

@ and allow 8 to vary around the path angle. Insert the value of C#J into (17) for M. The width contour is now generated.

A. Effects on Null Reference Glide Slope The null reference glide slope is examined first. A nominal 3.0" path angle is chosen along with Cat I tolerances and a value for A of 0.30466. The shaded area in Figure 9 shows the allowable reflection coefficients for which the null reference glide slope performs under Cat I tolerances. If one were to graph the system for other practical path angles, one would find that the shaded portions are virtually identical, provided the value of A is modified to provide a nominal 0.7" course width. This statement will prove accurate for all systems. Note that the out of tolerance sharp course widths are not graphed. In all systems, the path angle is already out of tolerance for all values of reflection coefficient at which the sharp course width occurs. The upper boundary yields out of tolerance path too low conditions; the lower boundary yields out of tolerance path too high conditions. The ellipse in the middle yields out of tolerance course width too broad conditions.

Figure 9. Null Reference Cat I Tolerance Limits.

B. Effects on Sideband Reference Glide Slope-

A 3:l sideband reference glide slope is examined next. A nominal 3.0" path angle is chosen along with Cat I tolerances and a value for A of 0.30467. The shaded area in Figure 10 43 shows the allowable reflection coefficients for which the sideband reference glide slope performs under Cat I tolerances.

-1

Figure 10. 3:l Sideband Reference Cat I Tolerance Limits.

Again, the upper boundary yields out of tolerance path too low conditions; the lower boundary yields out of tolerance path too high conditions. The third curve is the boundary for widths that are too broad. Again, the sharp course width is not graphed since the path angle is already out of tolerance.

The 2.5:l (A = 0.38085) and 4:l (A = 0.22288) sideband reference glide slopes are also graphed in Figure 11 and

Figure 12. Note that the shaded area is larger for lower height ratio, or when the lowest antenna is higher. The system can be said to be more robust to the effects of snow as the shaded area becomes larger. -i

Figure 11. 2.5:l Sideband Reference Cat I Tolerance Limits.

-i

Figure 12. 4:l Sideband Reference Cat I Tolerance Limits.

C. Effects on Capture Effect Glide Slope Last, the capture effect course array is examined. A nominal 3.0° path angle is chosen along with Cat I tolerances 45 and a value for A of 0.30467. The shaded area in Figure 13 shows the allowable reflection coefficients for which the sideband reference glide slope performs under Cat I tolerances. Again, the upper boundary yields out of tolerance path too low conditions; the lower boundary yields out of tolerance path too high conditions. The third curve is the boundary for widths that are too broad. Again, the sharp course width is not graphed since the path angle is already out of tolerance.

Figure 13. Capture Effect Cat I Tolerance Limits.

Figure 14 shows the same capture effect system under Cat

I11 tolerances. The allowable reflection coefficients are restricted significantly because of the tighter tolerances on path angle. -j

~igure14. Capture Effect Cat I11 Tolerance ~imits.

The intersections of these curves with the reflection coefficient expression from (28) allow the determination of critical snow depths for any type of snow provided conductivity and dielectric constant can be determined. Depending on the electrical character of the snow, there can be regions of snow depth that cause out of tolerance conditions. These are the shaded regions in Figure 15 and Figure 16 for null reference,

Figure 17 and Figure 18 for the 3:l sideband reference,

Figure 19 for Cat I capture effect, and Figure 20 for Cat I11 capture effect. There are certain trends that are common to all graphs. At relatively small depths of dry snow, one or more shaded regions appear where the path-length difference phenomenon occurs. These typically occur as specified in (32). The early portions of these areas are dominated by path low conditions, 47 the later portions by path high conditions. Selected portions in the middle, top and bottom, have width broad conditions. The large shaded area to the right indicates where the raised ground plane effect caused by wet snow dominates. Unlike the path-length difference regions, the critical snow depths for raised ground plane conditions are independent of dielectric constant, provided the conductivity is high enough. The type of system and the path high tolerances for a given Category are the prime factors in raised ground critical depths. Note the small unshaded areas at the bottom where the reflection coefficient contours sweep briefly through the allowable region.

FAA criteria for glide slope facilities state that 18-24'' of snow must be present for null reference and capture effect, and 6-8" for sideband reference before snow in front of the array must be plowed [4]. The data clearly indicates that snow with certain electrical parameters can possibly take each glide slope out of tolerance before FAA specifications for depth have been met. Note that because of the raised ground plane effect, any type of snow is sufficient to put the capture effect out of tolerance under Cat I11 conditions. All snow reflection coefficient calculations presented are made at 3 O. Since the reflection coefficient for snow over ground is angle dependent, the shaded regions in the graphs will expand or contract slightly for increasing or decreasing observation angles, respectively. 1 E-05 0 5 10 15 20 25 30 35 40 45 Snow Depth in Inches Figure 15. Critical Snow Parameters that Cause Out of Toler- ance Performance on Null Reference. Dielectric Constant = 1.4.

0 5 10 15 20 25 30 35 40 45 Snow Depth in Inches Figure 16. Critical Snow Parameters that Cause Out of Toler- ance Performance on Null Reference. Dielectric Constant = 2.0. 0 5 10 15 20 25 30 35 40 45 Snow Depth In Inches Figure 17. Critical Snow Parameters that Cause Out of Toler- ance Performance on 3:l Sideband Reference. Dielectric Constant = 1.4.

Snow Depth In Inches Figure 18. Critical Snow Parameters that Cause Out of Toler- ance Performance on 3:1 Sideband Reference. Dielectric Constant = 2.0. 1 E-05 0 5 10 15 20 25 30 35 40 45 Snow Depth in Inches Figure 19. Critical Snow Parameters that Cause Out of Toler- ance Performance on Capture Effect, Cat I. Dielectric Constant = 1.4.

Figure 20. Critical Snow Parameters that Cause Out of Toler- ance Performance on Capture Effect, Cat 111. Dielectric Constant = 1.4. 51 D. Probabilitv of Snow Tvwe

Tiuri [25] provides semi-empirical equations for the loss tangent of dry snow based on measured values. When the snow is wet, the dielectric constant and conductivity increase rapidly. The expression for dry snow is written as

where E,'E, and E,"E~ are the real and imaginary parts of the dielectric constant, E, is the dielectric constant of free space, f is the frequency in MHz, T is the temperature in

Celsius, and p, is the density of the snow. The conductivity, a, for dry snow can be solved from the loss tangent as

because the expression E,' in terms of p, is also included in the paper, provided p, < 0.5 (or E,' < 2). In order to write the conductivity in terms of the relative dielectric constant, rewrite p, in terms of E,'. For a temperature of -10°C and a frequency of 332 MHz, conductivity can be written in terms of relative dielectric constant. his is shown in Figure 21. It is seen that for dry snow, the expected conductivity passes through the path-length difference areas. The probability of the combination of snow depth and electrical parameters that cause the path-length difference phenomenon should be considered possible. This discussion does not apply to the raised ground effect found for high conductivity wet snow. The raised ground plane is the upper limit when the system will be out of tolerance.

Relative Dielectric Constant Figure 21. Conductivity vs. Relative ~ielectricConstant for Dry Snow.

E. Effects of Rouqh Snow Surfaces and Terrain Based on the monographs of Beckmann and Spizzichino

[26], the effect of statistically rough snow or terrain is to reduce the magnitudes of the reflection coefficients, r,, that contribute to the total reflection coefficient such that A rough surface scatters the wave randomly, causing it to be less coherent. This is observed as a reduction in the reflection coefficient. This reduction factor, p,, is written as

where a, is the statistical deviation of the nth surface, k,,, and 8, are the wave number and angle in the mth medium. This equation is graphed in Figure 22 for an incidence angle of 3 degrees where it is assumed that if &, = 1, then a, refers to roughness at the snow surface. Any other value means that the reduction factor is computed for roughness at the ground plane or the underside of the snow layer. Since the electrical length and angle increase with E,, use is made of (29) in order to compute 8 for k. If one assumes that the transmission coefficient for a rough surface is also modified in the same manner as (38), namely, 54 then the bulk reflection coefficient calculated from (25) is modified as

-2jk,dshB2 p 12~12+ p 12p 21p 23r23e R = -2jhd sin 0,

+ P21p23~12~23~

after simplification into the form of (26).

0 3 6 9 12 o in inches Figure 22. Roughness Reduction Factor.

The wave passing through the snow and reflecting from the ground plane is affected more by rough surfaces than the wave reflecting from the snow surface. The path-length difference phenomenon has a tendency to be quenched by rough surfaces. Anomalous effects of snow over ground may be reduced by rough terrain and snow drifts. 55 VIII. MONITOR DESIGN CONCEPT As has been shown, when certain conditions are met, it is possible that snow cover can cause each system to go out of tolerance before the criteria for snow removal have been met. The out of tolerance condition goes undetected by the current monitor system. Under wet snow conditions (the expected majority of cases), the FAA policy is too restrictive; the system is shut down unnecessarily. The author asserts that by accurately measuring changes in the ground-reflected image, in conjunction with the existing integral monitor (or the Mark 20 electronics) and continuation of ILS critical areas, the length and number of outages caused by snow cover can be reduced. In addition, any monitoring scheme employed should detect the more unusual dry snow phenomena that cause out of tolerance performance. The remainder of this document examines a novel system of antennas which can be used for monitoring the image radiation from an antenna over ground. This monitor has been submitted for publication in the IEEE-AES Transactions [27]. The monitor scheme employs an antenna system that transmits a continuous-wave signal to a calibrated receive antenna. The bulk reflection coefficient is then calculated from the total (direct plus ground-reflected) signal. The reflection coefficient is then compared against charts derived earlier in this document to determine if the glide slope is in tolerance. 56 Calibration and error budgets are also discussed. A monitor based on principles from image theory is proposed. In the presence of a flat, level, reflecting ground plane, a stand-alone antenna sends a signal to a receiving monitor antenna. The antennas are assumed to be high enough that ground waves are negligible, yet low enough so as to approximate an angle of incidence comparable to the designed path angle of the glide slope. See Figure 23 for a block diagram of the monitor.

V V M RCV Ant

Figure 23. Monitor Block Diagram.

The voltages at the terminals of the two antennas are sampled and measured with a phase sensitive device, such as a vector voltmeter. The relative voltage at the terminals of the monitor antenna can be computed from (3) as

where V, is the voltage at the monitor antenna, V, is the voltage at the source antenna. The first term on the right is the source field, and the second term is the ground-reflected image field, R is the reflection coefficient (complex ratio of reflected to direct signal), k is the wave number in free space, DSf is the distance from the source antenna to the monitor antenna, and DIf is the distance from the image of the source antenna to the monitor antenna. Dsr and DIr are written as

Since the locations of the source and monitor antennas are known and the total field can be represented by the voltage measured at the monitor antenna terminals with respect to that at the source, the reflection coefficient can be deduced from (42) as This formulation, in conjunction with a look-up chart derived earlier in this text, tells when the image radiation combines with the direct signal to produce a glide slope that is out of tolerance.

A. Monitor Error Budqets and Calibration In general, knowledge of the locations of the antennas is not exact. An inch displacement of antenna position can mean a phase error of 10 degrees at glide slope frequencies. In addition, cable length errors can also produce a bias in the measurement. In this section, an attempt is made to set bounds on errors in the measurement of reflection coefficient. The above errors appear as

where Ds and D, contain antenna positional errors, L1 and L2 are the cable lengths (which might not be known) between antenna and phase meter as in Figure 23, and y is the wave number in the cable. Assume for the moment that cable errors are negligible. 59

One can then solve for position errors by defining two- dimensional position vectors for each antenna as

which are best guesses at the horizontal and vertical posi- tions of the monitor, the source, and image antennas, respectively. Next, define the two-dimensional vectors for errors in position of the respective antennas as

the position errors for the monitor, the radiating source, and the image source. The actual position vectors are then

Now calculate actual and approximate lengths for use in

(45) as It can be assumed that the 6 are bounded by some value much less than either Dsf or D,'. These lengths can then be expanded in a Taylor series and written in terms of Ds or Dl.

where

The path-length difference between direct and ground-reflected paths is

where the horizontal errors tend to cancel when the antennas are far enough apart. This will be discussed further in the 61 section on siting criteria. For glide path angles, the vertical error terms in (49) will be small and can also be neglected. Since displacement errors are typically much smaller than the distance between antennas, assume that only phase terms are affected and rewrite (45) as

where

Now include the cable length errors and solve for R in the form of (50)

where a = exp(y(L2-L1)-jkp). If the antenna locations are known to some accuracy, the error introduced by using these quantities in place of their actual values can be removed with a single measurement under nominal conditions where the reflection coefficient is known. a is a quantity that can be determined in a single calibration measurement. This term is solved as where E, is the calibration value of EN at some known value of the reflection coefficient, q. If the reflection coefficient is unknown, the author suggests wetting the ground between the antennas. The reflection coefficient should tend to a value of -1, provided the terrain is not altered by the addition of standing water. It is important to note that all values except EN are fixed until the system is altered and in need of recalibra- tion; D, and D, are fixed by the antennas and do not change. To illustrate this point, write (52) as

where A and B are constants.

B. Monitor Sitinq Criteria It is important that the measurement of fields at the monitor correlate with the fields received in the far-field by the aircraft. Considerations are now given to monitor placement that will aid the monitor in accurately representing any changes in the image radiation. The snow covering the ground may not be uniform. In order 63 to ensure that the monitor is representative of glide slope far-field performance, it is essential that the Fresnel zone of the monitor coincide with that of the glide slope as much as possible. The source antenna for the monitor might be placed on the glide slope tower, so long as there is no interference from (or to) the glide slope antennas. The receiving antenna should be placed near the edge of the ILS critical area, and far enough from the runway so as not to be an obstruction to landing or taxiing aircraft. Reflections from nearby stationary or moving objects may cause biased or varying fields at the receive antenna which may affect the measurements. If the monitor is placed in the ILS critical area, interference from structures and aircraft should be minimized to a degree comparable to that of the glide slope. A directional antenna, such as a log-periodic dipole antenna, may be required to reduce stray signals. When discussing errors, it was stated that the antennas had to be sufficiently far apart so that the horizontal errors in the second term cancel. The author suggests that the horizontal spacing should place the monitor antenna in the Fraunhofer (far-field) region. The horizontal spacing, r, should be at least 64 so that this condition is true. For example, if the antenna height is chosen to be 2 wavelengths (about 6 feet), r should be at least 95 feet. Reflection coefficients are dependent on incidence angle. The angle between the ground-reflected image antenna and the receiving antenna should approximate the glide slope path angle. If the monitor angle is different, the phase variance with snow depth will change proportionately. The monitor height requirement can be written as

where 8, is the designated glide slope path angle, Z, is the height of the monitor antenna, and Z, is the height of the source antenna. Remember that the heights should be chosen so that ground waves are negligible. If the snow rises above the heights of the antennas, there will be significant ground waves. Surface waves tend to decay rapidly with distance for horizontal dipole antennas. The question then arises whether the close proximity of the monitor antennas to a poorly conducting snow layer will cause the system to measure a reflection coefficient different from what an observer in the far-field might measure. As indicated in Figure 4, the required spacing is small. Other considerations in the 65 placement of the monitor force the antennas to be greater than this distance. 66 IX. CONCLUSIONS Data has been presented showing the necessary conditions for a uniform layer of snow over a reflective ground plane to cause an image-type glide slope to go outside USFIM tolerances. The data indicate two phenomena causing out of tolerance conditions:

1. wet, conductive snow cover effectively raises the reflecting surface, which causes a reduction in the effective size of the antenna array. This causes an increase in path angle and broadening of the course width.

2. dry, poorly conducting snow delays the reflected signal, caused by transmission of the signal through a low-loss snow having a wave number higher than that of the ambient medium (in this case, air). This can produce a depth critical condition which can cause path lowering, raising, and/or width broadening. The raised ground plane effect is more noticeable for highly reflective, wet snow cover. Critical snow depths for Cat I image-type glide slopes are: at least 34" for null reference and capture effect, and at least 17" for sideband reference glide slope. Capture effect under Cat I11 tolerances requires only 13.5" of wet snow cover to go out of tolerance. In general, the capture effect glide slope is about as robust as the null reference glide slope. The path angle robustness of a system is related to the ratio of the lowest 67 height antenna to the designed path angle. The sideband reference, in particular the greater height ratio systems, are more sensitive to snow since the antennas are lower. The critical depth for the raised ground plane effect is propor- tional to the ratio of path-high angle to path angle for a particular system. Width conditions are a function of path angle, in addition to the conditions previously discussed. FAA procedures [4] state that the system cannot be used when snow cover of 18-24" for null reference or 6-8" for sideband reference exists. Under wet snow conditions, FAA procedures are overly conservative for Cat I systems because the system is shut down before it is out of tolerance. Current

FAA procedures appear to require revision under Cat I11 tolerances since the capture effect can go out of tolerance with less than 18" of snow cover. The dry snow phenomenon can cause each system to go out of tolerance in an unstable condition that typically migrates from path low, to width broad, to path high. As the snow depth increases, the system may return to normal until the raised ground effect becomes significant. The shape of the curves varies such that the loss tangent is nearly a constant for a particular system. The critical snow depth is reached when the electrical distance through the snow adds an extra 180 degrees to the path-length of the reflected signal. Although high dielectric snow requires less snow cover for outages, the range of snow depths that can cause anomalous performance is 68 reduced. At a minimum, current FAA snow removal guidelines appear to require revision since the system can go out of tolerance before the criteria are met.

The paper by Tiuri [25] indicates that nominal measured values of dry snow tend to have the values of conductivity required to meet the conditions outlined in this document for the path-length difference phenomenor, to occur. The possibili- ty of a combination of snow depth and electrical parameters that cause the phenomenon should be classified as likely. A rough surface scatters the incident wave randomly, causing the received wave to be less coherent. This is observed as a reduction in the magnitude of the reflection coefficient. Anomalous effects, in particular, the path-length difference phenomenon, may be reduced by rough terrain and snow drifts, based on equations from Beckmann and Spizzichino

[261

As stated previously, FAA procedures appear to require revision in guaranteeing nominal system performance in the presence of snow. Current systems do not monitor the image radiation that contributes to the formation of the path, nor do they measure the electrical parameters of snow to determine if out of tolerance conditions exist. Only the now-defunct near-field monitors could observe glide slope anomalous performance caused by snow, but they were prone to be too conservative in alarming the system. No existing monitor can 69 accurately determine glide slope anomalous performance due to snow. A novel glide slope monitor that measures the effects that ground plane changes have on image radiation has been presented. This monitor, in conjunction with the charts already presented, allows one to quantify the effect the ground plane has on glide slope far-field performance. The defunct near-field monitor measured CDI in the vicinity of the glide slope. To stay on airport property, the monitor antenna had to be placed so that the path-length difference between upper and lower glide slope antennas was 180 degrees shorter than in the far-field (inside the Fraunho- fer region). This gave accurate measurements for correlating width and path angle shifts due to transmitter faults. The system became inaccurate whenever the environment changed, such as in the presence of standing water or snow accumulation. Under these conditions, the 180 degree point migrated away from the monitor antenna. There was no way to recalibrate the monitor without yet another system to determine the changes in the ground plane. The near-field monitor would no longer give an accurate representation of the far- field performance of the glide slope. The proposed monitor has many advantages over the near- field monitor. It measures only the ratio between direct and ground-reflected radiation. The monitor antenna can be placed in the far-field and still stay on airport property. Since it 70 does not require the glide slope signal, it does not confuse transmitter faults and ground plane changes. The monitor can

work in tandem with the integral monitor (or the Mark 20 electronics), thereby determining what combination of transmitter and ground plane faults cause the glide slope to go out of tolerance. Cumulative errors caused by a combination of transmitter imbalances and the ground plane can then be obtained. Unlike previous systems, the new monitor does not need to know what the snow composition is, or if standing water is present. It only concerns itself with changes in the image radiation; this is the bottom line in measuring changes in system performance based on ground plane.

Recommendations for future work would include verifying the phenomena by experiment. The raised ground plane effect has been well validated over the years, but the path-length difference phenomenon will probably require scale-modeled laboratory testing to confirm. Difficulties associated with in-situ measurement of electrical parameters of snow and surface roughness may make the phenomenon difficult to observe. While these conditions may make the phenomenon difficult to measure, it also would make its probability of occurrence small. The odds that snow cover can cause the image-type glide

to go out of tolerance need a good statistical basis to be 71 computed. Use of the proposed monitor will help to determine the probability of outages in the future. BIBLIOGRAPHY J. J. Battistelli, !#The Design and Testing of a Glide Path Integral MonitorIt, No. 35, Avionics Engineering Center, Department of Electrical and Computer Engineer- ing, Ohio University, Athens, Ohio, May 5, 1972. Morehart, Jack B., R. H. McFarland, David C. Hildebrand, ItSnow Effects on Image Glide Path Systemst1,Report No. FAA-RD-72-85, EER 5-13, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, Final Report, July 1972. Itsiting Criteria for Instrument Landing Systemtt,DOT-FAA Order 6750.16B, June 17, 1985. U.S. Department of Transportation, Federal Aviation Administration, FAA Order 6750.49, "Maintenance of Instrument Landing System (ILS) Facilities, Chg 16, Chapter 5, Section 3, Sub-section 3, 5297 - Snow Removal Procedure, Oct. 1991 R. H. McFarland, I1Rationale for Locating the Field Sensors for Snow Depth Monitors at a Sideband Reference Glide Slope Siten, Precis 118, Avionics Engineering Center, Department of Electrical and Computer Engineer- ing, Ohio University, Athens, Ohio, August 6, 1990. R. H. McFarland, J. T. Gorman, D. A. Hill, D. K. Lutter- moser and D. A. Miller, "Earth Cover and Contour Effects on Image Glide PathsM, FAA-RD-68-60, 1, EER 5-7, Avionics Research Group, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, September, 1968. McFarland, R. H., D. A. Hill, D. K. Luttermoser, "Earth Cover and Contour Effects on Image Glide Paths, Phase IIIt,FAA-RD-65-30, EER 5-1, Avionics Research Group, Ohio University, Athens, Ohio, May 21, 1965. Luttermoser, Donald K., "The Effects of a Ground Covering of Snow Upon the Image-Type Null Reference Glide Slope Monitorw, Technical Report 5-3, Avionics Research Group, Department of Electrical Engineering, Ohio University, Athens, Ohio, June, 1966. (Also submitted as Master's Thesis January, 1966.) McFarland, R. H., J. T. Gorman, D. A. Hill, D. K. Luttermoser, D. L. Miller, "Earth Cover and Contour Effects on Image Glide Paths, Phase IIn, FAA-RD-66-39, EER 5-5, Avionics Research Group, Department of ~lectri- cal and Computer Engineering, Ohio University, Athens, Ohio, July 10, 1966. [lo] Miller, David A., ftComparisonof Near and Far-Field Snow Effects on Image Glide Pathsff,Master's Thesis, Avionics Research Group, Department of Electrical Engineering, Ohio University, Athens, Ohio, August 27, 1966. [ll] Gilchrist, Thomas A., "Investigation of Changes in the Near and Far-Field Glide Path Angle due to Layers of Snow on the Ground-Planeff,Technical Report 5-9, Avionics Engineering Center, Ohio University, Athens, Ohio, June, 1970. [12] Smith, G. E., "A Worst Case Approach to Glide Path Errors Caused by Snow Cover on the Ground PlaneN, Technical Memorandum 18, Avionics ~ngineeringCenter, Ohio Univer- sity, Athens, Ohio, March 8, 1971. [13] Smith, G. E., "The Effect of Varying Depths of Uniform Snow Cover on Glide Slope Angle of Null-Reference, Sideband-Reference, and Capture-Effect Image Glide Slopesn, Technical Memorandum S-37, Avionics Engineering Center, Ohio University, Department of Electrical and Computer Engineering, Athens, Ohio, March, 1977. [14] Redlich, Robert W., lfImageRadiation from a Finite Ground Plane in Two Dimensionsu, IEEE Transactions on Antennas and Propagation, Vol. AP-16, No. 3, May, 1968. [15] Redlich, Dr. Robert, "Effects of Snow on Glide Slope Signals: The Mathematical Formulationtt, Technical MemorandumOU/AEC 91-63TM00006/45-1,Avionics Engineering Center, Department of Electrical and Computer Engineer- ing, Ohio University, Athens, Ohio, February, 1992. [16] Marcum, Frank, "Effects of Snow on Glide Slope Signals: Mathematical Modelingw, Technical Memorandum OU/AEC 91- 64TM00006/45-2, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, March, 1992. [17] Walton, Eric K., '!Effect of Wet Snow on the Null-Refer- ence ILS Systemff,IEEE Transactions on Aerospace and Electronic Systems, Vol. 29, No. 3, July 1993. [18] Walton, Eric K. and Peter Tolley, lfThe Effects of Wet Snow on the Null-Reference ILS SystemN, Proceedings of the 23rd International Seminar of the International Society of Air Safety Investigators, Vol 25, No 4, Dallas, TX, November 1992. [19] Lopez, Alfred R., t8Commentson 'Effect of Wet Snow on the Null-Reference ILS Systemrw, IEEE Transactions on Aerospace and Electronic Systems, Vol. 30, No. 4, October, 1994. [20] United States Standard - Flight Inspection Manual, FAA Handbook OA P 8200.1, chg 46, January 1991. [21] E. C. Jordan and K. G. Balmain, Electromagnetic Waves and Radiating Systems, chapter 16, Prentice-Hall Inc., New Jersey, 1968. [22] Wilcox Instruction Manual - Glide Slope Antenna, Manual 704180-0300, Wilcox Electric Inc, Kansas City, MO, June 1976.

[23] J. A. Stratton, Electromagnetic Theory, p. 511-2, McGraw- Hill Book Company, New York, 1941. [24] Marcum, Frank, utEvaluationof Image-type Glide Slope Performance in the Presence of Snow Coverut.Technical paper submitted to IEEE ~ransactions on Aerospace and Electronic Systems.

[25] M. E. Tiuri, A. H. Sihvola, E. G. Nyfors and M. T. Hallikaiken, "The Complex Dielectric Constant of Snow at Microwave Frequenciestt,IEEE Journal of Oceanic Engineer- ing, Vol. OE-9, No. 5, Decercber 1984. [26] P. Beckmann and A. Spizzichino, "The Scattering of Electromagnetic Waves from Rough Surfacestt, Artech House, Inc., Norwood, MA, 1987. [27] Marcum, Frank, ItDesign of an Image Radiation Monitor for ILS Glide Slopew. Technical paper submitted to IEEE Transactions on Aerospace and Electronic Systems. RELATED MATERIAL A. Ohio University Documents 1. McFarland, R. H., @@Effectsof Snow on Image Glide Paths with Suggestions for Improved Monitoringff Technical Memorandum Number 13, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, July, 1970. 2. Battistelli, J. J. and R. H. McFarland, IfResults of Measurements of Far-Field Glide Path Angles at Sites Possessing Heavy Snow Covern@,Technical Memorandum Number 23, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, February 19, 1971.

3. McFarland, Richard H., @nInvestigationtoProvide Improved Glide Slope Operation During Periods of Ground-Plane Snow Cover - Summary and Conclusions@@,RD-74-69, I, EER 5-17, Ohio University, Athens, Ohio, April 1974. 4. Battistelli, J. J. and R. H. McFarland, @fInvestigationto Provide Improved Glide Slope Operation During Periods of Ground-Plane Snow Cover, Volume I1 - Details of 1972-1973 Investigationsn@, RD 74-69, 11, EER 5-18, Ohio University, Athens, Ohio, April 1974. 5. McFarland, R. H., nInvestigation of Effects of Ground- Plane Deep Snow Cover on Image Glide Slope 1974-1975", FAAReport RD-75-210, Report EER 24-1, Avionics Engineer- ing Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, August 1975. 6. "Glide Slope Site Classifications Based on Depth of Ground-Plane Snow Coversoo,Technical Memorandum S-13, Avionics Engineering Center, Ohio University, Athens, Ohio, April, 1976. 7. Mitchell, Lawrence H., and R. H. McFarland, @@ThePerfor- mance of the Null-Reference Glide-Slope System in the Presence of Deep Snow 1975-1976@@,FAA Report Number RD- 77-24, EER 29-1, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, January 1977.

8. Smith, G. E., @@AGlide Slope Monitor System for Snow Siteslf, Technical Memorandum S-45, Avionics Engineering Center, Ohio University, Athens, Ohio, June, 1977. 9. Mroz , Mark, "Glide Slope Facility Snow Data Compilationn@, Technical Memorandum Number M-1, Avionics Engineering Center, Department of Electrical and Computer Engineer- ing, Ohio University, Athens, Ohio, September, 1977. 10. Chamberlin, Kent, "Capture-Effect and Sideband-Reference Glide Slope Performance in the Presence of Deep Snow, 1977-1978tt,FAA-R-6750.1, AAF-420, EER 36-1, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, Final Report, July, 1978. 11. McFarland, R. H., ttAnomalousSnow Effects on the ILS Glide Slopew, Precis No. 2, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, February 23, 1979. 12. Longworth, Joe D., "Instrument Landing System Critical Area Studies: Phase I, Theoretical and Experimental Investigations of Boeing 747 Dual Frequency LocaliZer Scattering for CAT I11 Critical Area Determinationtt,EER 59-3, Avionics Engineering Center, Department of Electri- cal and Computer Engineering, Ohio University, Athens, Ohio, November 1982 (revised August 1983).

13. McFarland, R. H. , ItA Review of Image Glide Slope Perfor- mance with Ground Plane Snow Covern, Technical Memorandum OU/AEC 53-87TM-80789/1, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, November 1987. 14. McFarland, R. H., "Flight Data on SBR Glide Slope Responses at Asheville, NC, with Ground Plane Snow Coverm, Precis 73, Avionics Engineering Center, Depart- ment of Electrical and Computer Engineering, Ohio University, Athens, Ohio, January 12, 1988. 15. Edwards, Jamie S., "Evaluation of the Cheyenne, Wyoming SBR Glide Slope with Ground Plane Snow Covertt,Precis 113, Avionics Engineering Center, Department of Electri- cal and Computer Engineering, Ohio University, Athens, Ohio, March 12, 1990.

16. Edwards, Jamie S., "Flight Evaluation of the Null- Reference Glide Slope Serving Runway 31 at the Houghton, Michigan Airport in the Presence of Ground Plane Snow Covertt,Precis 140, Avionics Engineering Center, Depart- ment of Electrical and Computer Engineering, Ohio University, Athens, Ohio, January 6, 1992. 17. Edwards, Jamie S., ttFlightEvaluation of the Sideband- Reference Glide Slope Serving Runway 32 at the Bradford, Pennsylvania Airport in the Presence of Ground Plane Snow Covertt,Precis 144, Avionics Engineering Center, Depart- ment of Electrical and Computer Engineering, Ohio University, Athens, Ohio, January 6, 1992. 18. Edwards, Jamie S., "Effects of Snow ILS Glide Slope Signalstt,TechnicalMemorandumOU/AEC 92-65TM00006/45-FR, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, October, 1992.

19. Edwards, Jamie S., "Flight Evaluation of the Sideband- Reference Glide Slope with a Snow Covered Ground Plane, Runway 16, Binghamton, NYtt, Precis 157, Avionics Engi- neering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, February 22, 1993. 20. Edwards, Jamie S., "Flight Evaluation of the Sideband- Reference Glide Slope with a Snow Covered Ground Plane, Runway 04, Schenectady, NY", Precis 158, Avionics Engineering Center, Department of Electrical and Computer Engineering, Ohio University, Athens, Ohio, March 3, 1993.

B. FAA Literature 1. Tech Report DOT/FAA/PM-86-7.1, p.128. 2. Jackson, ~illiamE., ed., "The Federal Airways SystemN, Institute of Electrical and Electronics Engineers, Inc., Washington DC, 1970.

3. "Instrument Landing System Concepts: Student Texttt,FAA Academy Training Manual, Catalog No. 40233, US Department of Transportation Federal via ti on Administration, March 1986. 4. "Installation Instructions for the ILS Glide Slopett, Bureau of Facilities, 1st ed., June 1, 1959.

5. "Instrument Landing System Glide Slopett,CAA Aeronautical Center Training Series, Facilities Branch Manual 206, US Department of Commerce, Civil Aeronautical Administra- tion, Aeronautical Center, Oklahoma City, OK.

6. '*Terminal Instrument Proceduresw, FAA 8260.3A1 2nd ed., February, 1970. APPENDIX A. Tolerance Limits There are three categories of approaches. Each category guarantees levels of signal quality and limits of coverage. The limit of coverage is defined from the edge of the service volume to the pilot's decision height. Category I provides coverage to ILS Point B (the middle marker, typically about 3500 feet from runway threshold) and 200 feet above ground. Category I1 provides coverage to ILS Point C, typically about 1000 feet from runway threshold and 100 feet above ground. Category I11 provides coverage to runway threshold and 50 feet above ground.

Parameter Reference Tolerance/ Limit

Width 217.3306b 0.7O _+ 0.2' Angle 217.3306a(l) +10.0% to -7.5% of the commissioned angle CAT I11 Within 4% of commissioned angle Symmetry 217.3306~ CAT I 67%-33%. * CAT I1 58%-42%. * CAT I1 67%-33%. (Broad sector below path only) CAT I11 58%-42%. * Clearance Below the Path 217.3307 Adequate obstacle clearance at 180 uA or greater of fly-up signal in normal (150 uA or greater in any monitor limit condition).

* Broad sector either above or below path