UNCLASSIFIED           {SECRET NOFORN}

(10 HZ). These latter data showed, incidentally, the typical appearance of precursor noise and clutter-related noise.

Auroral Measurements

(U) As part of the planned experimental activity, a large quantity of data on the radio auroral effects was gathered during the life of the AN/FPS-95. This was taken from observations in the northerly beams (numbers 1 through 6). (12)   In addition, the data recorded during the special synoptic investigation of clutter-related noise from beams 1, 7, and 12 were likewise examined for their auroral contributions. (12)

(U) {(S)} Although the results showed that, particularly in the most northerly beams, auroral radar echoes could adversely affect the performance of the AN/FPS-95, there was no indication that these effects were related in any way to the clutter related noise, which by observation and definition, and in contrast to the auroral returns, always occurred at the same radar ranges as the earth surface backscatter.

One-Way OTH Path Test (35,35)

(U) {(S-NF)} Objectives of the One-Path Test were (1) to determine whether the spectrum of a known "clean" signal would be modified by effects associated with propagation via an ionospheric path and (2) to compare the signal received by various components of the AN/FPS-95 antenna (the antenna employed a ground screen thought to be a possible cause of clutter-related noise) with the signal received by specially constructed Yagi antenna. Aimed toward the Eastern Mediterranean, the antenna pattern resembled that of a single string of the AN/FPS-95 antenna. The Yagi had no ground screen.

(U) {(S-NF)} In the tests, which were performed for the 6-day period from March 6 to 11, 1973, signals were transmitted by the AN/FPS-95 in England and received in the Eastern Mediterranean, and signals were transmitted by an HF site in the Eastern Mediterranean and received in England. During the period of the tests, the HF site transmitted for a 2-hr period from 10:00 a.m. to 12:00 noon. The interval was broken up into eight 15-min periods, during each of which the transmitter at the HF site was turned off during the fifteenth minute to permit the receive antenna configuration at the AN/FPS-95 to be changed. The Yagi was used during the first and last 15-min periods; in between, combinations of the two strings of the AN/FPS-95 antenna that pointed toward the HF site in the Eastern Mediterranean were used so that comparisons could be made. During the following 2-hr period, from 12:00 noon to 2:00 p.m., the AN/FPS-95 transmitted and the signal was received in the Eastern Mediterranean by means of a gated receiver installed in a van located near the HF site. The dynamic range of the gated receiver and processor was 80 dB or more. All one way tests were performed at 23.145 MHZ.

(U) {(S-NF)} The shape of the spectrum of the signal received at the AN/FPS-95 was essentially the same as that transmitted from the HF site at the same time. The spectrum was clean down to a level above 70 dB below the peak, at which point skirts formed; the average level of the skirts then fell off with frequency on either side of the carrier. Lines seen at a number of discrete frequencies in both local and over-the-horizon spectra were related to the power frequency and to blower frequencies. No evidence of any contamination by the ionosphere is present in the spectrum received at the AN/FPS-95, and no differences of consequence were seen between the spectra received by the several receive antenna configurations used at the AN/FPS-95, indicating that
the ground screen of the system antenna was not contributing measurable noise. The spectrum of the AN/FPS-95 signal received at the HF site likewise was found free of noise down to the limit of dynamic range of the gated receiver, which was about 83 dB.

Sporadic-E Layer/F-layer Test (37)

(U) {(S)} The objective of the Sporadic-E Layer/F-layer Test was to explore whether one layer or another of the ionosphere was the unique cause of the observed clutter-related noise. The test was made in connection with the Scientific Assessment Committee program in the early part of 1973 relative to the Synoptic Data Task (Task 7).

(U) {(S)} The idea of the test was to compare the spectra of signal sequences that propagated via two-way sporadic-E refraction paths with the spectra of signal sequences that propagated simultaneously via two-way F-layer refraction paths. If the spectra of the signal sequence that propagated via

316 JDR                                UNCLASSIFIED                               {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

F-layer contained clutter-related noise and that via sporadic-E did not, then the F-layer is a possible unique cause of clutter-related noise could be inferred; if the signals that propagated via sporadic-E only had clutter-related noise and the F-layer-only signals did not, then the F-layer would be absolved and sporadic-E implicated, and so on. Because sporadic-E layers were not in evidence during the time interval within which the Scientific Assessment Committee's investigation was conducted (February and March 1973), use was made of data recorded in June 1972 in connection with Design Verification System Testing (DVST) Experiment 202, when sporadic-E was a frequent occurrence.

(U) {(S)} Analysis of the data showed clutter-related noise to be present in the spectra of signal sequences that propagated over two-way sporadic-E propagation paths and in the spectra of signals that propagated simultaneously over two-way F-layer paths. The characters of the noise and the clutter-to-noise ratios were roughly the same in the two cases.

Transmitter Power Reduction Test (25)

(U) {(S-NF)} Objective of the Transmitter Power Reduction Test which is relevant here was to determine whether the high power radiated by the transmitter was heating, and thus modifying, the ionosphere so as to cause the observed clutter related noise.

(U) {(S-NF)} The test was done by members of the on-site staff on June 3, 1972 with the radar in its normal operating configuration, transmitting in beam 7 on horizontal polarization at a frequency of 17.4 MHZ. All six transmitters were used. The transmitter power was reduced in steps of 3, 6, 12, and 18 dB, each step being maintained for one minute, and all measurements were taken within about 5 min. In the data processing, range bins 80 nmi in range extent were formed, and the returns in each was coherently integrated for 6.4 sec. Further processing then yielded average noise power in all Doppler bins from PRF/8 to PRF/2 and the average clutter power in the first, eight Doppler bins around the carrier frequency. These averages were computed for each range bin during each integration interval. 

(U) {(S-NF)} The result relevant here is the behavior of average clutter power and average noise power in a range bin set near the peak of the ground backscatter. Here the clutter power and noise power increased together as transmitter power was decreased, but clutter and noise were only 10 to 12 dB down for the transmitter power reduction of 18 dB. (The experimenter conjectured that poor calibration of the power reduction switch could have caused the discrepancy.) There was no sharp reduction in noise power at any point during transmitter power reduction. Both clutter power and noise power decreased smoothly and proportionately with transmitter power reduction.

REFLECTION EFFECTS

(U) {(S-NF)} Postulated causes of range-related noise which attribute the phenomenon to equipment, local environment, or propagation effects generally include the assumption of the earth-surface reflection as an element of the relevant two-way radar propagation paths. This reflection is regarded as that of a fixed reflector, however, which does not therefore alter the spectral composition of the reflected energy from that of the incident energy. The spectral broadening that accounts for the clutter-related noise is assumed to occur elsewhere. In contrast, this section discusses postulated causes of range-related noise in which the spectral broadening of radiation, which is reflected back to the radar receiver from distant locations, occurs at the actual point of reflection, This reflection point may be in the normal ground-clutter reflection area or at some totally different location. 

(U) {(S-NF)} As described previously and as seen in Fig. 16, the range-related noise was observed mainly in three well-defined regions of radar ranges, that is, a "short-range" region extending out to approximately 600 nmi, a "precursor" region in front of the ground-clutter return, and a region coincident with the ground-clutter return. This latter noise is named "clutter-related noise," and it is the one of highest importance in its effect on the observation of most aircraft, since it is at the ranges of the ground clutter that the lower atmosphere is illuminated and, consequently, where the aircraft echoes are to be found. The other regions are also of some interest, however, since their noise may obscure the OTH observations of high-altitude targets such as ballistic missiles, as well as those of target echoes generated via multi-hop ambiguous-range propagation modes. Yet

   {SECRET NOFORN}                             UNCLASSIFIED                                    JDR 317
     296-318 0 - 79 - 3

=============================================================

UNCLASSIFIED           {SECRET NOFORN}

Table 7. Reflection effects matrix. (Table unclassified.)

another reason for investigating the origin of the close-in and precursor noise was the possibility that there may have been a single common cause for all of the observed excess noise.

Postulated Reflective Causes of Range-Related Noise

(U) Possible explanations for the generation of range-related noise attributed to reflection mechanisms, together with the relevant observations that tend to support or disprove the theories, are listed in Table 7 and are discussed in the following paragraphs.

Meteors

(U) {(S)} It has long been known that meteors entering the lower ionosphere at heights of around 100 km produce ionization effects that reflect radio waves in the high frequency radio band. (38) Since these meteor-induced effects are transitory in nature and would therefore spread the spectrum of the incident radiation, they were obvious suspects as the cause of at least some of the range-related noise. Figures 15 and 16 show how the presence of the meteor reflections in the E-layer can give rise to the range-related noise identified as "short range" and "precursor."  The meteor effects would be distributed evenly through the E-layer, thus resulting in a precursor range- related noise power-versus-range profile very similar in shape to that of the ground clutter, but reduced somewhat in range. Both theory and experiment have previously shown that the close-in and precursor range related noise should and does exist. (39,40)  Furthermore, the calculated and observed spectra of these features are shifted toward the recede direction, a fact that is in agreement with observations made with the AN/FPS-95.  Figure 15, for clarity, exaggerates the vertical scale. In fact, the far tail of the precursor range-related noise would overlap the area G1-G2 occupied by the ground clutter and thus come under the definition of clutter-related noise. However, it would not give rise to a peak of noise within this region, as observed with the AN/FPS-95 and depicted in Fig. 16, and would not exhibit the symmetrical spectrum of the observed clutter-related noise.

(U) {(S)} There were, and probably still are investigators who believe that the precursor noise and the clutter-related noise are one and the same phenomenon, This view is not shared by the authors of this paper. Strong credence was lent to the meteor explanation for the short-range noise by the results of the short-range noise experiment. (34)  Evidence for the meteor explanation of the precursor range-related noise was less well established. However, one of the radar displays features an A-scope representation of the envelope of raw radar return signals, on which could often be observed the characteristic decaying transient signals typical of radar echoes from the ionized columns caused by meteorites. These transients occurred at the same ranges as the precursor noise, and the two phenomena were therefore assumed to be connected. The results of the land/sea experiment are not compatible with a meteor explanation of clutter-related noise, since it is

318 JDR                                    UNCLASSIFIED                            {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

highly unlikely that the meteor effects would exhibit abrupt differences in their reflection capabilities as a function of their geographical positions within the AN/FPS-95 coverage.

(U) {(S)} To summarize, it appears that the clutter-related noise is a different phenomenon from the close-in end precursor range-related noise, both of which appear to be caused by reflections of radar energy from meteor- induced ionization within the E layer.

Auroral Effects

(U) The term "auroral" is a very loose description of the postulated causes of clutter-related noise considered under this heading. Such causes include all those which may be attributed to radar reflections from ionospheric irregularities, whether magnetic-field-aligned or otherwise. It happens that most of such well-known effects occur in the high latitudes and are somewhat loosely correlated in position with visible aurora.

(U) {(S)} The radio aurora effects are known to produce radar reflections over a wide radio-frequency range, including the HF band. Furthermore, these reflections exhibit Doppler frequency shifts and spreading on the order of the observed clutter-related noise spectral widths. Over-the-horizon measurements in the Arctic have shown this "diffuse spectrum clutter" as a severe limitation to the detection of aircraft. (41)  Also, the ranges from the AN/FPS-95 to the zone of maximum auroral activity were such as to place the radar ranges of the auroral reflections within the AN/FPS-95 coverage.

(U) {(S)} Much information was gathered throughout the operational life of the AN/FPS-95 on the radar returns from radio aurora. (12)  In addition, more of these data were specifically gathered as part of the synoptic data collection during the investigation of clutter-related noise. These data clearly distinguished auroral effects from those of clutter-related noise in a number of particulars. First, the auroral returns, while occasionally coinciding in range with those of ground clutter, were generally to be found at ranges and with statistical frequencies that varied considerably, depending upon the time of observation, season, magnetic activity, operating frequency, and azimuth. Second, the spectra of the auroral backscatter were generally highly asymmetrical. And third, the amplitudes of the returns were found to depend strongly upon the radar frequency, being 10 to 30 dB higher at 8 MHZ than at 10 MHZ.

(U) {(S)} These observed characteristics of radio aurora reflections contrast strongly with characteristics of clutter-related noise, which include gradual variations in level as a function of beam azimuth and radar frequency, symmetrical spectra, and a close correlation in range with that of the ground clutter.

Aircraft Returns

(U) {(S)} Among the less plausible suggested causes of clutter-related noise was the possibility that the reflections from a large number of aircraft, entering the radar receiver through the antenna sidelobes, could be the source. It would be ironic indeed if the AN/FPS-95 failed to see aircraft because it was seeing too many aircraft! Quantitative calculations to examine this postulated phenomenon have not been performed, largely because of a lack of data concerning the numbers, velocities, and dispositions of aircraft about the radar. It does, however, seem extremely unlikely that within a given range cell, even within a large azimuth sector, there would have been sufficient aircraft to occupy all the Doppler cells (typically several hundred) and, thus, have given the appearance of broadband noise. Even if this had been the case, then the relatively small number of aircraft within the antenna main lobe should have been separately resolvable in Doppler frequency and would, on account of the large two-way gain differential relative to the sidelobes, have been easily discerned above the clutter-related noise background. One would also have expected to see marked diurnal changes in the noise due to the reduction in air activity at night.

Earth-Surface Effects

(U) {(S)} While there are virtually no objects on the earth or sea surface which have translatory velocities comparable with those of aircraft and which might therefore produce Doppler-shifted radar reflections to interfere with OTH aircraft detection, there are nevertheless many objects, particularly man-made, that move, vibrate, or rotate in such a manner as to modulate an incident radio wave, either in phase or amplitude, so as to generate sidebands in the reflected power. These sidebands could, if removed sufficiently in fre-

{SECRET NOFORN}                                UNCLASSIFIED                          JDR 319

=============================================================

UNCLASSIFIED           {SECRET NOFORN

quency from that of the incident power, resemble clutter-related noise.  Within a single AN/FPS-95 range-azimuth resolution cell (an area of approximately 10,000 nmi² for a 1-millisecond pulse at a range of 1,000 nmi) practically anywhere within eastern Europe, one would expect to find a large number of potential modulating reflectors, such as vibrating telephone wires, fences and power lines, rotating wheels, and moving vehicles which, either alone or through interaction with surrounding terrain or structures, would present time-modulated reflective properties. Qualitatively at least, it is plausible that such an ensemble of modulating entities could spread the spectrum of the incident radar energy to produce the clutter-related noise phenomenon. The assumption that these effects ere spread evenly throughout the AN/FPS-95 coverage would suffice to explain the observed correlation between the amplitude-versus-range behavior of the clutter-related noise and that of the ground clutter.

(U) {(S)} Another modulation effect could be generated by stationary reflecting objects, composed of sections between which the electrical impedance is varying. For example, the metal frameworks of buildings may be composed of sections that are poorly connected electrically. A mechanical vibration could cause such a connection to vary in impedance and thus produce a reflection with sidebands at the vibration frequency and its harmonics. Indeed, such an effect was used to calibrate the AN/FPS-95. The devices in question were located in Norway and Turkey and consisted of log-periodic antennas, pointing toward the radar, whose terminals were connected to modulated impedances. The reflected power from these switched reflectors, with its characteristic modulation frequency, could be detected and identified at the AN/FPS-95. It is possible to imagine many man-made artifacts that might conceivably contain such modulated impedances, including wire fences, vehicles, and even railroad tracks whose sections are connected by impedances that could vary rapidly during the passage of a train across the joints.

(U) {(S)} A third modulating agency, somewhat similar to that just described, involves the existence of nonlinear electrical impedances within the reflecting bodies. Such impedances, when exposed to local time-varying electrical fields, would also have the effect of modulating the reflections of incident radar energy, and thus have the potential for generating clutter-related noise.

(U) {(S)} Finally, it should be noted that the effects described here need not necessarily be limited to man-made reflecting bodies. Intuitively, however, the observed motions of natural objects such as vegetation would lead one to expect a spectral distribution of reflected energy that would peak up toward the low Doppler frequencies, in contrast to the broad, flat characteristics of the observed clutter-related noise.

(U) {(S)} There is no doubt that the effects described in this section do exist. Investigations have aimed at using such effects for the radar identification of man-made objects. (42)  Whether the effects are quantitatively consistent with being the source of the clutter-related noise observed with the AN/FPS-95 is, however, not known.

(U) {(S)} In order to test the hypothesis that clutter-related noise was generated by earth-surface effects, such as those just described, the Land/Sea Experiment was performed. This experiment was designed to identify the existence, if any, of persistent differences in the clutter-related noise levels from geographically separate areas of the AN/FPS-95 coverage. In particular, the experiment was designed to include a comparison of land and sea areas, since such a comparison should reveal large differences in clutter-related noise if the clutter-related noise was generated by man-made artifacts. The results of the experiment (see Appendix) were consistent with the theory that little, if any, of the clutter-related noise was generated by reflections from the sea areas, when compared with that from the land areas. In contrast, the clutter levels returned from the sea areas were roughly similar in power to those from the land areas. These two facts support the hypothesis that the clutter-related noise is generated by some reflection mechanisms operating within the land areas of the AN/FPS-95 coverage, and is not caused by spectral spreading of the radar energy occurring either before or after reflection at the land or sea surface.

Multihop Effects

(U) {(S)} It is known that, as expected, the effect of a rising or falling ionospheric layer is to shift the frequency of radiation reflected from the layer.  It

320 JDR                                      UNCLASSIFIED                             {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

Table 8. Propagation medium effects. (Table unclassified.)

was therefore suggested that the radar energy arriving back at the receiver, after many such reflections from multihop propagation modes, might be a cause of clutter-related noise.

(U) {(S)} One reason that this would seem to be unlikely is that such multihop returns would not generally coincide in range with that of the observed clutter-related noise, which is always approximately coincident with the ground clutter. Another reason concerns the fact that the observed single-hop Doppler shift due to rising or falling ionospheric layers is usually less than 1 Hz. Since the amplitudes of the returns from successively higher orders of hop would generally be attenuated, one would expect the corresponding spectrum to fall off sharply with frequency. It would also be rare to encounter the particular mix of rising and falling layers necessary to account for a symmetrical spectrum. From this reasoning, it appears unlikely that multihop effects could explain clutter-related noise: This conclusion is strengthened by the low PRF observations performed during the short-range noise experiment, (34)  wherein clutter-related noise was observed during radar operation at a PRF of 10 HZ. 

PROPAGATION MEDIUM EFFECTS

(U) The spectra of high-frequency radar signals could be corrupted by a number of mechanisms in passing from the radar antenna over the horizon to the earth's surface. It is the dual purpose of this section first to list the various phenomena that have been postulated as possible mechanisms for such spectral corruption and then to review the evidence for and against each case as the cause of observed clutter-related noise. Here we consider only transmission effects; reflection effects are dealt with in the preceding section.

(U) {(S)} The matrix of Table 8 lists at the left specific phenomena that have been put forward as possible causes in the transmission medium for clutter related noise. Across the top are the names of various experiments that were performed to confirm or deny one or more of the causes. Check marks (X's) signify which experiments relate to the various postulated causes. The method here will be to consider each phenomenon in turn and to review for each the relevant experimental evidence that was generated in the attempts to find and eliminate the cause of the noise.

F-Layer Vertical Motion and Waves

(U) {(S-NF)} The experiments which bear on the F-layer of the ionosphere as the unique cause of clutter-related noise are the One-Way Path Tests, (35,36) the Sporadic E-Layer/F-Layer Experiment, (37) and

{SECRET NOFORN}                                UNCLASSIFIED                                  JDR 321

=============================================================

UNCLASSIFIED           {SECRET NOFORN}

the Land/Sea Test. (33)  These tests are described earlier in this section.

(U) {(S-NF)} One-Way Path Tests, which consisted of transmissions both ways between England and the Eastern Mediterranean via F-layer refraction, showed no contamination of spectra induced by the medium down to the dynamic-range limits of the measuring equipment to excess of 80 dB. About the only ionospheric effects noted were simple spectral shifts of 1 to 2 HZ caused by ionospheric vertical motion.

(U) {(S)} In the E-Layer/F-Layer test, ground clutter returns via one-hop sporadic-E and one-hop F-layer propagation modes were observed and analyzed. Clutter-related noise was observed on both sporadic E- and F-mode returns. The characters of the noise and the clutter-to-clutter-related noise ratios in the two cases were essentially the same.

(U) {(S)} Finally, the result of the Land/Sea Backscatter Test showed that the clutter-related noise associated with the clutter return from sea surfaces was significantly less than that associated with clutter returns from land surfaces at the same range and in adjacent beams. A mechanism in the F-region that can create the spectral contamination at different levels relative to the clutter for land and sea reflection surfaces is difficult to conceive.

E-Layer and Sporadic-E Vertical Motion and Waves

(U) {(S)} Remarks along the lines of those given for the F-layer above relate also to the E-layer as the possible unique source of the contamination (that is, clutter-related noise) of the clutter spectrum when the clutter return passes through or is refracted by the E-layer of the ionosphere. One significant point is that clutter-related noise was always observed, even at night when the E-layer did not exist, whenever propagation support was strong enough to raise the clutter returns to 70 dB or so above natural background noise. And again, the Land/Sea Test (33) seems to rule out the propagation medium ns a whole as the source of clutter related noise.

Ionospheric Modification and Heating

(U) {(S-NF)} It has been conjectured that the observed clutter-related noise could result from the heating of the ionosphere caused by high power of the transmitted signal.  Ionospheric heating calculations, (20,43,44) the Transmitter Power Reduction Test, (25) and the One-Way Path Tests (35,36) relate to this possible cause, as do the Land/Sea Tests. (33)

(U) {(S)} Calculations predict (1) that, at the power levels associated with the AN/FPS-95 and for low-angle transmitted rays, caustics will occur at approximately E-region heights following F-layer refraction, (2) that the ionosphere will be heated in the region of the caustic, and (3) that the, spectrum of the radar signal will be distorted. The distortion, calculations show, should result in an asymmetric broadening of the central line of the radar signal spectrum on the recede-Doppler side. The amount of the broadening predicted is about 2 HZ.  Ionospheric heating would not, according to calculations, account for the approximately flat amplitude excess noise (that is, the so-called clutter-related noise) that fills the entire unambiguous Doppler-frequency region.

(U) {(S)} Further, any effect caused by ionospheric heating should change with power level, and changes in the level of clutter-related noise with respect to that of the clutter or in the character of clutter-related noise (for example, the spectral shape) with transmitter power level were never observed, either in ordinary day-to-day operation or in the Transmitter Power Reduction Tests. As before, the results of the Land/Sea Tests imply that clutter-related noise is not caused by ionospheric heating or, indeed, by any effect in the propagation medium.

Meteor-Induced Power-Flow Modulation

(U) {(S)} Some attempts have been made to relate both backscattering and forward scattering from the so-called "meteor belt" located about 100 km above the earth with the observed clutter-related noise. All over-the-horizon backscatter signals pass through the meteor belt four times, and the forward power flow of the signal conceivably could be modulated by interaction with meteors to create the clutter-related noise. The experiments relevant to possible meteor-induced power flow modulation as a cause of clutter-related noise are the One-Way Path Test (35,36) and the Land/Sea Test. (33)

(U) {(S)} In regard to the One-Way Path Test, none of the results supported the hypothesis that the meteor belt corrupted in a measurable way the spectrum of the forward-scattered signal.  During

322 JDR                                  UNCLASSIFIED                              {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

reception of the AN/FPS-95 signal in the Eastern Mediterranean, we looked for meteor-belt-induced forward scatter by slewing the range gate ahead of and behind the main received radar pulse. Results were inconclusive, perhaps because of a lack of adequate isolation of the gate. But the spectrum of the received direct signal was not measurably corrupted in any way. As before, the Land/Sea Test results imply that the cause of clutter-related noise is not in the transmission medium.

Aurora Power Flow-Modulation

(U) {(S)} The question here is whether the cause of the observed clutter-related noise could be a modulation of the signal resulting from interaction with aurora as the signal passed through the propagation medium.

(U) {(S)} To cause spread-frequency noise (that is, noise resembling clutter-related noise) to be present in range intervals containing returns reflected from the earth's surface, either the main signal would have to be corrupted in passing through aurora-disturbed regions or the corruption would have to be impressed upon non-main-path signal components (that is, transmissions of the signal over paths containing aurora in the antenna sidelobe direction, and so on) that arrived back at the radar receiver at times corresponding to those of the arrival of the main-path clutter returns. For the former - corruption of the main signal by passage through auroral regions - transmissions would have to be along certain beams, namely, the more northerly beams, and at times when aurora was present. One would then expect the clutter related noise to occur in northerly beams only when aurora was present. But clutter-related noise was present with clutter returns from northerly beams whether aurora was present or not. (12)  In fact, clutter-related noise was observed in returns via all beams, including southerly beams, whether aurora was present in the north or not. In regard to the conjecture that aurora-induced noise entered through the sidelobes of the antenna, studies (12) that analyzed the return in each beam as a function of time, range, and magnetic activity made it possible to distinguish and categorize auroral clutter. The studies also found, at times and for some beams, that auroral clutter could increase the noise level in the range bins containing ground clutter, but that mostly it would not. (12)

FEASIBILITY OF ELECTRONIC COUNTER
MEASURES AS THE SOURCE OF EXCESS NOISE

(U) {(S)} In the absence of any convincing conventional explanation for the clutter-related noise, some speculated that the noise could have been generated deliberately. After all, the AN/FPS-95 was engaged in a surveillance of the Soviet Union and the Soviet-Bloc countries, a function that could have been deeply resented. Perhaps this resentment provoked countermeasures to reduce the radar's effectiveness and ultimately remove it from the scene. Admittedly, the notion seems "far fetched"; however, it is not easily disposed of and remains a possible explanation for the noise. In this section, we explore this possibility and describe how it could have been done.

(U) {(S)} If countermeasures were employed, they were not of the conventional jamming type, because jamming in the ordinary sense would have been observed by the site personnel. Furthermore, such jamming would have both violated international agreements and incurred severe criticism. But a jamming technique not easily recognized as jamming might be a distinct possibility. Granted that the notion of "covert jamming" seems even more ridiculous, it is, however, not without precedent. There is a technique referred to by some as "Villard's Disclosure" that provides a basis for covert jamming in OTH systems.*  Over-the-horizon radars generally have large transmitting antennas and high-power transmitters, which combine to produce large power densities in the target coverage area. The actual return from the targets of interest is quite small compared with the incoming radiation and its scattered components from ground clutter. These target returns are detectable at the radar because OTH radar has a large receiving aperture; in the target coverage area, however, the target returns tend to be masked or covered by the large incoming and ground-scattered signals. In other locations, it is also difficult to discern the signals reflected from the target because of the large clutter return that covers the signal. These clutter returns are also present at the radar, but are removed by compli-

 ------------
* (U) Probably because the technique was disclosed by O.G. Villard, Jr., many years ago, but the authors do not have a reference to support this conjecture.

{SECRET NOFORN}                                  UNCLASSIFIED                               JDR 323

=============================================================

UNCLASSIFIED           {SECRET NOFORN}

cated Doppler processing equipment that separates target returns from clutter returns on the basis of their motion. Suppose, then, that an adversary generated a signal in the target area that is proportional to the radar illumination and somewhat larger than the returns from a legitimate target, but much smaller than the returns from clutter. Further, suppose that this signal spreads across the spectrum to fill both the unambiguous Doppler band and the illuminated range interval. The world at large would never see this jamming signal because, as a rule, ordinary antennas are small and no equipment is available to separate the jamming from the covering clutter return. At the OTH radar, though, the signal would not only be visible, but would also mask the targets of interest. It would, in fact, exhibit all the properties of clutter-related noise, triggering in turn a some what predictable chain of events. The first reaction would be that, since the clutter return was not fully cancelled, something must be wrong with the radar. If we found the radar to be fault-free, we would blame the ionosphere. If we exonerated the ionosphere, we would blame the clutter. In the end, we would lose patience and summarily cancel the program without ever discovering the cause. This is indeed covert jamming.

(U) {(S)} At this point, we will briefly describe how this jamming technique could be implemented.  There are of course many ways, but we describe only one.

(U) {(S)} The radar coverage of the AN/FPS-95 was a region spanning approximately 90 deg In azimuth and from 500 to 2,000 nmi in range from Orford Ness, England.  Select 15 sites in this region, anywhere in azimuth but separated by 100 nmi in range. At each site, install a linear array of 16 monopoles with a backscreen boresighted on Orford Ness. (Such an antenna has been built at MITRE for a cost of about $25,000, and it has a gain of 25 dB.) Each site would be equipped to recognize the AN/FPS-95 signal by its power, pulse width, PRF, and direction of arrival. (One element of the array could be used for a sidelobe cancellation device.) Each site would also be provided with a linear transmitter that would, upon the reception of the AN/FPS-95 signal, repeat its signal offset in frequency and in every Doppler cell and in every range cell for 100 nmi trailing the range of the site.  Each site that was illuminated would in turn fill its portion of the total area illuminated by the AN/FPS-95. The 100-nmi uncertainty at the beginning and end would hardly be noticed. Each site would sense its being either in a sidelobe or the main beam of the AN/FPS-95 and adjust the level of its transmitted signal to exceed that of legitimate radar targets by 10 dB, taking into account sidelobe/main-beam receiving gains of the AN/FPS-95.  By and large (12 out of 13 times), the site would be working into the AN/FPS-95 sidelobes, but when the main beam happened on the site, it would attenuate its normal transmission by about 20 dB. The power requirement can be determined as follows : Let

Then

                         PGXA
                S =  -----------            (2)
                       (4piR²L)²

and

                      jgAs
                J =  -------                   (3)
                     4pir²l

If

                J =  10S                       (4)
               

then from Eqs. (2) and (3),

                     10PGXr²l
               j =  -----------             (5)
                    4piR⁴L²gs

Now if

324 JDR               UNCLASSIFIED                 {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

then j = 2.78X10^-4 watts/resolution cell.

(U) {(S)} Since the shortest pulse used on the AN/FPS-95 was 250 microseconds long, corresponding to 20 nmi in range, each jammer must fill no more than five range cells. Since trailing cells require less power than the first cell, the power requirement due to range cells is less than five times the power requirement for the first cell. The highest PRF of the AN/FPS-95 was160 Hz, and with 10 sec of integration, there are no more than 1,600 Doppler resolution cells. Consequently, there are less than 8,000 resolution cells in total, and the worst-case jammer at 500 nmi would require less than 2.22watts. A site at 2,000 nmi would require 1/256 of this power. In either case, the jammer power requirements are quite small.

(U) {(S)} We are forced to conclude that the jamming technique is quite feasible, and is not clear that the experiments conducted at the AN/FPS-95 would have discovered the jamming had it occurred. If experiments confirming or denying the possibility had been conducted, they would have perhaps resolved the issue. They were not conducted.

SUMMARY AND CONCLUSIONS

(U) {(S)} The AN/FPS-95 OTH radar built by the U.S. Air Force on the North Sea Coast of England in the late 1960's was plagued by noise that severely limited subclutter visibility and, thus, seriously impaired the detection performance of the radar. All-out attempts to locate and correct the source of the noise in the relatively brief time allotted in late 1972 and early 1973 were unsuccessful: The source was not found. Subsequently, the program was terminated abruptly on June 30, 1973, after which the radar was dismantled and its components removed from the site.

(U) {(S)} A host of tests were performed on the radar equipment to see if it contained the source of the noise. In the end, the equipment was exonerated; furnished by RCA Corp., Moorestown, N.J., it was generally of high quality and was judged as almost certainly not the source of the clutter related noise.

(U) {(S)} Tests on the environment external to the radar seem to eliminate as causes of the noise all effects except what we have called earth-reflection effects. While the results of the Land/Sea Test, which explored the earth-reflection effects, are generally consistent with the hypothesis that clutter-related noise is present in returns from land surfaces and not present in returns from sea surfaces, the evidence is too limited, both in time and in regions examined, to be considered conclusive.

(U) {(S)} As this paper suggests, a few inexpensive simple, repeater-type jammers with a few watts of power output each, distributed over the radar coverage zone, conceivably could have produced effects like those identified in the paper as clutter related noise. No tests performed at the radar either confirm or deny the hypothesis that jamming caused the clutter-related noise.

(U) {(S)} The strange legacy of the AN/FPS-95 is the enigma surrounding the clutter-related noise. In all the time since the program terminated, the radar community - even including some OTH radar specialists - does not seem to have assimilated either the nature of the difficulty that beset the AN/FPS-95 or the details of the program that was mounted to try to find the cause.  There seems to be a feeling that the Cobra Mist experience was anomalous and that the affliction will not recur. The authors would caution against such a view.

(U) {(S)} The AN/FPS-95 experience may indicate that natural effects of some kind limit the subclutter visibility achievable in high-frequency OTH radars to about 60 to 70 dB. The AN/FPS-95 was the first OTH radar with enough power routinely to generate clutter returns 80 to 90 dB above external CCIR noise levels. Therefore, it is perhaps the first OTH radar to be afflicted routinely with clutter-related noise. But not the only one: During the Cobra Mist tests in 1973, members of the Scientific Assessment Committee visited another OTH radar site, bringing back data records that clearly showed noise resembling clutter-related noise in range bins containing ground-clutter returns. (7)  So, at least in 1973, clutter-related noise was observed at another OTH radar.

(U) {(S)} If the cause of clutter-related noise is an area effect - and some believe that it is - it can be overcome in design by giving an OTH radar adequate spatial resolution, so that the returns from

{SECRET NOFORN}                                 UNCLASSIFIED                                JDR 325

=============================================================

UNCLASSIFIED           {SECRET NOFORN}

objects that the radar is to detect appear at levels well above the AN/FPS-95 observed clutter related noise level. One might require, for example, that the spatial resolution of an OTH radar be such that the amplitude of returns from targets of interest be within about 40 to 50 dB of the earth clutter return. Moreover, regarding countermeasures, designers of future OTH radars, conceived to overlook the land area of an adversary, should remember that an OTH radar like the AN/FPS-95 would be relatively easy to jam and that the jamming would be difficult to detect.

(U) {(S)} It is the hope of the authors that this paper will stimulate informed discussion and debate about the cause, or causes, of clutter-related noise. The clutter-related noise anomaly should be pursued: Resolution of such anomalies almost always is accompanied by a significant advance in knowledge and understanding.

APPENDIX: THE LAND/SEA EXPERIMENT (33)

(U) {(S)} The objective of this experiment was to determine whether or not there were persistent differences between the levels of clutter-related noise within a range-azimuth resolution cell as a function of the geographic position of the cell. In particular, it was desired to compare those cells located over land with those over water.

(U) {(S)} Conceptually, the ideal way to carry out the comparison would be to hold all the other relevant radar parameters constant during the measurement and to varying the geographical area within the resolution cell. Since this was obviously impossible, the experiment was designed to minimize, as far as possible, the differences in these parameters while alternating between the two measurements. This was achieved by comparing only pairs of resolution cells that were close together, that is, at the same ranges, but in adjacent azimuth beam positions, It was judged that this method would be superior to that of comparing adjacent range resolution cells within one beam, because the noise was known to be highly range dependent and also because the absolute position in range of the cells was subject to some uncertainty because of ionospheric layer height uncertainties. As the two resolution cells were at the same range and adjacent, the respective radar propagation paths would have very similar vertical profiles (assuming a well-behaved ionosphere), and the take-off angles at the antenna - and by implication, the relative gains - would likewise be similar. The effect of temporal variations in the clutter-related noise during the cell-pair comparisons was minimized by making the second cell measurement immediately after the first and then averaging a large number of such measurements. Possible differences between the antenna gains at adjacent azimuth beam positions would be revealed by a comparison of surface-clutter returns in those areas where the two beams both straddled areas of similar surface, that is, both land or both sea. It was hoped that gain differences would be small, since five of the six antenna strings were common between adjacent beams. Many measurements were spread throughout the diurnal cycle. This had the incidental effect of requiring a range of radar operating frequency to accommodate different ionospheric conditions, which consequently produced differences in the take-off angles of the radiation from the antenna. It should be emphasized, however, that during any single comparison measurement between adjacent resolution cells, the frequency was held constant. The diversity in frequency and elevation take-off angle was not considered detrimental, since it would tend to smear out any possible (although unlikely) antenna effects that might conceivably generate clutter-related noise over a very small range of elevation angles more in one beam than in the adjacent beam. Since this effect would have resulted in a persistent variation of clutter-related noise as a function of azimuth and range, it could have led to a misinterpretation of the experimental results.

Area of Observation

(U) {(S)} The area selected for observation was bounded by beams 9 and 13, inclusively, and by ground ranges of 1,000 and 1,500 nmi (Fig. 17). It was chosen mainly for the reason that it contained a large portion of the Black Sea and had the requisite land/sea boundaries between azimuth beam positions.  Additionally, the area's terrain features and industrialization levels varied widely. Another important consideration was that the chosen radar ranges should not allow single-hop E-layer propagation modes, with consequent confusion between these modes and the normal one-hop F-modes.

326 JDR                                           UNCLASSIFIED                    {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

Figure 17. The geographical coverage chosen for the Land/Sea Test. 
 {(Figure Classified Secret)} (U)

The beam positions on Fig. 17 correspond to the nominal azimuth directions of the AN/FPS-95 antenna structure. The radial lines indicate the nominal one-way half-power bearings of each beam. Measurements of the actual antenna patterns (45) revealed that the beam positions squinted inward, so that the high-number beams were actually pointing several degrees north of their nominal positions. This fact is important in the interpretation of the experimental data. The range-resolution cells drawn in beam 12 are each 40 nmi long. Reference to these cells is by the numbers indicated in the figure.

Operating Parameters and Procedures

(U) During data gathering, the AN/FPS-95 was operated using the following parameters:

(U) For each run, the data were recorded first in beam 13 for 2 min. Beam 12 was then similarly treated, and so on down to beam 9. The 10 min of data thus recorded on magnetic tape were all taken using a single radar frequency. Subsequent 10 min runs would not necessarily be at the same frequency. Over the course of the experiment, approximately 8 hr of data were recorded and analyzed.

Data Analysis

(U) {(S)} A full description of the signal and data analysis is beyond the scope of this paper, and the interested reader is referred to Ref. 33.  Briefly, however, for each range cell the 2 min sequence of signal returns was divided into batches of 3.2.sec duration (128 samples) and submitted to an off line spectral analysis. This permitted the ground or sea-clutter returns, which are located in the vicinity of zero Doppler shift, to be separated from the clutter-related noise. Measurement of total clutter power was made in a Doppler band extending from +5 HZ to -5 HZ. Clutter-related

{SECRET NOFORN}                         UNCLASSIFIED                                     JDR 327

=============================================================

UNCLASSIFIED           {SECRET NOFORN}

Table 9. Median clutter-related noise power. (33)  {(Table classified Secret)} (U)

Table 10. Clutter power. (33) {(Table classified Secret)} (U)

noise was measured in a frequency band extending from +5 Hz to +15 Hz. The measurements of ground or sea clutter and of the clutter-related noise that  were thus produced for all of the 3.2-sec intervals within a 2-min recording were averaged to obtain smoothed estimates. These estimates were then paired with similar estimates obtained from the same range cell in the adjacent azimuth beam, which were obtained from measurements made 2 min after those of the first beam. From these pairs was calculated the fraction of the total, clutter or clutter-related noise in each pair of resolution cells attributable to each member of the azimuth beam pair. The fractions thus calculated were further averaged over all of the valid 2-min recordings. The number of fractions thus averaged varied from cell-pair to cell-pair and between clutter and clutter-related noise due to the application of data-validation algorithms. After rejecting invalid data, the number of fractions averaged ranged from a low of 18 to a high of 46, each representing 2 min of raw recorded data.

Results

(U) {(S)} The results of the experiment are tabulated in Tables 9 and 10 for the clutter-related noise and clutter, respectively. The interpretation of these tables is best explained by means of an example. Referring to Table 9, the letters L and S indicate whether a particular resolution cell was predominantly on land or on sea. Consider range cell 32 and the line separating beam 11 from beam 12.  The numbers 81 and 19 will be found straddling this line. This means that the noise powers observed in beam 11, range cell 32, and in beam 12, range cell 32 were measured to be in the ratio of 81 to 19.

Interpretation of Results

(U) {(S)} Examination of Fig. 17 shows that two of the resolution cells in beam 12 (cells 33 and 34) are completely over the water out to and beyond the half-power one-way beamwidth points. If it is assumed that the clutter-related noise is returned only from land areas, then an integration of the two-way beam pattern across, and well beyond, the half-power one-way beamwidth would produce the conclusion that the ratios of clutter-related noise in beam 11, range cells 33 and 34, to that in beam 12, range cells 33 and 34, should be much larger than indicated in Table 9. However, as mentioned previously, the actual high-number beam positions, as measured during limited airborne antenna pattern calibrations, were skewed around toward the north by amounts ranging roughly from 3 deg to 7 deg. With such beam shifts, the land on the northern side of the Black Sea would be positioned in the skirts of beam 12,

328 JDR                                         UNCLASSIFIED                        {SECRET NOFORN}

=============================================================

{SECRET NOFORN}      UNCLASSIFIED

thereby increasing the amount of clutter-related noise received in that beam. A numerical integration of the two-way antenna pattern for various assumed beam skews shows the following results for the expected ratios of clutter-related noise in beam 11 to that in beam 12 at the ranges of range cells 33 and 34.

(U) {(S)} This clutter-related noise ratios are seen to be not inconsistent with the ratios of 87:13 and 82:18 from Table 9, assuming existing beam skews of approximately 3 to 4 deg, which is within the range of the measured beam skews.

(U) {(S)} From the preceding arguments, it appears that the experimental measurements of clutter-related noise are fully consistent with the hypothesis that little, if any, clutter-related noise is returned from resolution cells corresponding to sea areas when compared with clutter-related noise returned from land cells. As Table 9 shows, the clutter-related noise variation between either adjacent pairs of land cells or an adjacent sea cell pair is generally much smaller than that observed at land/sea boundaries. The data in Table 10 for clutter returns are particularly interesting when compared with the clutter-related noise data in Table 9, for they show that at the land/sea boundaries, and unlike the clutter-related noise behavior, the clutter levels do not change appreciably. These facts do not support theories of clutter-related noise generation that propose that the radar energy is modulated during propagation to form clutter related noise either before or after being scattered back from the land or sea surface. If such were the case, there would be little difference between the clutter-related noise returned from the land areas and that from adjacent sea areas. 

ACKNOWLEDGMENTS

(U) The authors are grateful to the following of their MITRE colleagues for assistance in the preparation of this paper: William K. Talley, Adolph Fejfar, Nicholas M. Tomljanovich, Fred G. Benkley, and Dominic J. Marino. Special thanks are due Jeanne R. Rainoldi, who did much to make the manuscript more readable. Bobbi Statkus, who produced the typewritten text, and Mary Mills, who prepared the tables and illustrations for reproduction, also deserve our thanks. We are indebted to Charles M. Brindley and Sheldon M. Paskow of RCA Corp., Moorestown, N.J. and to, John Schneider of the U.S. Air Force Rome Air Development Center, Griffiss Air Force Base, N.Y.  Finally, it is a pleasure to thank Col. Erlind Royer, U.S. Air Force Electronic Systems Division, Hanscom Air Force Base, Mass. for agreeing to sponsor the paper.