Technical Report EL-93-9 US Arm'y Corps AD-A269 Iii~lllil ...

Technical Report EL-93-9

AD-A269 956 JuneUS Arm""'y Corps Iii~lllil•!llill!lof EngineersWaterways ExperimentStation

Environmental Characterizationfor Target Acquisition

Report 3New Concepts for Evaluating Low-GrazingAngle Radar Measurements

by John 0. Curtis, Bruce M. SabolEnvironmental Laboratory

Approved For Public Release; Distribution Is Unlimited D T IC• ELECTE

SEBFP. 2, 91993,

93-22511

Prepared for U.S. Army Aviation Applied Technology Dircztorateand Headquarters, U.S. Army Corps of Engineers

DESTRUCTION NOTICE -For classified documents, follow theprocedures in DOD 5200.22-M, Industrial Security Manual. Setion H1-19. orDOD 5200.1 -R, Information Security Program Regulation, Chapter IX. Forunclassified, limited documents, destroy by any method that will preventdisclosure of contents or reconstruction of the document.

The contents of this report are not to be used for advertising,publication, or promotional purposes. Citation of trade namesdoes not constitute an official endorsement or approval of the useof such commercial products.

SPl•'lMrlD ON RECYMMED PAPMER

DISCLAIMER NOTICE

THIS DOCUMENT IS BEST

QUALITY AVAILABLE. THE COPY

FURNISHED TO DTIC CONTAINED

A SIGNIFICANT NUMBER OF

COLOR PAGES WHICH DO NOT

REPRODUCE LEGIBLY ON BLACK

AND WHITE MICROFICHE.

Technical Report EL-93-9June 1993

Environmental Characterizationfor Target Acquisition

Report 3New Concepts for Evaluating Low-GrazingAngle Radar Measurements

by John 0. Curtis, Bruce M. Sabol

Environmental Laboratory

U.S. Army Corps of EngineersWaterways Experiment Station3909 Halls Ferry RoadVicksburg, MS 39180-6199

Report 3 of a series

Approved for public release; distribution is unlimited

Prepared for U.S. Army Aviation Applied Technology DirectorateAviation Systems CommandFort Eustis, VA 23604-5577

and U.S. Army Corps of EngineersWashington, DC 20314-1000

US Army Corps_of EngineersWaterways Experiment N

Station

LABORTORY

Berry, Thomas E.

Environmental characterization for target acquisition. Report 3, Con-cepts for Evaluating Low-Grazing Angle Radar measurements I byTommy Berry, Salvador Rivera, Jr., Bruce Sabot ; prepared for U.S.Army Aviation Applied Technology Directorate, Aviation Systems Com-mand and U.S. Army Corps of Engineers.

247 p. :iii. ; 35 cm. -- (Technical report ; EL-93-9)Includes bibliographical references.I. Target acquisition -- Remote sensing. 2. Infrared imaging -

Military aspects. 3. Military topography-- Remote sensing. 4, Remotesensing -- Military aspects. I. Rivera, Salvador. II. Sabol, Bruce M.Ill. United States. Army. Aviation Applied Technology Directorate.IV. United States. Army. Corps of Engineers. V. U.S. Army EngineerWaterways Experiment Station. VI. Ttite. VII. Series: Technical report(U.S. Army Engineer Waterways Experiment Station) ; EL-93-9.

TA7 W34 no.EL-93-9

S.. .. .. * ,Berry, Thma iE. I II

Contents

P reface . . . . . . . . .. . . .. . . .. . .. . . .. . .. . .. . . . . . .. . . .. . .. . . . .. . . . .. . . . .. . v

Conversion Factors, Non-SI to SI Units of Measurement .......................... vi

1- Introduction ....................................................... I

B ackground ....................................................... 1Objectives of This Study .............................................. IScope of Report .................................................... 2

2- MSFD Radar "Images" .. .............................................. 3

Data Supplied to W E3 ........................ ...................... 3Creation of Radar "Images" .. ........................................... 3

3-Metrics Approach To Quantifying Target-Like Features ........................ 12

Philosophy ........................................................ 12Target Feature M etrics ............................................... 13Representative Results ................................................ 14

4-Backscatter Modeling and Discussions of Terrain Effects ....................... 33

A Simple Backscatter Prediction Model ................................... 33Vegetation Overlays ................................................. 35

5--Summary and Recommendations ........................................ 38

Appendix A: Scene Metric Statistics ........................................ Al

Appendix B: Target Metric Statistics ....................................... BI

Appendix C: Relevant MSFD Test Site Data and Analysis Results ................... C1

SF 298

'II

List of Figures

Figure 1. MSFI radar test high site data covrag ............... ........ .5

Figure 2. MSFD radar low site data coverage ................................ 6

Figure 3. Measure of power returned from test MS6603 ........................ 9

Figure 4. Range-azimuth displays of the "measure of power" metric, MS6603 ......... 15

Figure 5. Range-azimuth displays of the "signal-to-clutter" metric, MS6603 ........... 16

Figure 6. Range-azimuth displays of the "variance" metric, MS6603 ................ 17

Figure 7. Range-azimuth displays of the "polarization" metric, MS6603 .............. 18

Figure 8. Range profiles for target metrics, azimuth mark 18 ..................... 20



Figure 11. Graphical representation of global target prominence (GTP) ............... 27

Figure 12. Cumulative histograms of power data (all tests) ....................... 28

Figure 13. Cumulative histograms showing range effect .......................... 29

Figure 14. Cumulative histograms showing repeatability (heavy fog) ................. 30

Figure 15. Cumulative histograms showing repeatability (moderate weather conditions) ... 31

Figure 16. Schematic of the backscatter model ................................ 34

Figure 17. Backscatter model predictions for the high site location.....................35

Figure 18. Backscatter model predictions for the low site location .................. 36

Figure 19. Backscatter prediction results and vegetation overlay for test MS6603 ........ 37

iv

Preface

The study reported herein was conducted by the U.S. Armi, Engineer Waterways Experiment Station(WES) during fiscal years 1991-1992 as part of the Environmental Characterization for TargetAcquisition Program. This program was jointly funded by the U.S. Army Aviation AppliedTechnology Directorate (AATD), Fort Eustis, VA, and by Headquarters. U.S. Army Corps ofEngineers, Washington, DC. Mr. Nyle Wilcocks was the AATD Technical Monitor.

These efforts were under the general supervision of Dr. John Harrison, Director, EnvironmentalLaboratory (EL), Dr. Victor Barber, Acting Chief, Environmental Systems Division (ESD), EL, andMr. H. Wade West, Chief, Environmental Analysis Group (EAG), ESD. Mr. Bruce Sabol, EAG, wasthe Principal Investigator for this project. Dr. John 0. Curtis and Mr. Sabol, EAG, prepared thisreport. The labor-intensive task of assembling Appendix C was capably performed by Mr. SeanBrewer (EAG contract student).

At the time of publication of this report, Director of WES was Dr. Robert W. Whalin. Commanderwas COL Leonard G. Hassell, EN.

This report should be cited as follows:

Curtis, J. 0., and Sabol, B. M. (1993). "Environmental characterization for target acquisition;Report 3, New concepts for evaluating low-grazing angle radar measurements," TechnicalReport EL-93-9, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS.

XlTIS S •,oDTr' Top(

Ju:;. ! .: c

.<Z/_DI. IDta'. VC" ...

UL;!__11 l I l I l I i I I I I I I I II I I II I IIy

Conversion Factors, Non-SI to SI Unitsof Measurement

Non-SI units of measurement used in this report can be converted to SI units of follows:

Multiply By To Obtain

degrees (angle) 0.01745329 radians

feet 0.3048 meters

vi

1 Introduction

Background

One possible scenario for modem military helicopter operations involves the use of both natural and

man-made ground features to provide cover for the aircraft until the pilot decides to pop up andattempts to locate and fire on a suspected target. Target acquisition and weapon delivery procedures

will ultimately be automated, making use of a number of different sensors whose data may be fused to

provide the greatest opportunity for success. The design of such sensors and sensor fusion algorithmscan be optimized by a thorough understanding of terrain and target interactions with each other and

with each sensor under various environmental conditions.

To this end, the U.S. Army Aviation Applied Technology Directorate (AATD), located atFort Eustis, VA, managed a field demonstration of multiple sensors called the Multi-Sensor FusionDemonstration (MSFD) whose purpose was to collect target acquisition perfcrmance data on targets onthe ground using data from individual sensors as well as combinations of sensors. The MSFD tookplace in February and March 1988 at Fort Hunter Liggett, CA. Sensors were provided and tested bytwo organizations, Martin Marietta Orlando Aerospace and a combined team from Hughes AircraftCompany and Texas Instruments. The U.S. Army Engineer Waterways Experiment Station (WES)was initially tasked by AATD to provide ground truth information for these tests and to conductstudies on the impact of environmental conditions on sensor performance, particularly within the visualand thermal infrared portions of the spectrum.

Objectives of This Study

After completion of the MSFD, WES was additionally tasked with providing input to AATD onhow test sites might be characterized with respect to sensors that operate in the millimeter wave por-tion of the spectrum. The WES staff felt that the first step that should be taken to respond to this lastwking was to closely reexamine existing data collected by millimeter wave sensors with the goal ofassessing the impact of terrain conditions on the sensors' ability to locate and identify targets of inter-est and not simply measuring their ability to find those targets. In other words, the objectives of thisstudy are to determine whether or not the terrain presents target-like features to the sensor and todetermine if that behavior can be quantified and modeled so that one might be able to predict futuretest and/or battlefield performance for a given millimeter wave sensor.

Chapter 1 Introduction

WES has developed an image metrics approach to characterizing te.t sites within the visible andthermal infrared portions oft Uk• •Octrurn. Because radar data like that collected during the MSFI)can be displayed in an "image" form (slant range versus azimuth position), the most logical approachto quantifying target-like features within these images was to apply the appropriate target featuremetrics (filters) to the radar data collected at Fort Hunter Liggett. Therefore, one path taken in thisstudy was to acquire as much MSFD r-•'.tr data as possible, to Lonvert those data to "images," and toapply the simplest radar target featu, metrics to those images to quantify target-like features withinthe environment.

The second path taken to characterize test site terrain with respect to radar sensors was to use avery simple reflectance model to predict the backscatter response of radar energy that is due only toterrain surfa,. geometry at the test site. A point light source was placed at the two locations for theMSFD sensor test-beds and a simple Lambertian scattering model was used to calculate backscatteredpower from a three-dimensional finite element representation of the test site terrain surface. Thesepredictions were compared qualitatively with real data to determine the relative impact of terraingeometry on radar returns. Furthermore, where available, overlays of the locations of trees and signifi-cant shrubbery were prepared to supplement the qualitative assessment of the impact of environmentalconditions on radar data.

Scope of Report

Chapter 2 contains a description of the radar data from the MSFD that were acquired by WES forthis study as well as some details about how those data had to be processed to produce "images." Adescription of the target feature metrics used in this study are found in Chapter 3 along with the resultsof applying those filters to one of the MSFD tests. Chapter 4 discusses the light model and vegetationoverlay studies, and Chapter 5 summarizes the results and makes recommendations for fiture studiesand data collection methodology.

Three appendices are included. Appendix A contains a summary of statistics associated withapplying the target feature metrics to all of the available radar scenes. Appendix B summarizes targetfeature metric statistics as they relate to the man-made targets within each scene. Appendix C is agraphical output summary that contains scene filter results, terrain contour maps, backscatter predictionresults, and vegetation overlays for all of the scenes.

1 B. M. Sabol and S. Rivera. (1993). "Enviromnental characterization for target acquisition; Report 2. Analysis of EOimagery (in preparation)." U.S. Army Engineer Waterways Experiment Station, Vicksburg. MS.

2Chapter 1 Introducion

2 MSFD Radar "Images"

Data Supplied to WES

Through the efforts of the AATD, radar data for 31 separate MSFD tests were supplied to WES byMartin Marietta. Of these 31 files, which represented all of the MSFD data possessed by MartinMarietta, only 24 could be read on WES data processing systems without errors. Data were not avail-able from the Hughes/Texas Instrument team. Details on the format of the Martin data and the proce-dures required to generate radar "images" from these data are presented in a later section.

When reduced to image format, the Martin radar can be thought of as illuminating a patch of terrainroughly 420 m in range depth and encompassing an azimuth sweep of about 24 deg.' Table I con-tains a listing of all of the useful Martin data by MSFD test numbers. Also contained in that table isan average range and an azimuth angle (clockwise from north) that can be used to approximatelylocate the center of the illuminated area. The "High" and "Low" site designations on this table signifythe location of the radar system for each test. Depression angles from the radar locations to the centerof the test sites never exceeded 6 deg.

Figures 1 and 2 contain ground plane representations of the approximate area covered by each ofthe MSFD tests for the radar located at the high site and low site, respectively. The elevation contourlines superimposed on these figures are at 200-ft intervals.

Creation of Radar "Images"

The 31 MSFD test files sent to WES by Martin Marietta consisted of range-corrected (but not cali-brated) inphase and quadrature frequency domain data for the K,-band circularly polarized radarsystem used at the MSFD. Data files ranged in size from 23 MBytes to 73 MBytes and containedmultiple azimuth sweeps of each test area and usually two sets of different range gate data. The term"range gate" as used in this report refers to a set of data that is collected by the receiver and recorderwithin a very specific range of time delays following the transmission of the wave form. Each rangegate contained 20 coarse range cells that were defined by the pulse width of the radar wave form.Furthermore, the data were collected by stepping frequencies in both an up and down direction, thetotal bandwidth of the sweep determining a theoretical limit on range resolution. Data were also col-lected with the center of the beam at two different elevation settings.

1 A table of factors for converting non-SI units of measurement to SI (metric) units is presented on page vi.

Chapter 2 MSFD Radar "Images" 3

Table 1MSFD Test Data Used for These Studies

Range Central Az Central ElTest No Radar Site m rn Angle, deg Angle, deg

MS5502 HIGH 4120 1616 283

MS5503 HIGH 1780 2325 568

MS5504 HIGH 2460 192 2 4 78

MS5601 HIGH 3015 2381 363

MS5602 HIGH 3960 1718 297

MS5603 HIGH 3650 2429 3 17

MS5604 HIGH 3040 174.7 372

MS5702 HIGH 4460 161 9 2,70

MS5703 HIGH 3935 251.7 2,77

MS5704 HIGH 2340 202A4 4 75

MS5801 HIGH 2430 232.7 3.58

MS5802 HIGH 4460 162 1 2.72

MS5803 HIGH 1450 262.3 5 50

MS5804 HIGH 4450 162.1 2.20

MS6001 LOW 3880 284.7 0.23

MS6002 LOW 3995 286.4 0.33

MS6004 LOW 3840 285.8 0.32

M,86101 LOW 3960 285.6 0.27

MS6102 LOW 3920 2865 0.27

MS6301S' LOW 1960 277.1 0,75

MS6301L' LOW 3050 277.1 0.42

MS6403 LOW 1555 297.1 0.70

MS6603 LOW 2540 2879 0.37

MS6701 LOW 2440 275.8 0.67

S = short range; L long range.

4 Chapter 2 MSFO Ridar Images'

SO RC (5 24 30, 79')) 46CONTOUR~ INTERVM IS 200 FT

x -

( HIGH SITE

Figre1. SF raa tethg ie aacvrg

Y.MlrIe

()i

C

| i i

V J

Figure 1. MSFD radar test high site data coverage

Chapter 2 MSFD Radar lmages'

.tW (.f i2'),'820 -'xAJf I L', 400 13H)

(:ft4TOOR INTI WVVA( P', i1$0 I T

,. .'" .

'z'

LwSITE

Figure 2. MSFD radar low site data coverage

The point of listing all of these variables that were a part of the data collection process is toemphasize that each data file contained an enormous amount of useful information. For the purposesof this study, much of those data were redundant. This is not to say that sophisticated data processingalgorithms could not make good use of multiple terrain sweeps, overlapping elevations, and over-lapping range gates, but the study reported herein was not intended to be a target detection optimiza-tion effort. Before any of the target feature metrics that form the heart of this site characterizationprocedure could be applied to a given test, some kind of data compression had to be performed.Acquiring a clear understanding of how the data were arranged and developing a procedure for pro-ducing a smaller useful data set was a nontrivial matter.

Martin Marietta provided WES with FORTRAN codes that could be used immediately to readrecord headers for these data files but that lacked the proprietary subroutine that performed the inverseFourier Transforms for converting to the time (or range) domain. In addition, it was not discovereduntil much later that the way in which the data are stored in the frequency domain resulted in a wrap-around of data in range; i.e., one had to know the algorithm for locating the center of each coarserange cell. This algorithm was finally obtained from a classified Martin Marietta report.

Without going into any great detail, the basic procedure for compressing the Martin files into some-thing more useful for these studies was to create a fine-range resolution, two-dimensional map for eachradar elevation setting and for each test. The fine-range map was established by first identifying a pairof azimuth sweeps, one in each of two successive range gates, with azimuth marks on the record

6 Chapter 2 MSFD Radar ImageS"

headers that were most compatible. This was done hN reading all of the headers and computing theminimum sum of squares of differcnces between successive range gate sweep azimuth marks, in otherwords, find the pair of azimuth sweeps in successive range gates thai best line up. An inverse FastFourier Transform routine was written to produce fine-range resolution data from each coarse rangecell, overlapping data were eliminated, and a continuous string of fine-range results was produced foreach azimuth setting. The end result for each radar elevation or depression angle setting (the elevationdesignated "barO" is a depression angle greater than the elevation designated as "barl") was a multi-dimensional array of data measurements that are proportional to power returned to the radar from theterrain and/or targets. There were 1,400 fine-range cells within each of 33 different azimuth settings,with a separate "image" for each of the two radar polarization combinations (right circular receivedand left circular received for right circular transmitted signals). In this way, data files that consisted oftens of MBytes of data were compressed to two files (two elevations) of about 370 kBytes each thatcould readily be turned into two-dimensional "images" in range-azimuth space.

The intent of these studies is not to find a way to improve signal-to-noise ratios to better detecttargets within natural backgrounds, but to answer the following question: Given a single sweep of theterrain, how many target-like features are out there? Therefore, an attempt was not made to optimizethe data set beyond selecting the best azimuth alignment of successive range gates. Although nottested during this exercise, a simple averaging of the multiple sweeps (subject to some kind of azimuthalignment reasoning) is highly likely to greatly improve the quality of data for stationary targets.Some kind of comparison of data from the two radar elevation settings probably could also be used tominimize terrain background response and to enhance man-made target response.

Numbers alone do not have the impact of graphics in transmitting information to the user. There-fore, several FORTRAN routines were written not only to apply metrics to these compressed data setsbut also to display data and results of analyses in both two- and three-dimensional formats. All calcu-lations for these studies were performed on a MicroVAX II 630QE computer system, while colorgraphics were generated on a Raster Technologies Model One/360 graphics display system.

Naturally, in range-azimuth space, each fine-range cell will be represented by a circular arc, and theentire test data set will be represented by circular segments like those shown on Figure 1. However,because of the somewhat limited capabilities of the graphics display system used for these studies andthe very fine arcs required to represent each fine-range resolution cell, a conscious decision was madeto display range-azimuth data in a rectangular format to prevent any loss of visual information. Thedrawback to this, as will be seen in a following chapter, is that one distorts the planar geometry andmakes comparisons with the results of light model calculations more difficult than if both could bedisplayed in range-azimuth space. Although the shape of the data display in the rectangular format isnot going to be correct, an effort was made to make the width/depth ratio for the displayed data thesame as the azimuth/range depth ratio in the real data.

Another concern about how to best visualize the data and calculation results for each test is how tomake certain that one test can be immediately compared with another. In other words, nothing doneduring the compression of data or the application of metrics to those data should change the relativemagnitudes of the results. One should be able to look at a color display of power returns for testMS5502 and one for test MS6701 and be confident that what appears as a given color on one test iscomparable with the same color on another test. To this end, some effort was made to determine thebounds of data and metric applications to all of the data before selecting a color bar. The absolutemagnitudes of test data are meaningless because the data are uncalibrated.

Chapler 2 MSFD Radar *Images* 7

• • , i i i I I I1

An example of the data is shown on Figure 3, which contains three differcnt representations ol themeasure of power returned to the radar from the terrain and targets at the site (-or test MS66(J3. On alinear scale, only a few features appear as bright colors representing strong returns. As will be seenlater, however, these are not all man-made targets. This type of display is a kind of data thresholding,and is very much a function of the strongest signal within the scene. It would be a simple matter todefeat a sensor system based simply on the magnitude of power returned to a radar through the use ofradar corner reflectors that are optimized for the wavelength in question. The accompanying three-dimensional representation of the linear data shows just how complex the terrain returns can be for thistest site. It is not at all apparent that there are four man-made targets within this patch of naturalterrain, three of which are totally in the open and one is camouflaged with brush. More will be saidabout this observation later. Terrain elevation data contained in Appendix C reveal that the lowreturns at the longest slant ranges are due to shadowing.

The preferred representation of data in the range-azimuth plane is logarithmic such as that shown inpart (c) of Figure 3. Here it is apparent that there are several areas of strong returns and others thatare either in shadow or that absorb most of the millimeter wave energy. The black ovals represent theapproximate locations of target vehicles within this scene. Although many target detection algorithmsprobably operate in the linear domain, there appears to be much more informational content in a loga-rithmic display, particularly when one is trying to relate terrain features to measured data and theresults of calculations. Note in the two-dimensional displays that a histogram of data values has beenincluded. These will be significant in determining the quantitative relationships among various testsand will be further discussed in the next chapter.

8 Chapter 2 MSFD Radar 'Images'

NEflSURE OF POWERMS6603_BAROe RR*• 4 LR**2

88-

68

48

*a lap 29 34 40 see 6 ;9 Soo 900 9

NLTRIC VALIU

a. Slant range-azimuth plot, linear scale

Figure 3. Measure of power returned from test MS6603 (Sheet 1 of 3)

Chapter 2 MSFD Radar "Images" 9

II

b. Three-dimensional representation, x = azimuth, y = slant range, z = measure of power (linearscale)

Figure 3. (Sheet 2 of 3)

10 Chapter 2 MSFD Radar *Images"

ME''-SURE OF POWERMS6603_BARO RR*2 + LR**

25-

15

-aig 16 22' go 46 me 4S so ]1a0 166 9

MITRIC VALUL

c. Slant range-azimuth plot, logarithmic scale


Chapter 2 MSFD Radar Images"

3 Metrics Approach To QuantifyingTarget-Like Features

Philosophy

Aided Target Recognizer (ATR) algorithms for modem sensor systems often begin the process oflocating and identifying targets in cluttered backgrounds by eliminating substantial portions of the datathat are not likely to contain real targets. This process, referred to as segmentation, usually involvesthe application of a gross (in a spatial sense) signal-to-clutter filter to coarse range profiles andassumes that man-made metallic targets produce strong returns compared with the natural backgrounds.Only after areas having high probabilities of containing a target(s) have been identified do the algo-rithms then pr"-s data at its finest spatial resolution to complete the search and identification proce-dure. To answer questions about the ability of natural target backgrounds to produce target-likefeatures, one must not throw away any data. Thus the approach taken in these studies is to generateand retain the finest spatial resolution information that is available from these data.

It would be an enormous task (as well as an infringement on the proprietary nature of many sensorsystem algorithms) to apply all of the target feature algorithms to these data that are currently beingused or being considered by ATR developers. Therefore, a decision was made to apply only the mostcommonly used and simplest features as a first step in tabulating statistics on the occurrence of target-like signatures in the MSFD test site backgrounds.

The basic approach, then, to quantifying the occurrence of target-like features in natural back-grounds is to first select several simple, widely accepted target filters, or metrics, for further study.These metrics are then applied to the finest resolution range profiles for a representative set of test sitedata, and the results are tabulated and stored in data files for further processing. Color graphicsdisplays of the results are generated in range-azimuth space for qualitative interpretation. The metric,or filtered, data files are then analyzed using statistical routines, to determine moments of the histo-grams and other relevant information such as the relative positions of target data within the filtereddata histograms. This latter step requires placing range windows over the known target locations andeliminating the target-related data from the background data to develop background-only histograms orto simply quantify the target signature with respect to the background.

12 Chapter 3 Metrics Approach To Quantifying Target-Like Features

Target Feature Metrics

Following numerous discussions with sensor developers, four simple target lfature metrics wereselected for application to the MSF') fine-range resolution test site data files. These are listed inTable 2 along with some comments about their general applicability. Even though the amplitude met-ric does not require any additional data filtering, it proves to be a strong indicator of target presence;as long as all of the test site data sets are properly range corrected, amplitude measurements fromdifferent sites and backgrounds can be compared.

Table 2Simple Target Feature MetricsMetric Definition Comments

Amplitude Measure of power for each fine-range Appropriate for range corrected data. Mostresolution cell. useful for quick-look, hot spot identification,

Signal-to-Clutter (S/C) Ratio of average measure of power in a Data need not be range corrected if clutternumber of fine-range cells forming a target window is not too large. Typical applica-window to the average in a number of similar- tions are region-of-interest identification andsize clutter windows. One buffer window is target-size S/C for detection.skipped on either side (in range) of the targetwindow

Variance Variance of signal (not measure of power) An indicator of the presence of one or morewithin a window of fine-range resolution cells, strong scatterers. Data need not be range-

corrected.

. (. - 1)

where x, is the signal (voltage) amplitudereturned from the i" fine-range cell.

Polarization Ratio Ratio of total measure of power in the left Man-made objects should exhibit strong,receive, right transmit channel to the power in double (even) bounce behavior. Relativelythe right receive, right transmit channel. smooth terrain should have strong, single

(odd) bounce characteristics. Foliageshould be neither. Large tree trunks andrack outcroppings are expectec to causeproblems.

There was some concern regarding how many range-resolution cells should be included in eachmetric. For example, the signal-to-clutter metric consists of a target window, a buffer window oneither side of the target window, and a number of clutter windows outside of the buffers. After muchexperimenting with the size of these windows, a somewhat meaningful filter size was chosen thatincludes a 2-m target window, one 2-m window on either side, and four 2-m clutter windows outsideof each buffer. Although this may not be the optimum arrangement from a sensor developer's per-spective, it formed a reasonable tradeoff between spatial resolution and metric amplitudes. Larger sizewindows appeared to do nothing more than smear the results. Other filter sizes included a 10-m win-dow for variance calculations and a 2-m window for calculating the ratio of even to odd polarizationreturns.

Chapter 3 Metrics Approach to Quantifying Target-Like Features 13

The procedure for applying each target feature metric (except for amplitude) was to set the filter atthe shortest range for each azimuth, calculate the value of the metric, and insert that value into the

fine-resolution cell at the middle of the filter. The filter was then moved out in range by one fine-

resolution cell and the calculation perlbrmed again. This was repeated for each azimuth until the filter"bumped" into the farthest range cell. In this way, some data is lost at the near- and far-range edgesof the range-azimuth data set.

Representative Results

Range-azimuth displays

Figures 4-7 contain graphics display results of the application of each target feature metric to thesame test discussed previously, MS6603. Each figure contains the slant range-azimuth representationin both linear and logarithmic scales along with a crude histogram. Open circles added to the logarith-mic displays approximately coincide with the locations of man-made targets at the test site.

Several observations are in order at this time.

a. Of the four different representations of target feature metrics, only the polarization metrcdisplays a log-normal distribution of terrain signatures. Power returns from natural terrain areoften assumed to be distributed log-normally, but the power returned from this test site is mostassuredly not log-normal. This observation, which is generally true for the other test sites, isprobably due to a combination of low-grazing angles, relatively sparse vegetation, and thepresence of small hills and depressions that result in numerous "shadows" within each test area.

b. Returned signal strength and the amplitudes of filtered data alone are not enough to ensure thedetection of man-made targets. An examination of all of the MSFD data reveals that if oneknows where these targets are located, the data will often show a strong signature at about thatpoint in space; but other numerous strong features appear to be caused by either the terrain orby man-made clutter within each test site. As for man-made clutter, a consensus exists amongMSFD participants that all clutter was not adequately logged, and this study was initiated toolong after the demonstration to gather any reliable information by revisiting the site.

c. The polarization ratio concept is of no value whatsoever as a means of separating man-madeobjects from natural backgrounds. After some reflection on this question, it appears that even ifone had an ideal target-size double-bounce object within the radar's instantaneous field of view,its effect on the returned signal would still be minimized because of the fact that the volume inspace occupied by a fine-range cell and the half power beam width of the radar is large enoughto result in a volume averaged response that is essentially depolarized. For this test caseexample, the half power beam width covers an arc with nominal length of over 30 m. Thesame argument could be made to account for why so-called target amplitudes are not greatlyabove those of the natural background in many test cases.

14 Chapter 3 Met"ics Approach To Quantifying Target-Like Features

-KWI C !f It

j',tf :I'3l.t'j ' Fl t I~l~j k'

40-

*. a. ta. S.l *• W* e* .q a. •l I

a. Linear scale

rlU'OSURE oF POWERKS&W3 OW.t3 PR**2~ L* 4

20-

15-

b. Logarithmic scale

Figure 4. Range-azimuth displays of the "measure of power metric, MS6603

Chapter 3 Metrics Approach to Ouantifying Target-Like Features 15

., W ITHI 10.11 !1[ W : L ltfAIi. .

40

I NW~. . St

a. Linear scale

'61 hi I TH [BiIF V Wi J NI))

I'5l

_7

b, Logarithmic scale

Figure 5. Range-azimuth displays of the "signal-to-clutter" metric, MS6603

16 Chapter 3 Metrics Approach To Ouantifying Target-Like Features

V ýit- I f! 1A I L( I

Pq. W V N V 4 4,

i i a im m

a. Linear scale

t'o

S.~~~~P tat W* N ffft* ** V

b. Logarithmic scale

Figure 6. Range-azimuth displays of the "variance" metric, MS6603

17Chapter 3 Metncs Approach to Ouantafying Target-Like Features

a. Linear scale

H-: aA - -ý -i T (,

-SAM BW-O -

0. Logarithmic scale

Figure 7. Range-azimuth displays of the "polarization" metric, MS6603


d. While all of the target feature metric cxamples shown arc Ir MSFD data that arc derivcd fromsums of squares of measured voltages for the two diflfrent polarization combinations, the statis-tics will show that neither of the polarization combinations by them,ýelves consistently per-formed better than the other when looked at from the target feature metrics perspective. This isa reflection of observation c.

Range profiles

Consider now another visualization of these target feature metrics, this time as a trace of the metricsalong a given azimuth mark. Several of these range profiles are shown in Figures 8-10. It is obviousfrom the logarithmic representation of power for test MS6603 shown in Figure 4 that a man-madetarget should be quite apparent along azimuth mark 18 (the 18th vertical strip counting from the left);those profiles are shown on Figure 8. The apparent target position is indicated by the dashed line.Another target should exist along azimuth mark 21, but it appears to be obscured by strong foregroundreturns. Azimuth mark 21 profiles are shown on Figure 9. Finally, there are numerous significantreturns from azimuth mark 24 where there should be no targets of interest. Its profiles ame shown onFigure 10.

As with the two-dimensional maps, several observations are in order relevant to these one-dimen-sional traces.

a. With respect to the target in azimuth mark 18, its measure of power does exceed all otherreturns along that profile. The other significant return at about range cell 530 is simply a spill-over from the target located in azimuth mark 17. At this point it is impossible to speculate asto the significance of the shape of this profile because not enough ground truth data exist toindicate whether the reduced returns on either side of the target are due to the shadowing effectsfrom adjacent targets or dips and rises in the local elevation, or whether this is just a strongtarget signature superimposed on a flat terrain return. A closer examination of the target profilewould have to deal with higher order target detection algorithms, anyway, and that was not theobjective of these studies.

b. Regardless of what influences the power profile data, application of the signal-to-clutter metricto a strong isolated return has to produce the characteristic profile shown on part (b) of thefigure. For this target, under these test conditions, nothing in the background dominates thesignal-to-clutter space quite like the real target. There is one minor anomaly at about range cell900 that requires more ground truth than that available to explain.

c. By its very definition, a calculation of the variance of a signal within some window is ameasure of how extreme the variations in the signal are. One is therefore not surprised to seethe strong signature in variance space because of this rather isolated and dominant target. Largeshadows will cause the severe depression of the variance metric seen between range cells 900and 1100.

d. Even for the well-defined target in azimuth mark 18, the polarization metric did not reveal anyanomalies, as expected.


25 dE

MP

LITU

E

-25 dB •

0 1 1400R•NGE CELL

a. Measure of power

25 dB

r'iAI

Mi

PL

TU0E

-25 dB

I RANGE CELL 1400

b. Signal-to-clutter

Figure 8. Range profiles for target metrics, azimuth mark 18 (Continued)


d6

2

-25 dB

0 10

RANGE CELL

Fiura. (Coinclued

Chate 3 etis Apoc oOab~ agtLk etrs2

Z'5 dE,

L

II

TU-0E

-25 de

I 10

RANGE CELL 10

a. Measure of power

25 d8

MP

L lTLI0E

-25 dG -0 1400

0 RANGE CELL



22 Chapter 3 Metrics Approach To Ouantifying Target-Like Features

UD

2'S d e1400

RA~NGE CELL

a. Variance

25 dB

r

U0)E

PANGE CELL 1400

b. Polarization ratio

Figure 9. (Concluded)

Chapter 3 Metrimc Approach to Quantifying Target-Like Features 23

I';

E

aPHGE CELL

a. Measure of power

25 dE

MP

TUDE

-25 de.

F'4NGE CELL 1400




`5 dE

A

-2d 1 IRANGE CELL 140

a. Vardance

E5 dB

TUDE

- 2 5 d B .. ..I I 1 " t ! ' . . I " I

RANGE CELL 1400

b. Polarzaion ratio

Figure 10. (Concluded)

25 25

Chapter 3 Me~lice Approach to Quantifying Target-Like Features 2

S. ... . m . m m •• m m |i |||||mm

e. Whereas the strong target signature on azimuth mark 18 was very isolated, the target locationon azimuth mark 21 does not stand out in any' of the target feature metric profiles. The testsummary indicates that this target should have been in the open but was turned at an angle tothe radar. In other words, target signatures are not always the strongest signatures on any givenprofile. In fact, almost everything in the foreground of this target gave stronger returns. It willbe shown later that the strong foreground signals can be attributed to terrain geometry andvegetation.

f Finally, consider the target feature metric profiles shown on Figure 10 for azimuth mark 24.Without urther ground truth information, it appears that there are no man-made targets alongthis profile, but, at least in the signal-to-clutter domain, there are strong indications of target-like features between range cells 500 and 900. The very elevated power measurements withinthe first 400 range cells are uniformly high and produce no target-like features. It appears thatanything within the natural terrain that could cause a short range jump in power return couldproduce a target-like signal-to-clutter response. Within the terrain at Fort Hunter Liggett, suchbehavior could be the result of the isolated trees at each test site.

Statistics

During early discussions with the AATD staff, it was suggested that by taking a hard look at a goodset of existing data and applying the metrics analysis approach, a quantitative way might be developedto characterize a given test site within a given portion of the spectrum. After a great deal of effortwas expended on this task, the conclusion to date is that a simple, clean way of quantifying a test sitethrough the "eyes" of a given radar system has not been found. There are, however, hints that such anapproach might still work. Ideally, what would be required is well-documented data from other geo-graphical locations that were collected using the same sensor system. Such data do not exist. Never-theless, the following observations derived from statistical analyses may prove useful for futuremeasurement and analysis programs.

Global target prominence. Literature on imaging ATR systems' has defined a metric conceptwhich quantifies the conspicuity of a target portion of an image relative to the entire image. This isreferred to as a "global target prominence" (GTP) and is measured by determining the percentile loca-tion of a target feature value relative to the histogram of this feature for the entire image or for theportion of the image devoid of targets. This is illustrated in Figure 11. GTP is interpreted as theportion of the image less conspicuous than the target for a specific feature. Values range between 0.0and 1.0. A value of 1.0 indicates that the target has a higher feature value than anything in the image;conversely, a value of 0.0 indicates that the target has a lower value than anything in the image. Thisconcept is directly adaptable to the radar features used in this study.

Table 3 contains average GTP values for each of the radar target feature metrics. The "all data"column represents the average of target prominences when the target data is considered as pan of thehistogram for each test. The "backgrounds only" column uses histograms for which the target signa-tures have been masked. What these numbers say is that the targets, in general, are among the domi-nant features within all of the test sites. However, given the number of fine-range cells

I J. Beard, L Clark and V. Velton. (1985). "Characterization of ATR performance in relation to image measurements,"unpublished paper, AFWALIAARF, Wright-Patterson AFB, Ohio.


TARGET FEATURE

LVALUE

VALUE OF FEATURE X

Figure 11. Graphical representation of global target prominence (GTP)

Table 3Average Global Target Prominence (GTP) of Radar Features

(Based on 88 Targets In 24 Scenes; Total Measure of Power Data)

Measure of Target-Uke Features All Data Backgrounds Only

Power 0.994 0.994

Signal-to-Cluer 0.990 0.990

Variance 0.973 0.975

Polarization Ratio 0.925 0.925

contained within each test site data set, there are still a large number of other features whose signa-tures are equally as strong.

Cumulative histogram inferences. One of the most promising approaches to quantitativelycharacterizing test sites with respect to a given radar sensor was to look for predictable patterns in thehistograms of measured and filtered data. The following figures and paragraphs contain some observa-tions regarding the cumulative histograms of measure-of-power data for all of the test sites containedwithin this study.

Figure 12 contains all of the MSFD test site cumulative histograms presented on a logarithmicscale. There is no obvious clustering of these curves that could be attributed to something uniqueabout the terrain at different sites within Fort Hunter Liggett. In fact, disregarding the two anomalousresponses from tests MS6701 and MS5803, all of the remaining test results for power measurementsare quite closely clustered. There are, however, a couple of observations worth noting.


CUMULATIVE HISTOGRAMSHISTTYP=-w/otargets

PERCNT100... ."

70 "

60 / i/I50/

40

30

20

......... .................................................,I"

-30 -20 -10 0 10 20 30

VALUE

TESTNO - MS5502 ....... -S5503 MS5504 --- MS5601- M-- S5602 .... 1S5603 - M- 1s5604 M- S 45702

--- �S5703 - MS5704 M.S5801 --- 1S5802............. S5803 ....... NS5804 . . 56001 5MS6002.. NS6004 . 1... 56101 1456102 M 1S63011

MS63012 ---... 5 MS6403 . .S6603 M.S6701

Figure 12. Cumulative histograms of power data (all tests)

Figure 13 is another plot of cumulative histograms for test sites whose average range is differentfrom all of the others. In a gross sense, increasing range to the test area appears to cause a shift to theright in the cumulative histogram. The net result of such a shift is to decrease the extremes of contrastwithin the test area that could, in turn, lead to less apparent target signatures. All of this could beexplained by the large range cell volume averaging concept discussed earlier. When one combines astrong target signature with a larger number of worker background signatures, the resulting averagetends toward the weak background levels.

28 Chapter 3 Metrics Approach To Quantifying rarget-Like Features

CUMULATIVE HISTOGRAMSHIST_TYPw/otargets

PERCNT100. -

90'

80.

70// , Average Range (m)70 / ", ' 1 -1450

60 2- 1780

50 3 -2340

/O 4-246050/ 5-3040

30 6 - 3960

20 7 - 4120

8 - 4450

10

-30 -20 -10 0 10 20 30

VALUE

TESTNO - NS5502 - S5503 M5 04 MS5602-- MS56004 ---- S74 - - NS5803 M- ?S5804•) 3) t,*) (9)

Figure 13. Cumulative histograms showing range effect

The final observation in this section on statistical perspectives is that the data collected atFort Hunter Liggett do appear to be quite repeatable. Figures 14 and 15 show the histograms of testswhose target arrays were located at about the same position relative to the radar but whose data werecollected at different times of the day varying from predawn to the middle of the afternoon. Thisclearly shows, also, that environmental factors such as air temperature and relative humidity changeswithin reasonable limits have little effect on radar measurements taken during MSFD. There is sometemptation to attribute the differences between the histograms shown in Figures 14 and 15 to grossdifferences in weather conditions, as the data represented on Figure 14 was collected during heavy fog


CUMULATIVE HISTOGRAMSHIST.YP-w/otargets

PERCNT100-

90,

70

60

50

40

30

20

10

-30 -20 -1o 0 10 20 30

VALUE

TESTNO - MS56001 ------- MS6002 .MS5004

---- MS6101 - _- MS102

Figure 14. Cumulative histograms showing repeatability (heavy tog)


CUMULATIVE HISTOGRAMSHIST__TYP=w/ot argets

PERCNT

80'

70

60 IS~:1

50

40

30 4

20

. ....... ... .. .. ... .. .. I-rr

-30 -20 -to 0 10 20 30

VALUE

TESTNO - MS5702 ------- NS5802 ..... MS5804

Figure 15. Cumulative histograms showing repeatability (moderate weather conditions)

Chapter 3 Metrics Approach to Quantifying Target-Uke Features 31

conditions and those shown on Figure 15 to a more moderate set ol relative humidi1N conditions.However, the magnitudes of the data do not support that contention. Water Aill tend to absorb and

forward scatter the radar energy. If moisture was the controlling factor in the dillerence between thesetwo groups of tests, then the data in Figure 14 should be shifted to the left of the data in Figure 15,and that is not the case. The range effects noted above seem to be the best argument for thedifferences.

32 Chapter 3 Metncs Approach To Quantifying Target-Like Features

4 Backscatter Modeling andDiscussions of Terrain Effects

A Simple Backscatter Prediction Model

The second approach taken in these studies of characterizing the interaction of the MSFD test sitebackgrounds with the radar sensors was to develop and exercise a simple backscatter prediction modelto qualitatively assess the impact of terrain geometry on measured data. In other words. "How muchof the radar return can be predicted by some reasonable knowledge of the test site surface geometry?"This was thought to be a particularly relevant question for the MSFD because of the relatively cleanenvironment of the tests compared with what might exist at test sites in other temperate parts of thiscountry or in western Europe, or even Central America.

The backscatter model developed consisted of the following elements that are schematically drawnon Figure 16. First of all, available 25-m terrain elevation data were converted into a three-dimen-sional finite element surface of triangular facets for which vectors nc;mal to each facet are easilycomputed. Next, a point light source was positioned at the l3cation of t;: radar system (either thehigh site or the low site), and the angle, 4, between the light source and the normal vector to eachterrain facet was computed. Under the assumption of Lambertian scattering, the intensity (or power)of the backscattered light was calculated to be proportional to the cosine of this angle.' A ray-tracingscheme was developed to ensure that terrain facets that could not be seen by the radar would not pro-duce a returned signal. Because none of the available Geographic Information System packages avail-able at WES possessed a finite range point light source simulation capability, this software packagewas developed by the first author.

The point light source simulation of a radar is reasonable, because light energy and radar energy arethe same physical phenomenon; they differ only in frequency. The Lambertian assumption for scatter-ing is not unreasonable, because at the frequency of operation of this radar system, the terrain surfaceis certainly rough. Ultimately, the acid test for this model is its comparison with real data. Resultsindicate a strong correlation between the two.

Representative results for this simple backscatter prediction model are shown on Figures 16-18.First of all, ivr-ýs 16 and 17 show the terrain backscatter predictions for the entire areas withinwhich MSFD tests were conducted, with Figure 16 depicting the radar located at the high site (Site 8)

' M. Born and E. Wolf. (1980). Principles of optics. 6th Edition (with corrections). Pergarnon Press, Inc., Oxford, UK.

Chapter 4 Backlscatter Modeling and Discussions of Terrain Effects 33

INTENS1TY = 10cos(p

FACET NORMAL

VECTOR TO RADAR

T R IANGULARTERRAIN FACETS

Figure 16. Schematic of the backscatter model

and Figure 17 showing results for the radar located at the low site. There is a black dot on each fig-ure at the location of the radar, and 40-ft contours are superimposed on the figures. The simple modelpredicts intensities that vary between 0.0 and 1.0 in magnitude. These results were then converted to a40-db range (which is comparable with the range of measured data) and displayed with the same colorbar used for the displays of target feature metric calculations shown in the previous chapter. Even onthe gross scale of these figures, it is readily apparent that the model behaves correctly in the sense thatsteep slopes visible to the radar cause strong backscatter returns and that terrain that cannot be seen bythe radar is placed in shadows (white background on these figures to facilitate viewing of the elevationcontours).

But the real question is whether or not the model can qualitatively simulate the measured data ateach test site, thereby establishing that surface geometry is a major contributor to the test results atMSFD. Returning once again to the same test for which results have already been discussed inprevious chapters, Figure 18 contains a comparison between measured data and predicted results usingthe simple backscatter model. When viewing these images, keep in mind that the measured data weredisplayed on a rectangular format, while the outline on the simulation results depicts the polar natureof the radar operation. A similar set of figures is shown in Appendix A for each MSFD test. They alldemonstrate a strong correlation between predictions based on simple surface geometry and reflectancemodels and the actual data measured by the radar. For example, note that on Figure 18 strong returnswere measured by the radar and displayed on the lower right hand area of the range-azimuth plot thatcould not be attributed to the presence of any man-made objects. The simple backscatter model pre-dicted the same area of strong returns because of the fact that the terrain surface is nearly normal tothe radar at that location. In fact, most of the significant features of the test site radar response are

34 Chapter 4 Backscatter Modeling and Discussions of Terrain Efftos

1 I*4101W I"II *'WI I'

"4 v HIGH SI

Figure 17. Backscatter model predictions for the high site location

qualitatively simulated by this simple reflectance model. Hill slopes facing the radar are illuminated,and shadows exist where depressions in the terrain have been measured.

Vegetation Overlays

Also shown on Figure 18 is a 10-ft contour map with a vegetation overlay. This was produced byenlarging a set of aerial photographs of the MSFD test areas taken within the next year after thedemonstration was conducted to the same scale as the measured data and prediction displays. A con-tour map was generated on a transparency and laid over the aerial photo. As can be seen by the sitephotographs contained in Appendix A, the crowns of the trees and other significant vegetation areeasily recognized and were simply colored in on the transparency. Such an overlay would probably bemost useful for those test sites where the terrain geometry was not such a strong contributor to thebackscatter signal as is true for MS6603.

Chapter 4 Backscatter Modeling and Discussions of Terrain Effects 3

1AIIMI 13 Oi)H'H )o, . 18.ýiAi i,'A1141i

I:IUtT0I* INTf RVsAtI ~1; 40 1'

LOW SITE

~0 v :

Figure 18. Backscatter model predictions for the low site location

36 Chapter 4 Backscatter Modeling and Discussions of Terrain Effects

W"UPI ULF? 1 1ý1w '[IJI"M!,6603 HARP4 Rks? *IRS

a. Measured data

b. Backscatter predictions

C. Vegetation overlay

Figure 19. Backscatter prediction results and vegetation overlay for test MS6603

Chapter 4 Backscatte Modeling and Discussions of Terrain Effects 3

5 Summary and Recommendations

In an attempt to develop a scene characterization methodology for long-range (2 to 7 kin), low-grazing angle (less than 10 deg) active millimeter wave radar systems, a twofold approach was taken.First, slant range versus azimuth maps of radar backscatter measurements were filtered using the targetfeature metric technique to collect statistics on the occurrence of target-like features and to relate thosesignatures to natural terrain conditions. Second, a prediction of radar returns from the terrain wasmodeled by a point light source at the radar location and the assumption of Lambertian scattering fromthe terrain facets. Model predictions were combined with overlays of vegetation at each test site andcompared qualitatively with the measured data to assess the impact of terrain conditions on backscatterresponse.

Application of these analysis tools to the Martin Marietta radar data collected during the MSFDresulted in the following observations:

a. Man-made targets within the terrain backgrounds found at Fort Hunter Liggett are reasonablyeasy to locate in range-azimuth "images" when their positions are known, but might not beeasily found by automated methods applied to either raw data or filtered data because of thestrong target-like returns from the terrain at some test sites.

b. Not enough data on different types of target backgrounds are available from the MSFD trials todevelop an effective statistical tool for quantifying background conditions for differing terrainconditions. The use of cumulative histograms for achieving this objective is still a potentialtool, but more data from different backgrounds and collected with the same radar system isneeded to pursue this train of thought.

c. A simple point light source model with assumed Lambertian scattering qualitatively correlatesvery well with observed data. Vegetation overlays serve to enhance that correlation. Thisimplies that for the type of terrain offered by the Fort Hunter Liggett test sites, even crudesurface geometry models and vegetation overlays will explain much of the backgroundresponse.

d. Many questions about the impact of terrain on radar returns under the conditions stated aboveremain unanswered in part because of insufficient supporting data. Future tests in which radardata will be closely studied must include more comprehensive ground truth measurements. Forexample, had it been known prior to the execution of the MSFD that a detailed look at the radardata was going to be taken, a complete set of low-altitude, oblique, high-resolution color pho-tography and video of all of the test sites could have been collected (from at least two of the

38 Chapter 5 Summary and Recommendations

cardinal directions). This would have answered any questions aboui vegetation conditions orabout the occurrence of man-made clutter and such natural features as rock outcroppings.

Having developed the data management and analysis tools for this study, it would be relatively easyto test other higher order target feature metrics. In fact, a database now exists for this one set of ter-rain conditions to take a closer look at existing target detection algorithms or to develop and test newones. One very straightforward task that could be undertaken and that would require nothing morethan to revisit these existing data would be to examine the impact of averaging multiple azimuthsweeps or comparing results from different elevation settings as suggested in Chapter 2.

In the area of modeling, it appears from the simple exercise conducted within this study that modi-fications of the point light source model could yield a valuable tool for predicting terrain signatures.Examples of such modifications would include more realistic scattering assumptions from grass-covered terrain and a coupling of tree and bush spatial information with a separate module for estimat-ing an average backscatter from each species.

Chapter 5 Summary and Recommendaions 39

Appendix AScene Metric Statistics

The following pages contain a summary of common statistics for the log-scale histograms associatedwith each of the 24 MSFD radar scenes. There is a table for data that represent the total powerreturned to the radar receiver (BAND=total power) as well as a table for the power measured on theright receive-right transmit channel (BAND=polarization RR) and for power measured on the leftreceive-right transmit channel (BAND-polarization LR). Each of these tables has, in turn, separatesheets for each radar target feature metric (METRIC=measure of power, METRIC=signal to clutter,METRIC=variance). The final table summarizes results for the polarization ratio metric(BAND=-LR/RR, METRIC=polarization ratio).

Apedix A Scene Metric Statistcs Al

40 Sm 00 0 03 m Cq0w00 0 w 0

go

*0 40 n,0

It C W 00 ~ Q W .* 'r ~ n

4Wt-4 arr s.N cý C t c !c049 Oa! 1 9 9c! o llý

431

4D

V o CC c2 o . . .

ea 0q----

44 a" l w w a- 4 I a"A a nnA

.0~~E R a 2 EYRE00 0 00 0 00 0 0

a2 ApeUi A Soone Metric StatisticsQ04nm

C!0!0 ! C 0 0!0! !C440 440 . 0 0 .. 4!,tV!C!1! ! 0! 04

44 . 0 00 0 0 0 0 0 0 0 0 0 0.4

a f000 0 * 4 Cff0 0~.04

00>4

n in

a .0

* 404000000000000000 OO4r C!0

an "444 m "A00w0RE FEMER MEU.R F

a 0 0300 000 0000 0000 000

aped A0 Scene Meft Sta...40.N404b40

to vI4 nII #5 5 $ 1 1 5 I- I I 'l cIIw~

1.

u

.0 1 - 0 0 0 0 00 0 0 0 0 0

a

u 54

L) t= o c o C o nI 0 -tow W000 0 00 00 0

0 9;

o0 10 1.CQ 4 1 15 . .- V .4

.54

"" 0 0 0 - - T 00-10 C4 '1 03 M 0

94 5i C0 ...4- , III5 .~-. t .5 Qý . 4.-4N.4!0!1 .5

I ~V C! n Si S I ! Iý S I S

.4 4. ~ # O 5 0 9 0 5

o ~ ~ W *r oI %m In.5 J i-.5.sn ori i wtn.4w w.Jw r4o54 5 4g 5 s u I 5 5 9 5 5

A4~~~~~~ ApedxASee en tfsc

Iý m C) - tNC

I~ ~ 0 1!.4N a

040

0 - 0000N0N000 00000- 000-44 0 0 0'N4 N 0 0 I .0I0)I

1a £

S 4.

*1 IN "I00 0 0 0 0 0 0 0 0a U OuO o '00 a 0 O 0 0 0 0

a U ... 000 000 ..'...a..... ....o-a4! -aCC 9~ !!!!9~~!. Q

0 £40

*c~l~o o 0 CIO 00 0 00 0 09 0000 0000000

41a CD aQ a) 0 a a a

£4

Appendx ~ I A SewMf tbs A

00000O00000000000q�O00�4..E.4 �

I 33UE

I -4 000000000000000090000000* .4 flNONOOflO'.0Nt.N00N30014.'fl

4 ' Inr�3�3fl3.1�.3.1flflar1flflnn0�1nnn.#. 444*�*0433

0.

* .4 000000000000000000000000* - 0000n040030lCO.�4.000300

.34 30 3.4.4 l�4I0.

0.

* P 0000000000000000000000000 01.W�N01�O.4N300404

0 0* a ooooooooOOOOOOOOOOOOOOOoo I I 433433 I'3

U.33 0 000000000000000000000033'0

* U nnen.gwnqnWnW *tfl131p�3�C01030301

* *0 000000000000000000000000hI 3 P 3 3 4 4 3 3 3 3 33333 31114133 II

01 34

C 43 .4 00000000000000000000000044 44 ...3 0000100101434"C0

4433 4 3 0 NN.4N.3N.4N4 N.-1-"4 3�4NNN3)O (3 V 3P 3 3 3 I 3 3 3 4 4 I 3'34 NOo 34

* £3.30 �o .4 U3-3 U C U

> 01 .3 000000000000000000000000.4 .3 0tO4t.r03N0flI�flN31*30C0030

33 U .3.4 * I £3 *ttetCV�.c43fl3*ta.t�01V3V101N3.3 Ci flU 343331333333133311333333

(31 44 3.4 (3(31 9(3.z *�

S 000000000000000000000000VI 0101 � 01 � 0003 .-3 0010w 0300010101 3( 030�4

3.4 U0 C 44 01�N�01010�00 003�3i 003�001o 0 4�4.4r�1r3---N4.4.33..4NN,4.'3U *.3 -4 3 3 3 33 3 3 33 33 3 3 3 3133313343U

34 0101 .4o 34 U 01.4N0101.efl300r.NNN001010001�.1013'd3.�44 U .34 �* .4 U fl...3.t-..p�01...000101N.40.N.-3000

34 000.40000N0N0000..N=1e.-...

* 3 U �NN01NN00�.4..400N.3-3t1t3,N001U 3 U 0Q0101P..N01.3q�flNflfl001010013.1.43..013.,

3D U �@00130.#N01N0...fl0....flN4lN0.3..01...0U �0 3 U 000000000000000000000000

3 .01 333 333333333333343333 '3*3 3 U

44 3U 3 043 3 00 000000000000000000000000* 3 14.3 3fl.l01.0.1VlNfl4flo.4�flr40..33nfl��

3 P443 t U �

NNNNNNNNNNNNNNNNNN£31NNNNN3 44U3 U�

4 0 00000000.00000000000000003 U 01O.q01010101W01.q0301N01W01..01o003.-3�3 U3 B 000000000000.-4000000.-300.4..�3 I 3 3 3 1 3 3 3 3 3 3 3 3 3 3 34 3 3 4 3 3 3 I

3 03 0013 '.03 .3-43 @43 0000000.....43.3.-boOo..3000003 >433 U.3 *-3UI U

- N3 NC.4Nfltfln..ebn...4Nc.-3N -4-4flfl�3 OOOOO0000000000oo00�o�o�3 .3 flflfl01010101O0OOOOO�$�.W�* U 30flflflflflflflV3V4fl433(3V301010101010101

0 101

0 14 P03 443-

A6 Appeo�x A Scene Metric Stabsti��

a 0 00 0 000 a! cý 00mC, C

ET -

. to OCoomowo A ooo oco ooo

-4 Cý NCn D .I I r1 .C! W! V! C! -

.4.

.0.

m*n 000w000 000000 00000 0 0

1!0 !* !C 0! r!Of l 100! C! .00c o 0 0. c f ,

~owo

Apw- A sw Memo- Nttsfc A744. 4f4 - .N4 nNI

* 0 00 0 0 00 0 0 00 C O

0 44

PC .-.-.eg...

a .4 0000 0000 000 C.00 00.00

am WC O C W v, t O N .

A8 A penc A S ene ~tncStatst3

0, 0! 00 0! 01 00 V!01 000 a0 000v

* - C~w N.lmr) 4~n.Ow~

m m m4-. n -M0 mm4M

Sl

034

m a. *I D o"m4

s -4 ww Nn

U ~ ~ ~ N 0 0 N -0 I 3 I 4

N N 14 94 C4 I ~ CI CQ eq11 m NI r44 4 4 4N

*~ II

*o o -c o - - - -000 01 0- 00o .- o

Apeni A 6bn Mat StftI A9 ~ I I I II I ,

7. 7

G C4 m

u 1J 4 4 4 4 4 4 44 4 1 4 1

60 1

4 f

0CL

.4 .

*, 6

4 CD0a I .I

1 44

4 wi ýn o w

A10.N44~4-N'4tt 4M4444.

44~~Apec A Scn 4eh 4 44 4 444 44 44

0 4 05~ . 0 ; *C

0.

I~C .4 ioo ooieoo4oo oec C O O

In0 -,V

* ~ ~ ~ ~~ .4 .0C. . . . .f5 . I .. )0

a w 0 o m0.0 do0icn oo. t on a r4 o r4 n cjm w

0, ..,, .

id I

* .oooo eeoo o00 0 00O- 00ma

0 5Z Mm w

* ~ ~ ~ E E RE RENN-~ N 4

App~xA Sen Mtri SaftoaA0

Appendix BTarget Metric Statistics

In an attempt to make this document as useful as possible, statistics on radar feature metrics as theyrelate to the actual targets were compiled and are included in this appendix. The procedure for com-piling these statistics was the following. First of all, a mask was created for each scene that servedtwo purposes. One was to remove the influence of a man-made target on the radar return by effec-tively removing from the scene a portion of one or more azimuth slices that was centered on the likelylocation of each target. The other purpose was to identify the strongest return from that segment asthe target return, or metric value. The target metric value was then compared with the target-freebackground to produce the global target prominence for that target that is defined in Chapter 3 of themain text.

As with the previous appendix, data are presented in terms of both the type of received signal con-sidered (total power, right receive-right transmit (RR) polarization power, left receive-right transmit(LR) polarization power) and the metric being considered (measure of power, signal-to-clutter, vari-ance, and polarization ratio).

Appendx B Target Metiic Statcs BI

I ~ ~~~ .M .00 0 0 0 . . . 0 . . .O.4 . . . . .

N~~~~C C4 100 NfI . .4 .N .l .

co a. a_ 0 Z, - Z ow 0.0 a 0ama.m am w 0a -*

* ~ ~ SM C1.040 0 0 0 000 10'1 00DN

I ~ ~ ~ a m m am. amo@. momaO,0 0100 00

O 4 uH610 10 0* 00 a a W m a0000000

0000~*AC 00000r.0N0 rMwr-05* 0 0 I110. ==Am=O~I0e,.

* I C- AIA~l~ N@N10I~lA~*~t~01001001

H I r.NtI NN.."* "F

Am *9 112, DA IK -3 04 n cc a 1"X "sc .4 0 oo

OS ~ ~ ~ ~ ~ - m00 0 t- 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0H~~~ ~ o. aH a00 0 N 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

B2 Apenh B01OTargeteI00IMetric*N00IStatistics10V I

------- -O , oo 0 oeeO wzeCOgzwww eow eCC w SEF 1P E 2 gMVVYRM

2 9 E

--0- N00N

C! C C OO ! Z C C! CýCOCtC!C!C!C!C!t=5 14 o O ~ C C o c o c O O e c c c~

0 0- 6Cmo 34$ -- an~a , vn WN mgO-0 20

"0 Cý1 n t 1 i1inC! h ~ Ca! 09 Ct C!O O4 ! ý 9 ... ! ýC!14 43 Oc~ flf mf

91~~~Aw 2 A ---

omXma e c~c ceec oc C C CO OCOC COOC COOC! .r~ .OC OO O .O . C . .. . C . . . . . OO.. O..

*~a 96a

* W N ONO is N~INN0Zne go Y A n ~ o ~ C

* ~ ~ ~ Z.9 F 6 s.., 000WCO~b ,. O aa o o~ o*4 2Nis N N N ~ Nt. NN~* isi NNN NNNN02

~~~~1V ao. .10mcecc O c e ce ~ e c ~~ e ~ c c~

a. C O C O C C CC O O O C C O CC C C C O C 2

Li 44

Ap en i 242 Ta g tM ti ttsc N NOeO , a ~ ~ O. O i ~ o e ~ c e~ B3

C! Ca 00000lc

*;

.4 0 C* Co

0 m0

11 001000O0000!

*3 -W

B4~~~~C Apech TretMtrcSttstc

00000000.4. . . .040 . .00..0

000 ~ ~~ .0 00 0 00 00 0 0l 0 00 00 00 Z 0 0 00 010 00 00 0

0 000 0 0 00 0 000. .0S 0008

ONI* 0CC. m N w n Co O a 0 aCCCONt. 40 0N . 0NNg~C

C w CC0O t4 00 0 00 00 0

460

so a wm4 .4 C 4 4 0 0 0 N C C Crg C * C % 0 . C C ~ . C C N C

* COIDI... 0NC I....... ...O.

40

a 05 0 00000 N . 0 0 0 0 000 0 0 0 0 0 0 0 0 0 0

It um u l . .4 1C00 it0 0 00 4O CN IDaC CCItCO 4NzC

.33

6itNE 1II4 U k 112 2 -.3 9-KNC 4N C N0 C 0C 0 NO 4 NN

AP"k 5B5 CNw00e 004 Metric StaCfCCNtt. CCC CB5

I ~~~ ~o C 00 0 0 0 0 0 04-.00ý000000e0 0000

0~~~~~~ 09! 00 0 0 00 0 0 01 0 0 0 0 0 00 000 01 0

Z 0 kQ ww M 0 Mw w0000a *0

641

*~~ c 00000000000 000 0 0 0 0 0 0 0 000 00 00 00 a0t0 0!00 0 0 0 0

.4 CS 24.3Z- NC n. . .4 0 0 0 0 ~ S

*0 * - 4 40000. a . 0* 4 04 lk4 2 * C0 C 0 V 0

aaa ~ 000 0000 00 00 0 00 000 000 000

U .U3fa 0 933t0 0 0 0.~ 0 t* O 00 0 4 0 0 0 r

* ~~ ~- "B L'. 05 .. ONC....t ix4 A. C C O O O O

ai x~ x0 00 0 0 x00 00 E z00.0 E000Z0I0E

B6*o-~. tN.. tP. .t % tt1 N.N 1 N N-Apec B Tre Mti taitc

.44

o a

10 A 0

to I - vgv

to

S .00

HH

*1B7

Apeci B Tage 0eri0Sats0c

I -40

. ..... fn..44Ni0 fllf....e .N NN 4lfffl......-N

EO* 020220O W C N N0 r~f 0

.. .t C. . .2 . . . .N .4 .N .0

I 400

* 09 0 n 0 S 020 . .

mm ft £4OO~ 420-~ mrNNWN.Nenb C t,1

A A.0 I 4. t4 2 4 4 0 0 N 0 N 2 0

I £5 4 O=2 M I@ 0 N NW 2 e C* 4 £,* 09 4020. oc0m 0220,,Cn

z t . . . .21 10 0 .N C 0 0 !0.N.... .C.Cat

n*. C; i 4C iat* !1

.4 £.4 *0

0 020000 0000-020010 oCo On-.N~ wc

o ~ ~ ~ ~ ~ ~ Ls W N 2 N.0 4. 0 0 0 2 03 H ER.20 0 .n . wvr0

*1 E

B8£ - 0 0 0 0 0 0 0 0 0 00 2 0 0 0 0 0 0 0 0 0 0

.46~~~ ~ ~ ppnx8 we 40

2 400.40 0N N N @0N 00202020 0 . c Sais

-0 r

1! 4, 0 O OV! O O!O . IO.. . . . O. O OO.OOC!C

.0.

an-- - W- ft 4-NN

I ~ ~~~~ 00 0 0 0 0 0 0 0 0 0 0 0 0 00000 0 0 0 0.4~ ~ ~ ~ ~~~~ C C C C Q O O C O , . . 4 i if f O;t t

. f!40 peo oa o c P.O gc coec cooc oI ~~~s .. 40 0 0 0 00 0 c O 0 00 0 0 0 0

.34SOi-.431 r, .4% 0 O I.-.4 * O 4

Ap e c 84 Target40 W4 t* Metri Staisic B0t 4 . rf.4 4

>4)* 0** .40

.3 W40,.rN- *0,0C(CID*0

. �I; 00000

* 4)13 00000

* 00 44fl40)tV02 IflV1fl*�

'V - 3�03"IV�4�O I

0re.-. 00000I'CM 00000

U S 53 00000* I £0.0 I 04) flOote

130 IQOIN0�4* I C'-' *OtC

* I *-� 0C)00000 .?p..flNN

14 * CI) 0N�WC* I) 004) C 144) *OClC*

* 0 W=�C�o 4) 0 .r.trS.

W .4 14 NNNNNU 0

.3

� 009 C .� 0 ,..OaOC *��0

4)4)13 kO *� 00000

- &34 014-50.

5- *3* 0 13

00000ON 3� 50 N* C 04� WI � IAN 0 N 41.4I

13 I 00. I .3

.4 I* IU S

.3 3 -.4-J

* I UAS-. 35

.3C' 000043

30.4004

C 0

C 0* OOU 0- 1l.S�vI 13.14 5.54)44 5.5o .i..s� ..J

0* 4310�14 04.�41

� � '�

810 Appendix B Target Menc Stabsbcs

* 0.4 oeoo---------Oo- -------

0 0 0

pi -lfff~lf .fl ff4 tt N f - - - - - - - -I ~ ~~~~~~~ OO O C OO O C COaOOO O C OC C CIOO CI ~ ~ ~ ~ ~ ca :a aT1 ~ ~ l C C C W C C 0 b r ~ r r r . C C

* 4 -a .e . .e. .o .o.

80 0-41 C~VCD 000f C.40 0 00000Vft

4300

"94 IC..*O 1-4 '0 a OOad 04 3 A m

*2 2O offl EnS2e4.-.V9eTe t.-4N fa

Appenx 8 argetMetr c Statistic O4. tl

000a

a 0 0 0 0 0 0 0 0 0 0

0a0 00 0 0 00 00 o o oo0o oI0 C10 0000 000w000

-V .0 00 0 0 0 0 0 0 0 00 0 0 0 0 0 0 0

04ko4 9 0~ V~l.. 05. 0. C . * 4 c e 5V, m n n0 fl n ,0 0 N 0..-r..Mac~v.

*44C4 4 4f4 4445* 454 a' a C; - -- -

.000

840 O..,-~~~~~0.4 0 4 ,r a.. 16 - ~ ... n a . oOS tV~ ln .t4 .C ft ft 44 504 .N 00000

040 W 0000 00 400 04 0 000 0000 000NN444fr. 1I~.N. . ~ 4 . N

0~~~~~~~~ R000 0 00I00 0 00 00 0 000to0 00

SW ~~~~ ~ t m 00 0 00 0 00 0 0to00 0 00 0 00 0 00 0A084 ~~~ ~ ~ ~ ~ .a 00 0 0 0 0 0

0 0 0 0 0 0 0 0 00 0 0 0Nco

aa w aj

84 0 0r 0C 00C00 4.4 . ~ .... a...4 Iee..Dc

--- -- --

B12 ~ ! Aped 8Tre Mti taitc

* ow

-c0

* 00000* *3 C C C!

04

0000

0000

* 0.. 4 0010Appnci 8Twgt Vmd *Stas

-- - - - - - - -

za EER 0000.0 m 0 M22 VM 22 22 EP M H I0

.0 . . . .

0140000 00 0 00 0

0 1 00000N 40*taw0N 0N ao,0 a 0 a0V0 0a00Q00Oa.. S. Ct Nf 0 .. 4 C0 0t0 00V.

DQ Q* toc maa 40 M C

a! a .o . .s .n C! 134 C! n !I! .. ! 2m ,aZ,41 CO13 ., C1 Qa4 o .a 0as0Q

toa n- e 000 lom 00000 11a0000U ~ ~ r IC m.4 ito.4V. 'Olh 0Q lt toQW.g

a a 3.) N..~ato .. nto or~ tool ew

B145aped B agtMticSaitc

0w

.30

* .43 00000000000 00400 00000 00000

Is .4

* ~ ~ ~ ~ a c o s C Q 00 0 00 0 0 000 00 000 0 0 0 000 0 00 0 0 00 0sa a0 000 00000 000

* 54 n...MWC~flM. C . ...... NOCN . =- aM O

a ,. * C !M M S . ! .. ... . . t C fll. CC.CV.

10 C. C;. 0 000 a0 0000 000 0 0a0 0 0 00000000000000; a

.M I 3.05oooo OOOO OO 0 0 00 000o eeoo Moo

P3 - wnam 4c -n. 00 00 00 00 0a 00 0000000 .0000004*ýC N

'.3 a

I I3a I~ 00000000000000000000000000000000000001 w 509Jw000

Cpe cx8 T re Meri CStat c B00 0 0 00 0 0 0 0 00 0 0 0 00 0 0 0

0 0 a

44 ! C! 4 4~

00m

00000

00000

o 40

B1 Ape4xB Twe oi Sta00000

*4 - m 000ý0

40 D 0

1! 1 !. !3 . . . . . . . . . . . .

3 6. 000 000 000. 4.40 .4 C 0 0 0 0. o.. . . . .. . .. .. .

C40 w060C0 m0 0000000MAW 3) 44 P. a W 0 0~ O .0

* ~ ~ ~ ~ ~ ~ ~ ~ C 14-O N C . 0S C 0 0 0 04 4

to. 4 .4 0 0 V 0 00 0 0 0 N

*04 6 g w 004.4== OC N-ess: N0..C! * ! at 0!440 . . . . . .-4 .* 4!

Coco ~ 0. 00 000 e000 0000 lot0

I u -- ,,N------ ---- aNOR

(4~~~ ~~ Nea ag40 4 N W ON = .40"MO W" -, 1 11 1

.0 2 13a N* 20 E 4OeP4Vre OOO4-00m4=2

*ped 84 Target* Metric St.N04%0N0B17

I ~ ~ ~ 0 009 o o S o .OQeO~ OeO a: 4000000009V 0o !C !C ! 0 .C ý C ýC

. 0 8 0 0. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

00 00 00 00 00-0 00 0 0 0 0 0 0 0

No 'w4 00 0 00 0 0 00m"00 0 0 00 0

.11.

e40evnnr.o~ o 4

*a so 104 w

.0 ~ ~ ~ ~ ~ ~ ~ ~ d NE N33t0 eN C ~ .0 Ne O * 0 l 0 0 0 .,* 4. fl NNO4 010 .0N CINO NO N~e.NN ~ .NO .40

* .5 .t¶N ~ lC C .. tI '4C 0. fl~rIV....-~tOOV2* ~ S I- *O S @ 0 0 00 0 0 0 OI 0 C O O

* ~ ~ ~ ~ ~ ~V N ~ 0 0 e 0 0 0 0 0 0 00 0 0 0 0 0 0 0 0 0 0

B1 Ape3h Bl Target0 Metric00000000000000000,4

04-

4 00.00

4 mI 00000

* 0000*0 a a

OU 0T000

46 Q~ 0

14 19

Apwx8TrgtMti Siitc

w C4

w . w. 000 000a00 Q.40 w 00000...4

* 5*ý* 0

*~- C 0WN Q.202-0 3 NN ,.0

a ~ ~ 1 401 at.242 0.0 02 2 0#0 0 440N

1. k 4 00m0C - ae 0we ga~~

.N w 'm0 0.-20 COD to a a 0 0 A

sd a .1 N202 W% to I a 0 w 02020wa 4

a. a, 9

020

I 3r*~~~ a * 000eS0000000000000000

0*0 02 to 6

.3 ~ ~ ý ac. .w 2ffl 0 24 O4aCOKA 19-1 Ct0W-j Z4S0P).MN0fl02

* ~ ~~~~~~~ Mimi;lr-300302~-00 Na CC 02e~0~4 C 1

B208 N.~---a=ra0a2o2n.~fc2t0 @.N2222lv02 ~ ~ ~ ~ ~ ~ ~ Apni 84 Tage Metric02220*0222000 StatistimNNNN .

* ~ ~~~~ a. ..... 0 0 00 0000a0a.a 4..00OOOaOaO 00001=0

0040 CO0000000

* 1000000000000 00 aaO osoC. cooo* 100

ogooooo0oooo 0.00 0000000000

0 00 00 S000'.00 00 0 00 So 00c 000000000000000000000000 0 000 0 0 0 0000040000

340s 0a- -a..N a- 04 psm In(4pp.0c p..C, 0N d e eV "~ W.e "lm.. 2: =ao Oasamaaa m-m 2a w "2XMmam amme

p. pppp N . . .N . . . p. . . . C!C t IC

44 40:000000": 00 a 0 a aa m CU3 0,0, m0 000a8a*. . . . . . . . . . . . . . . .. .4..4.. .a m p. m a. .. . .

04 0I b0 0 0 0 0 0 a0 0 0a3a0 01.4 0 0 0 I00 0 aa0 0

aomt-OOVa.004b

00 .m 4oema - ma a n 00 go a w 4"o 0o m 0 0 a 4D aw o a " 3 a m -

0~~~ on 00 eo~ S .op.Zo 1H

34 0

1 mo mIn N m

.4 3 O .A. &

A0md Tare 000i 0ttsc B21

C-CC.) 00000

I 000CC* .8 4OWr.r"-

S ICIC8DIOW

8 � �* 484 00000I .854 *.8 O88480�* 55 .4fl�@V* 59 VS�WC

*0 I -8 r88�r4�41

o8�* I E� 00000M I 8C5 00000o .884 00000S ASA I �

I 84S W0�.0V�* 09 .4'4C�1V�

C- 008waa

*.. 00000a .CS *-.8I�-N

�k* 048 84.8 tenOr

* S @000*4o 84 6 .rbO.

a �J � -o -8-8 6 5

S 48 8.4�

C @0000

� 5� .ZI 00000-40088a

5.8 .8* 5 488, U - 00000a - 5�* 84 .86 088.000a S 90 0888088.- -4 *..8 00e)�0

Sa� *o-�u-

8 �

a84 8O 8

-4 8S 8A48 8-4 I 0a a

8 0.8 20KWC �* 000I �o �8-.H�S.

C 9..o i-i...� ha

C a-* 18 a-na-- 34.

�.8ha P888884 088.40fl

B22 Appen�x B Target Metnc Statistics

_0-- - - - - - - -

Vkv EV O2i 2 OV

Z o Q04$ 0 0 0 0 w 0a. 4 0 wa 0000 000

.40r

:V Qm

a .64

a! C!3 .v!09C!ce ca9 C! C!Ctc!,,0a..a

C ' aa ~ -- 00 00 2 0 0 W.00

r4 a M~ 00 0 a a DN 40n a o 0 nOS0 44 . 0, I-CO M0 ,-rn0r 0.aNOO"aWo aOCQ .N S4SU Oemm.W W4C4~ N.4 0N mN.I4 o0 o

.0 0 .3. N. fl . - Op oo 4D Ch In

.a! at at00 0 c! .4 N O O O N O C S' . . . . . . . . . . . .o . . . . . .

4D1

a. 4

4 1 11 Flag aOO OO O O CO C O OO O O OO O C

SO~~~ ~~~~~ u) MM 0 0 0 0 N OO O O O O O OQ O O O C C O C

a. ~ ~ ~ ~ u to "NN00 0 0N4 ~ O O O oO o C ~ O OO C ~ O egol. 6011

.3~ ~~ ~~~~~~~~ 0 Ia. a: sK cc cc0.4 llf ~ l0 0 r0 0 4S l..l V1 .

V42 0 1 . 2$3~ * r r00fn-N4wC-Sa.e-aa.lv,0- H H22.w..o~~

Oped B Target Metric Statisbcs B234C0 00.l~0 0nV .V

Orr4

0 0 ,0! 00 00 000a0 0 00 0 Cý, 0 0 0,c 0 Q0! 0 0! 0! ! 0 0 c0 0 0 C 0 0 0!

-. 000 CC O C0 0 0

n en 0..

a IV! t 0! .~ . . .0 0C, .0c! Q9c . .C C OC CCCC Cý 0! CC

0000000000 00 0 00000000

n o t~ - nt .wC? V) wC , C 0, C C

rC 0. O CIr 00 Q. CIS E 0E .- i- ". .E.-1

C - ~ $tCICI*' 00o noCICI 40

B2

2C~~ppnd 00 0 00 0 0 000ge 00m 000000 0000 0

* 00

Zo

o * 000040

00 0

* a 00.N

a.

400-. .,s

£- a

o825aptd 8 are Mem 00000oc

-- - - - - -- - -

0

*~. . .. .~ .00 V 0 4 C . . .~ . r. . . .*4 4) 10W W

.WO.

I ~ ~ ~ --- - -0CrP -.04 00 0 000 0 4

v .0 aA ; !! 2 243 w o.4 I0 041100 0 4c0N oo000P.00q ý . a

I ~~ OW a411 r 0400 0P. 10r. 11P00

* C ro..n O r.r-~on o .. d.. .......

*~~~ ~~ ~~~~~~ 0 0 0 0 0 0 0 0 0 0 0 0 00 0 0 0 00000a0

so am 0 .0 o 0 0 0 r02 0 0 0 0 0 0 0 00 0 0 0 0 0 0

43~~~ ~~~~~~~ C 4O0 0 0 0 1 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

PC -3 ..3 ~ ~ ~ ~ ~ C me 14. .4.4 cc 4O .0044 V~fE00 4w 0. Cfr4 1 1 4

53 0 0 0 1 0 f - - - 0 1 - I . -- a P V r 0 0 0 4 1 S

o~~~ ~~~~~~ *2 'M1..4.Pf4P01100 P..41 l 0 0 4 1 0 V V 1

B26 Ape-k .t.4..C P fffflltn8fg4I~ Target Metnc Statsti

I~~~~ Uo 0000.00 0 0 0 4'0 0 0 00000000

*0 w

0 0 0 00 0 00 000 0 000=-Q 000 0 .00 0 00 0 00 0 0

430

En.. ~ ~ ~ ~ I aoo eo o e 000 00 000 00 000 00~0U 000 000 000 0002 0-- 0a0 00 0 0 00* ~ ~ ~ ~ ~ ~ ~ " Q o~ M40 00 0 0 00 0 0 0 0 0 00 0 0 0

I~~~~~C - 903.I. 0s.e 04~ In cooWWVt

ID1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

HIM 14lo- .O

*~~~C 0- ,w O ,O O 00 0 0 00 0

.4ed are e tasc 2

1 Cc 0000

ca N * 4 eq IV

00000

00000

C41 00a0

I E a 0 *Cý* - nfl.S V

CC

0d f

828 ApwCb

00age000eSttstc

-C'

.* . .

0 0 0 000000000t000000n

* * c 00 Q C,

* *E .04 0 m~1 . 0 -. ~ . n

to 0 0 0 N 4 0 0 0 0 0 0 0

c; C; 000000 000; C0 00 0 a 0

Zo -0 dI0 %O4 0 4 N 0 0 0

14 1 N N . N NN . N

U -Ca 0001 24N00000000000000000000000

.~ a

* 00Ne f fm i R I -a 'Jig 'j =02."192 9- 2. 41 1C 0 ICI 0.419 . 90 !

O~ ~ ~ ~ ~ ~ ~~i E N44.~i.001100f...W~lCOnC0,1

Ap w ) Ta rge M e ri St t tc B 29. 4 N f f nm ii f f f f ff 4 4 S l e O t ) N

C! C ! C CC ! C !1 ýQ C ! 1 !1 ! 1

104

.340 m 1 4. NNN4 www

aa IeD 00000 0000000 a 0 0 In0 0 00 0 0 00 0 0 0 w 00 f0t

03

eoo82o~ 000 00 0000000000000

I e O OO O~ Ego eH".0 0 00 0 0

* ~o i,. NNO C.O N4 SN nr NCO. N44040C2

.00 m1 0 4 0 0 , 4 l 4 ~ 0 0 0 4 0 . 0

No

C~~~Apod 8 Tage Meri 00 *0 00 0 00 00 0 00 0 00 00

C!

104

-C

0 1 CC 0

I 00,:00 4

w q 0000

C St

o041 " 0 0

I 40 00 0

o * ~ 0000* S vO.0 ).4 100.4a

Appendx B4 TreMercSaitcs B314

Appendix CRelevant MSFD Test Site Data andAnalysis Results

Contained within this appendix is all of the information used to study the radar returns from each ofthe 24 MSFD test sites for which good data were available. Arranged in numerical order by test num-ber, each packet of information contains a test summary sheet, Charge Coupled Device B&W obliquephotographs (if available), a B&W down-looking aerial photo and a 10-ft contour map, both two-dimensional and three dimensional (3-D) linear scale representations of total measured power, slantrange-azimuth representations of measured data and filtered data from the target feature metric exer-cises, range-azimuth representations of backscatter modeling results, and another site contour map withvegetation overlays. The approximate locations of targets for each test are shown as circles on thebackscatter prediction results.

Appendix C Relevant MSFD Test Site Data and Analysis Results C1

4 GO

< al

1-22 -oe -: EE[S

(-) LU

4dQ

0 v

CY iq -v $ I Il0 z

z 0 AJD )

LOU

L(I'

100U1-

C2 ApenlA C RelvantMSF Tet Sie Dta ad Aalyss RsLl

Overhead photo

Terai cotor (1Vinevl

Appendix~ C eeatMF es ieDttn Aiii ~ut

H-

Measured data, linear scale

NEASURE OF POWEFms1l5502_bara lat II2**2 + LR.*2

3-D plot of measured data, linear scale

CS Appendix C Relevant MSFD Test Site Data and Analysis Results

POWER

V.@ -le -. 0 -45 *0 .9 1**1r f

SIGNAL/CLUTTER

38- 0 K TIP U II4RUN

S Ut LflTf P hlIR4OWi

Figure C1. (Sheet 4 of 6)


VARIANCE

-.00 -t -to 16 -946V -90 -0 -e0 It ? 0 0

MI TR! f VA4I I

- - ------

- -mow

- 1 a -0 -ý4-- -o0 w 06 6 s

Figure C I (Sheet 5 of 6)

Appendix C Relevant MSFO Test Site Data and Analysis Results

Measured data, log scale

9K

Backscatter predictions, log scale

14 01-Ifs A/

CJ

Terai cotor an egttinovr

Fiur C1. (See 6of6

Appendix ~ ~ ~ ~ ~ ~ ~ ~ ,, C eeatS etSt at n nlssRslsC

0<

U) J~4V. Hz

LI,

C0

0/

IýN,

'-I-

LC) D

C8 Appendix C Relevant MSFD Test Site Data and Analysis Results

CCD photo



00 *

Overhead photo

IVV0N

Terrain contours (10-ft interval)


CIO Appendix C Relevant MSFU test Site Oata and Analysis Results





POWER

- --

SIGNAL/CLUTTER404

io- H M TjTP jU 14)(IIW

H (tI En TI kT U I INIU'.,



- m

VARIANCE

|.%=~~ n. I MI 1 I' R1116I

- - -.--1-. -

POLARIZATION RATIO

I9- - -4 9 -9 v9 I v"I *l 99

#1% TV:I VAI 14



'N 117) P 0F F- F-': . * F--



SAP

Terrain contours and vegetation overlay


014 Appendix C Relevant MSFD Test Site Data and Analysis Results

LCV)

< r)

C Li -

000

to m V) to r

F-- 4

Li wLi

7I

U :13COE LI >30 =) E u w w

Ce > l,0U.Li ix <

W> -W 6r4-j -j <0L'i MC -jx x

I.- wx (1w L(U

ICON6

-0 -W -O

OhL

U,

___________________________________ ________ O

Appenclix C Relevant MSFD Test Site Data and Analysis Results05

Overhead photo

AiN

"Trri cotor (10-' inteval

Figure~~N C3 (het2of6

C16~~~~~~~~~~~~~~ ApedxCRlvn '"Ts St aaadAayi eut





POWER

A bpUv1d1

am I

fit TV' I I YVA I a

SIGNAL/CLUTTER

•(,4- ,"14 II TII. IJI~41 )Il

to-

S.. .. ... .. I iY" VA •1I4



VARIANCE

r-7n

9~~~~~P Itli -,'ii iýi w* 14 ai A,*i

im TIý is vci ,


App~ndix C Re~evant MSFD Test Site Oata and Analysis Results C19


! • -• -x " .' , " "

0I





0<<0 Y-i

cL J0ýN

LO 0!10'

E"10

0 CS il

0 OC~s

Ll.AE --- q 0 --

UL

000

____ ___ ____ ___ ___ ____ ___ ___ ____ __ ___ ____ ___ _ L

ApenixC eevntMSD et it Dt ad nayssReuls 2

CCD photo


C22Appendix C Relevant MSFD Test Site Data and Analysis Results

Overhead photo


Figure 04. (Sheet 3 o1 7)

Appendix C Relevant MSFD Test Site Data and Analysis Results 023





POWER

40-

30- P 0 14 T1R 1i I 4Jou

a CLUMT[ &a94)M4S

-E 1 Ay *I0 -1As -90 -So -1 9 so 'f too isi

ME RI C VAtL L

Figure 04, (Sheet 5 of 7)

APpendix C Relevant MSFD Test Site Data and Analysis ResultsC2

VARIANCE

LA ILfll

-106 -to -. 9 -19 -HO a 9 of 9 I 9 q 9 to9 11

Pfih'IV VA LK

Figure C4-Set f7

C26 =-peni C=~ ReeatM es ieDt n Aayi eut


0I


* r

Vw .



Appendix C Relevant MSFD Test Site Data and Analysis ResultsC2

I- .- 1 _0 00Q0 < CY MoZZAA

V . 0z Cd

LUJ

C28C

-II zL~U zm jmi LjjU)

Lii

03

L))

LU C

0--

A- E


CCD photo



Overhead photo

, <• / -,,

f







Appendix C Relevant MSFD Test Site Data and Analysis Results G31

S POWER

-- A . -

SIGNAL/CLUTTER

4 0 1 0 4 9 - 0



-MI

VARIANCE

pi 4 1 A

l '1 '

Ut , -9? -It' *.* -~9 6*AS . It.

POLARIZAION RATI

to it -- -W

Fiur U.-- (Set6-f7

Ap_ _ch C Reevn M-Ts ieDt ndAayi eut 3

ZmI



C344Apeni CeeatMF etSt aaadAayi eut

C'Y)0<

IC.0 E88 Li

;!Vf ; -H 08 L

V) V03

I.S.

w-1--D c0LOw

co)

.~, <

Erl Li

F-x

CO Lo

C 0)

____ ___ ___ ____ ___ ___ ____ ___ ___ ____ ___ ___ ____ ___ LA

Appendix~~~~~~~~~~~~0 C eeatMF es ieDt nayi eut 3

CCD photo


C36 Appendix C Relevant MSFO Test Site Data and Analysis Results

4'JV

w' '

o~

ov4erhead P~hOIO

Te r i c-lu '

in e v i

F~igure Cro. (Sheei 3 01 7)

C37

Appfl~~0 Re8V~'~MSFD Test SIta Data and MOPS Rs ult


11*




POWER

P*Twlf Val &

SIGNAL/CLUTTER

404



25-

VARIANCE

I0-

5-

MORIC YALL&

- -m - - ----

POLARIZATION RATIO

M* CLUTTTER W'I$OWS


C40 Appendix C Relevant MSFC) Test Site Data and Analysis Results


--- N---®


d -- ---- --

e e

*~#~ *1f~C41

Appenclix ~ ~ : C~ ReeatM etSt aaadAayi eut

ce)

CIO ce w s "a ýo mV

U) Lj 4088

ztIz LXJ -m4 a

D~L~j~mZ&.

-4-j 0 w'>

Cx) NJW

L.Lj LLJ0000 a

W-- 1 xCjjacc oj 2 z

oL x a) LJLLiC

C-) 404LO ~ I I*

,~ (D

-c/uo

co.

E

_o

U)

r~r


CCD photo


Appen~dix C Relevant MSFD Test Site Data and Analysis Resutts C43

Overhead photo





4,f


Figure C7 (Sheet 4 of 7)


POWER

40-

- -- -- - IýNC

M -P I ~ I-A1

Figure C7T (Sheet 5 of 7)

C46Appendix C Relevant MSFO Test Site Data and Analysis Results

VARIANCE

15- P. -n* I, k* WS* IS -. DOV*

mv-f -- 9 -

- -ý

POLARIZATION RATIO

15

OWTPIC VewoF


Appefclix C Relevant MSFD Test Site Data and Analysis Results C47


(AMkBackscatter predictions, log scale

IVJ

"Trri cotor an veeain vra

FiueC .4 (Set7f7

C48 ~ ApedxC Rlvn SDTs St aaadAa i etS

H LAJ

aI 989 Li

LiJZ z

U Li

F-

0LO

4: J10-0<

0

CL,

00

U

LL

Appendix C Relevant MSFD Test Site Data and Analysts Results 4

U

Overhead photo








POWER

oiwirtIC VAt JF

SIGNAL/CLUTTER40-

fe-0 f!VTFW WHUOO1

-4 C1 JTTF k IIINE~OW

10--



VARIANCE

- _

Ih- t _ ETR ))

MERI _ _ -

POLARIZATION RATIO

19 4 TF Vk I" i




FiuriC. (Shee 6 f6C54, Apeni C eeatIIF etSt aaadA ayi eut

ddco p -ca I i zi

01 - .Lil los -

0018o

C\2 co

C)u

-orLU

LoI

Apeoix C Rlvn SDTsieDtaadAayt eut 5

Teraicntors(1-ftineral

41*

t110

Overhead photo


C56 Appendix C Relevant MSFD Test Site Data and Analyis Results


=4

I



Appendx C Relevant MSFD Test Site Data and Analysis Results C57

POWER

-- - -- - -a -

SIGNAL/CLUTTER

RI T•I,'++I• ..

40 s - 5 0 4 * 9 9 9 3



VARIANCE

a. Ra e. -10 0 -ma4 0 110 -. s -a. 2* * lot

METRIC VALUE

-n -

mETTIc VALUE





Fiur C9 (See 6 f6

C6Apeni C/eeatMF etSt aaadAayi eut

LLJ

H-oLU)

LLJ)

-4- > >LU Ui -; *21

LL) 04LU

0 L 0 z 1

0C) N

Li

z

ID OD

1010 .010

LU

(0 C,,

co >

________________ E


Overhead photo

\ nX

NN





II




POWER

15-

**-6 -IS S -*S St S IS i?,P it, at

SIGNAL/CLUTTER40--

'H--


C64Appendix C' Relevant MSFD Test Site Data and Analysis Results

VARIANCE

I10 I0 1 TI P W I HX)43

25

POLARIZATION RATIO208

15- W iP~lP bIRf)I

Figure C1 0. (Sheet 5 of 6)

Appendix C Relevant MSFD Test Site Data and Analysis Resut~t C65





066 Appendix C Relevant MVSFU Test Site Data and Analysis Results

C)3

LUJ

0-CC) jii94dLo CE~cc-W

SEEStic

0 '-j-JilW

Q)~~

LA4,

CC

< C3

r-3

LLJ,co*____ ___ ____ ___ ____ ___ ____ ___ ____ ___ ____ ___ ____ _ o

Appn o Z ReeatMF etSt aa rdAayi*eut 6

~e?4:4 ,PO

~A

Ovrhadpht

Teri cotor (1-tinevl

Figure ~E C11.See f6

C6 pedi eean S DTstSt aa n nlsi eut


-I


Figure C11, (Sheet 3 of 6)

Appendix C Relevant MSFD Test Site Data and Analysis Results069

POWER

-- V -ib wi9 .09 -09 00 *W 0 @ so , Ib m 0

METRlIC V&I.I

- 7-

• •.- - ~ ---.. !_._- _-._.-._ -_ -_

SIGNAL/CLUTTER

il--

-. 'S*$ -,.'1 f l.0 iSO -.e9 -,O 0 , IVq , II, IO

Figure Cl!, (Sheet 4 of 6)


VARIANCE

POLARIZATION RATIO

10-

-S 9 9 -9?9 5 a9 9 *9 19 9 s.

Figure C1 1 (Sheet 5 of 6)

Appenclix C Relevant MSFD Test Site Data and Analysis Results C71




Figure C1 1 (Sheet 6 of 6)


CV 4

0 H <

4 00 Li

zz >.> M

H-

Si -532j

4m 4

LC)i04V

U

CC),

0

0) -

00Cu

N

00 0<E Go___________;j____ _____________ LO

Appedx Relvan MSF Tet Sie Dta ad Aalyss Rsult C7

Overhead photo


Figure C12. "ýheet 2 of 6)



LL




POWER

'iim,

50-

Figure C12. (Sheet 4 ol6)


VARIANCE

16 e PETR WIUM~JI

pKvrRIC VALLE

POLARIZATION RATIO

Figure 0 12. (Sheet 5 of 6)



I "'j



Figure C12. (Sheet 6 of6)


0

IV

CIO

E-4-J

Z U 0

V) '4

<

E-o z 0 z

__ ajLAJ)

LO M: tr v 1

coCco ka.9- ,).

0ý0

S Cl3

000

V'T* I ILL

Appendiii ~~~~L C- ReeatMF et ieDt nfnayi eut 7

Overhead photo

II

Figure C13. (Sheet 2 o1 6)

O80 Appendix C Relevant MSFD Test Site Data and Analysis Results


A

- V




"•-!•_ POWER

M lvit V¶ItIJ

25-

SIGNAL/CLUTTER

9 m 4rwI u imVUw

)L 'fI frTrWi WI lSl0-•

MTWIIt VAI;q



1iI-VARIANCE

10014T PUI O

Mr TR IC Vat LI

POLARIZATION RATIO

-I~ -91 -61 o. f? 9 q 9 * 9 Ile is*

rd TV I VA1. Lf


Appendix C Relevant MSFD Test Site Data anid Analysis Results C83





C84 Appendix C Relevant MSFD Test Site Data and Analysis Resutts

0~Q I a00 1W

ccc

z - COSJ I

ZZ

z~~ C)> C4-) HR

V)~~~c ---- Z IX0ý'-w

> LjCliiiCL C L go -C-jX-,

Q) >- Cu0 LLJn V0

j :IZ M Hd ZE!

LLJ M: 00 L- > -

L) ~LULJ

LLIL

C8

AppeiidNx C eeatMF es ie3t n Aayi eut

Overhead photo

~~I /1 / I', J ; ,

i • • i~ 1 / /,.

"I,,, ,,' ,

'. 1





AMI



Appenidix C Relevant MSFD Test Site Data and Analysis ResultsC8

POWER

- - - _ _ _ .-.. mmilm

SIGNAL/CLUTTER40-

PR-i - I I •! " H l'

11-4



VARIANCE

MMTIC VALUE

POLARIZATION RATIO


Appendix C Relvant MSFD T'est Site Data and Analysis ResultsC8



/ ..L



C90 Appendix C Relevant MSFO Test Site Data and Analysts Results

zz

0 V0 a

I <Z82

L IiLiia

x UT,

LLO

co'

LI3IzC

CC) z


IItt

CCD photo


C92Appendix C Relevant MSFO Test Site Data and Anatysis Results

Overhead photo



Appendix C. Relevant MSFD Test Site Data and Analysis Results C93


F!




POWER

M4TVir VAIII

- --

SIGNAL/CLUTTER

A UIIlrrT[R~hI~U~

"MtTRIC VAL I

Figure 015. (Sheet 5of 7)

Appendix C Relevant MSFD Test Site Data and Analysis Results 9

.59 . ... .

VARIANCE

40-

ME~TRIC VALUE

POLARIZATION RATIO




"oi~ /:

Backcattr prdictonslog cal

<' *

Terrin cntous an vegtatin ovrla

Figure~~ CI5/See f7

Appndi CRelvat MFDTes Ste ataan Anlyis esltsC9

mai W I s 111 1IX

C/,3

~I > "g

a'gzt.

ILI-

is

CI))

cco

CD

0

LL

C98 Appentdix C Relevant MSFD Test Site Data and Anatysis Results

COD photo

Figure 016. (Sheet 2o017)

APPsndfx C Reievant MSFD Test Site Data and Analysis Results 9

94

Overhead photo

w.--• .-


Figure C16. (Sheet 3 ot 7)

C1 00 Appendix C Relevant MSFD Test Site Data and Analysis Results


TIT





A

3-D plt of measured data, linear scale

Figure C16. (Sheel 5 ol 7)

C102 Appendix C Relevant MSFD Test -;tte Oata and Analysis Results

i-5

POWER

15-

S-

is b VU nip Ile

50- SiGNAL/CL UTTER

40-


Appendix C Relevant MSFD Test Site Data and Analysi, Results C103


Backscatter predictions, log scale//P /¼j

Teri cotusan/eeaio1vra

Fiur C16. (Set7f7

C- 04/ApedxC Rlvn/SDTs St aaadAayi eut

Q0J Eelnigi

w-

Cl) ::f £IE LI )m

< it5: In L .92'- $ aL4E--I

4L

'-4 g111

coV'I Egli

0-0

C\2 _ _ _ _ _ __

LL

APPendix C Relevant MSFD Test Site Data and Analysts Results 0105

cco photo


C1 06 Append.x' C Relevant MSFrJ Test Siý- ;ata and Ar.alysts Resili

Overhead photo







C1 08 Appendix C Relevant MSFD Test Site Data and Analysis Resutts

POWER

-

SIGNAL/CLUTTER

, - TII T Ji I

U UJVTLI R i 1 4XVj¶

"14-

,", -. 'S • -JS -11lAS •ii ,, I S. SI ,S 15,v

M• TRI• c VIZ (It


Appendix C Relevant MSFD Test Site Data and Analysis Results C

VARIANCE40-

614-

-- IN -a

Ps-

POLARIZATION RATIO


Cl 10Appendix C Relevant MSFD lest Site Data and Analysis Results



K -.. , 'r/p "/7/,,.-. i •-- • ./ • ." ,. --- ;

"1/._.. '•.,t ." ,~~~~- -'I -- .U b •/ , , f -. * /"" ""1/

--7 -... --..- 7. ... . .-Z - ,"



Append -C Relevant MSFD Test Site Datl and Anaiysis Resuits C111

0ý lip

~14:jig ellCcavil

0-7

cv

a a)

cl 121Apeni CReean MF TstSteDaaan Aay-r-Rsut

CCOD photo

Figure Cie. (Sheet 201f 7)

APPendx C Relevant MSFO Test Site Data and Analysis Results C1 13

WAIFI

Terrain contours (1 0-ft tnterval)


C1 14Appendix (C Relevant MSFD Test Site Data and Analy'sis Results




Appendix C Relevant MSFD Test Site Oata and Analysis Resufts C115

POWER

-L-

SIGNAL/CLUTTER

w


CA 16 Appendix C Relevant MSFD T"est Site Data and Analysis Results

S VARIANCE

30- to 9 CTER WINDOW

a.~- -100 0 4 9 9 * 4 4.

_ --

POLARIZATION RATIO


01 17Appendix 0 Relevant MSFO Test Site Data and Analysis Results



f OP

Vol

-~ V-i-.



C1 18Appendix C Relevant MSFD Test Site Data and Analysis Results

C\ UL RU

V--4'- 11 NIN

C.0

Sal 119 0,"S LN

it' Avg-

zci

LI.-

C31Apped, Relvan MSF Tet Sie Dta ad Aalyss RsC3

CCD photo


C1 20Appendix C Relevant MSFD Test Site Data and Analysis Results

Overhead photo








POWER

t0-

--

Irl I• rt' p/• u IFv

------ ".-_tim•'n- - _.

_ -,.- -.-. : • u,,, =: ._

50-

SIGNAL/CLUTTER

MTR!I Vfttg


Appendix C Relevant MSFD Test Site Data and Analysis Pesults C1 23

VARIANCE40-

30-

__--

-1

POLARIZATION RATIO20-

15 ~2 0 METER W~INDOW6

METRIC VALLE


0124Appendix C Relevant MSFD Test Site Data and Analysis Results


/A

FiueC9 Set7 of 7)~

Appndx ReevntMS TstSie Dtaan Aalysscte pReduction Cog1c25

Eat liass aco gig

$so fll X89 L:~~~& S uiggga *1

z

00

I-i

cv -

C) X)

C\22

0ý0

LI (

mxm mo

- i.-. tvu;~ /co0

F- C-


Figure C20- (Sheet 2 of 7)

APPendix c Relevant MSFO "1"t Sit Data and Analysis Result, 12

Overhead photo

,'- ,,-

- -

9 - -



Cl 28 Appenc�x C Relevant MSFD Test Site Data and Analysts Resulta


IN

1, I

3-13 plot of measured data, linear scale

C1 29


Appendix C Relevant MSFD Test Site Data and Analysis Results

POWER

min1I VAIIvIn

SIGNAL/CLUTTER40-

"4H

ioi 01- n T1 Y• I WO

'54.,eI' IC • --i --.. _M-s-

10--

110 ClA ,0



VARIANCE

P11 Wrt Vali-

POLARIZATION RATIO

11 l 41 )11


APPendix C Relevant MSFD Test Site Data and Analysis Results 0131





Cl 32Appendix C Relevant MSFD rest Site Data and Analysts Results

(0 ms 1I1 s "

Lag 10 fil

ZI

a w)

00

0 LI-

.1 1 1 --

Aped C RlvntMF es S ut Daaad nlsi euls0 3

CCD photo



,.to

Overhead photo

/



Appendix C Relevant MSFD Test Site Data and Analyis Results C135





POWER

PlMTW VnIIF

- -)

-5-

- t1•r - - -

SIGNAL/CLUTTER

.1 IjUiTr|tI WIP|h}w•

Pt I t•h VrI V


Appendx C Relevant MSFD Test Site Data and Analysis Results C137

VARIANCE

ý0 0 TFW i"V

-100 -#-fW 104 -1.4 -10f -t 0 -,'4 0' IV.1

-VV& ----

-77

POLARIZATION RATIO

*4 4* 4 Ti4 W 1i I ill 11W

Figure 021 (Sheet 6 of 7)

C 138 Appendix C Relevant MSFD Test Site Data and Analysis Results



P&Z

0'- -4-

-/4/

Ai -~



Appendix C Relevant MSFD Test Site Data and Analysis Results C1 39

Cr')US0w1114 d V! 9-11: o O

z 0

0 - 1

LC) I - UZ

00

CI))

00

C3'

COu

CL'

U-

Cl 40Appendix C Relevant MISFD Test Site Data and Analysis Results

CCD photo



Overhead photo

Teri cotor (104 inevl

Figur C22 (Shet 3of 7

C142IAppendix ~ C-eeatMF etSt aaadAayi eut

VARIANCE

Se ~19 e PETER WiItOO

1-~

'-

-L-

-

VIETRIC VALU.E


Appendix C Roeis 'MSFD Test Site Data and Analysis Results C1 43

POWER

Ze-

rrn

50-

SIGNAL/CLUTTER

40I

Figure C22, (Sheet 5 of 7)



00

N/




C146Appendix C Relevant MSFD Test Site Data and Anatyis Results

5 aI

CI) u O .log ;:LI ISIS L

I" 9 6

c z1 1> ass&&

U2 ~-Li

I-..

ot S00

cL)

C ~C9

L F-I

LO f 0

cui

ApPedidx C Relevant MSFD Test Site Data and Analysis Results C1 47

N41',~' ',to

''At

COD photo


01 48 Appendix C Relevant MSFD Test Site Data and Analysis Results

Overhead photo

Jw

Terrin cntous (0-ftinteval

Figur C23 (Shet 3of 7

Appendix ~ ~ C eeatMF etSt at n nlssRslsC4





POWER

40--

L ------- -- -

Fiur C23 (Set5f7

Appendi C~ Reevn MS- TetSt-aaadAnlssRslsC

VARIANCE

.me~ ~~~~ It6 7F'A SW k4$ 11,1 W4 ' 0 t

1-~

POLARIZATION RATIO

F~ WI Vwj I


C1 52 Appendix C Relevant MSFO Test Site Data and Analysis Results






030

(~.0 d

zz

5! IT

am~g IL)Liij

LOiN-

C ~ j' ~ .

UU

LID,

0

LOI

cli


i *=

CCD photo



.14

Overhead photo

Terrain contours (¶0-ft interval)


C156Appendix C Relevant MSFO Test Site Data anid Analysis Results


3-D pk~t of measured data, linear scale

Figure C24. (Sheet 4 of 7)C157

Appendix C Relevant MSFD Test Sit, O)ata and Analysis Results

POWER

PIETRIC VALLE

SIGNAL/CLUTTER

15 28 U.IPt LDIN(OW

METRkV VALý*


C1 58Appendix C Relevant MSFD Test S~te Data and Anlalysis Results

VARIANCE

I.. -. I I . I .

POLARIZATION RATIO

~4. S 4V - 111! ,vt k 4 4


C1 59Appendix C Relevant MSFD Test Site Data anid Analysis Results





C1 60Appendix C Relevant MVSFD Test Site Data and Analysis Results

Form ApprovedREPORT DOCUMENTATION PAGE OMB No. 0704-0188

Public reporting burden for this collection of inl0,rmatin is estimated to average I hour Per (e~prise includingCt the trime for fte-evin inst truction searchingQ irxisil data source,gathe,,ng and ,al•tantn- the oata nee•e•. rid comnteting alnd reviewmrgq itfe (Otf•e"ion of itorma'ntrO tio n d comment, ,.oatocn t,) burden .sttvatC or any Other iS OC t ttt

colection of intormation. including sugqestons lor reducng itn, 0uraen. to Walhnriton Heaoquaert, Setrvces. Direc•oate lof frifon-alOn Operations and Repons. 121S )effititotnDavis H•ghway. Suite 1204. ArlItrngo. VA 22202.4)02 anti To the Office ot Manaqemnent and Budget. Paprtutork Reduction Projet (0704-0158) Washirngton DC 2050)

1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE I3 REPORT TYPE AND DATES COVERED

I June 1993 Report 3 of a series ______4. TITLE AND SUBTITLE S. FUNDING NUMBERS

Environmental Characterization for Target Acquisition; Report 3, NewConcepts for Evaluating Low-Grazing Angle Radar Measurements

6. AUTHOR(S)

John 0. CurtisBruce M. Sabol

7. PERFORMING ORGANIZATION NAME(S) AND AODRESS(ES) B. PERFORMING ORGANIZATION

U.S. Army Engineer Waterways Experiment Station REPORT NUMBER

Environmental Laboratory Technical Report3909 Halls Ferry Road, Vicksburg MS 39180-6199 EL-93 -9

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADORESS(ES) 10. SPONSORING/MONITORING

AGENCY REPORT NUMBERU.S. Army Aviation Applied Technology DirectorateAviation Systems Command, Fort Eustis, VA 23604-5577;U.S. Army Corps of EngineersWashington, DC 20314-1000

11. SUPPLEMENTARY NOTES

Available from National Technical Information Services, 5285 Port Royal Road, Springfield, VA 22161]

12a. DISTRIBUTION/AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Approved for public release; distribution is unlimited.

13. ABSTRACT (Maximum 200 word)

In an attempt to develop a scene characterization methodology for long-range (2 to 7 kin), low-grazing angle(<10 deg or .17 radians) active millimeter wave radar systems, a twofold approach is taken. First, slant rangeversus azimuth maps of radar backscatter measurements are filtered to collect statistics on the occurrence oftarget-like features and to relate those signatures to natural terrain conditions. Second, a prediction of radarreturn from the terrain is modeled by a point light source at the radar location and the assumption of Lambertianscattering from the terrain facets. Model predictions are combined with overlays of vegetation at each test siteand compared qualitatively with the measured data to assess the impact of terrain conditions on backscatterresponse. These approaches are applied to a set of test site K.-band radar measurements made at Fort HunterLiggett during the winter of 1988.

14. SUBJECT TERMS 15. NUMBER OF PAGES247

Lambertian scattering Scene metrics 16. PRICE CODE

Radar backscatter

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACTOF REPORT OF THIS PAGE OF ABSTRACT

UNCLASSIFIED UNCLASSIFIED I ..NSN 7540-01-280-SS00 Standard Form 298 (Rev 2-89)

Prescribcd by AN£1 SIt Z39-1-19S.107

Technical Report EL-93-9 US Arm'y Corps AD-A269 Iii~lllil ...

Documents