F AD-A095 748 ANALYSIS AND TECHNOLOGY INC NORTH STONINGTON CTO FIG 6/7EVALUATION OF NA TIONAL SAR MANUAL. PROBABILITY OF DETECTION CUR-ETCIU)SEP aG0 N C EDWARDS. T J MAZOUR, R A BEMONTIS DOT-CO-8I-79-2038

UNCLASSIFIED USCG-D Ni1A NL

IL

00

et

9- "t

r 7'Technicsi R ,e*" Z cum nt etion Paq*CG- " , -'80

I91 *CJf l t g9E? JOeelftla

',-g e ,* Oe+O- 3 te. .*ee,, e,

Evaluation of National SAR Manual. / / Septefner .S 198 -,/_Probability of Detection Curves • '.o

-: ... .. "WN. C./.Ea:warcs, Jr.' ,T. . azour- 7 0

R. A./BemontkS. ClOsmer

U.S.C.G. R&D Center Analysis & Technology, Inc. IAvery Point P.O. Box 220 ... .Groton, CT 06340 North Stonington, CT 06359r 6/'

I nterimXepErt.Department of Transportation'- - 4 Janurary 1978-September 198p,U.S. Coast Guard .Office of Research and Development ' " "Washinoton. D.C. 20590

This report is the fourth in a series which document the Probability ofDetec ion Task of the SAR project at the U.S.C.G R&D Center. USCG R&DC 3/80

VSince September 1978 three controlled visual detection experiments providing966 target detection opportunities in 322 searches have been conducted by theU.S.C.G. Research and Development Center. Theseexperiments involved 82 and 95-footcutters, 41 and 44-foot boats, helicopters, and fixed wing aircraft searching for16-foot boat and life raft targets.

This report compares the detection performance of these search and rescue units(SRUs) with the probability of detection (POD) curves of the National Search andRescue Manual, and recommends revised predictions based upon these results.

Experiment results indicate that actual SRU detection performance falls below

that of the present SAR Manual POD curves (based upon the inverse cube law of detec-tion) and above that of the uniform random search curve.

Recommendations are also provided for methods to predict POD directly from*,probability of detection versus lateral range curves for use with the Computer-Assisted Search Planning (CASP) model.

Search and Rescue, Visual Search, Document is available to the U.S. publicSurface Target Detection, Probability through the National Technical Informationof Detection Service, Springfield, Virginia 22161

Unclassified Unclassified

n =Plez7d n i2 :.s suthcr-ad

j / 77(

VII

31!

1,

-.-. --

I~*.E - 1~-ow

YI Ii

TABLE OF CONTENTS

Section Page

* Executive Surmmary .. .. ... ..... .... ..... ........

Chapter 1 Introduction

1.1 Scope .. .. ... ..... .... ..... ........ 1-11.2 Background .. ... .... ..... .... ....... 1-11.2.1 Sweep Width. .. .. .... ..... ..... ..... 1-21.2.2 Coverage Factor .. .. .... .... ..... ..... 1-41.2.3 Inverse Cube Law of Detection. .. .. ..... ..... 1-51.2.4 Uniform Random Search .. .. ... ..... ....... 1-7

Chapter 2 Conduct of Experiments

2.1 Introduction .. .. ..... ..... .... ...... 2-12.2 General Description of Experiments. .. .... ...... 2-12.2.1 Time Frame .. .. ..... ..... .... ...... 2-12.2.2 Location .. .. ..... ..... .... ....... 2-12.2.3 Participants. .. .... .... ..... ....... 2-32.2.4 Navigation .. ... .... ..... .... ...... 2-52.2.5 Search Tracks .. .. ... ..... .... ....... 2-52.2.6 Targets .. .. ... ..... .... ..... ..... 2-92.3 Microwave Ranging System. .. .... .... ....... 2-92.4 Data Acquisition. .. .... .... ..... ...... 2-92.5 Description of Experiment Conditions. .. ... ...... 2-102.5.1 Summnary of Detection Opportunities. .. ... ...... 2-102.5.2 Range of Environmental Parameters .. .. ... ...... 2-11

Chapter 3 Analysis Approach

3.1 Introduction .. .. ..... ..... .... ...... 3-13.2 Organization of Raw Data. .. .... .... ....... 3-13.2.1 Determining Opportunities and Detections. .. ...... 3-13.2.2 Determining Nature of Target Distribution .. .. ..... 3-23.2.3 Sorting Data. .. .... .... ..... ....... 3-33.3 Empirical POD Versus Coverage Factor Curves .. .. ..... 3-33.4 Comparison of Experiment Conduct with Theoretical

Assumptions. .. .. ..... .... ..... ...... 3-43.4.1 Target Location .. .. ... ..... .... ...... 3-43.4.2 Search Methods. .. .... .... ..... ...... 3-63.4.3 Constant Sweep Width .. ... .... ..... ..... 3-6

4. Chapter 4 Results, Conclusions and Recommnendations

4.1 General . . . . . . . . . . . . . . . . . . . . . . . .4-1j4.2 Comparison of Results by SRU Type .. .. ... ....... 4-4

4.3 Target Distribution Effects on POD. .. .... ...... 4-44.4 Influence of Lateral Range Curve Shape on POD Model .4-7

4.5 Conclusions. .. .. ..... .... ..... ...... 4-84.6 Recommnendations .. .. ... ..... .... ....... 4-10

TABLE OF CONTENTS (Cont'd)

Section Page

References ...... ... ............................ ..... R-1

Appendix A Raw Oata ....... ....................... ..... A-IAppendix 3 Metric Conversion Factors ....... ............... B-i

A, q(iofl For

*N'Wl (;RA&I:,IC TA-jjaido'. ed [1T-,st i f 4.c t iorl-

Distribut ion/

Availability Codes

Avail and/orDist i Spcc tl

.A-

, ii

iti

p.

LIST OF FIGURES

Fiaure Page

I Comparison of Experimental Results to POD Model(All SRUs) .... ... .... ....................... 2

1-1 Definition of Lateral Range ................ 1-21-2 Relationship of Targets Sighted to Targets Not Sighted 1-31-3 Graphic and Pictorial Representation of Sweep Width . . . 1-41-4 Probability of Detection Versus Coverage Factor Curve . 1-51-5 Comparison of Inverse Cube Law and Uniform Random Search

Models ........ ......................... . . ... 1-8

2-1 Search Area and MRS Configuration ............ 2-22-2 Actual ard ntended LORAN A Search Tracks for

HU-16E (Fixed Wing Aircraft) .... .............. .... 2-6

3-1 Search Area with Normalized 5 X 5 Coordinate Grid .... 3-2

4-1 Comparison of Experimental Results to POD Models(All SRUs) .............................. 4-1

4-2 Effect of Deviations from Track on POD ......... 4-34-3 Comparison of Best-Fit POD Models for SRU Types ..... 4-54-4 Comparison of POD Models for 1978 and 1979 Experiments 4-74-5 Example of Peaked and Flat Lateral

Range Curves .... ...................... ... 4-84-6 Comparison of POD for Peaked and Flat Lateral

Range Curves ..... .. .. ...................... 4-9

LIST OF TABLES

Table Page

2-1 Search Unit Characteristics ...... ............... 2-3

2-2 Participating Units/Facilities ...... ............. 2-42-3 Summary of SRU Resources ..... .. ................ 2-112-4 Range of Experiment Environmental Conditions ...... ... 2-12

3-1 Comparison of Actual Operations with POD ModelAssumptions ....... ... ....................... 3-5

4-1 Goodness of Fit of Models Evaluated ............. ..... 4-24-2 POD Model Fits by SRU Type ....... ............... 4-44-3 Distribution of Targets .... ................. ..... 4-6

i iii

r~TT

EXECUTIVE SUMMARY

INTRODUCTION

Coast Guard search planners currently use the National Search and Rescue

,Manual (Reference 1) probability of detection (POD) curves to predict search

and rescue unit (SRU) detection performance. These curves are based upon workione during World War II by the U.S. Navy Operations Evaluation Group (OEG)

(Reference 3). Among the assumptions associated with this theoretical model

are uniform coverage of the search area and the instantaneous probability ofdetection being inversely proportional to the cube of the sighting range.

Visual detection experiments, conducted by the Coast Guard Research andDevelopment (R&D) Center during 1978 and 1979 primarily to develop improvedsweep width predictions, provided 966 life raft and 16-foot boat targets of

opportunity from 322 searches. This report compares the demonstrated search

ability of cutters, beats, helicopters, and fixed wing aircraft to the SARManual predictions in order to evaluate the need for alternative predictive

models.

RESULTS

Figure 1 shows plots of experimental detection results, the SAR Manual

inverse cube law model, a "lower bound" model (random search curve), andtwo alternative models. The empirical data is based upon actual detectionperformance and sweep width estimates reported in Reference 2. Regression

analysis indicated that the alternative models were much better fits to theexperimental data than either of the two theoretical curves.

For lower coverage factors' (0.8 or less), the data is well representedby the inverse cube law model, while for higher coverage factors (greater

than 0.8) the empirical data falls between the inverse cube law and the random

'Coverage factor Swewit WTakspacing S5

INVERSE CUBE LAW (91J) (5/5)

1.0T

0.9" (17114)

0.8-

z0.7- 210

0.6

U. (20/400 0.5-

0.4-

0. 0.3- ( )

0.2

- -POD - Tanh (0.943C)

0.1- POD =1 -e - 130C

(012) ( ) DETECTIONSIOPPORTUNITIES RATIO

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8

COVERAGE FACTOR (C)

Figure 1. Comparison of Experimental Results to POD Model (All SRUs)

search curve. The lower observed detection performance is attributed

primarily to navigational inaccuracies of search units and other operational

factors which lead to non-uniform coverage of the search areas.

CONCLUSIONS

a. The present SAR Manual POD versus coverage factor model overesti-

mates POD, particularly at higher coverage factors. Therefore,

there is a clear need to revise the SAR Manual model.

b. With the development of emoirically derived lateral range curves,

the standard method of determining PO should utilize these lateral

2

range curves by driving them through the search ar,_' and determining

the expected value of POD directly. The Computer-Assisted Search

Planning (GASP) System should be used to accomplish this and a manual

version of the GASP model should replace the present POO prediction

method in the National SAR Manual. It is necessary that any backup

manual POD model provide predictions consistent with the standard

computerized technique.

RECOMMENDATIONS

a. For use with the GASP model, it is recormmended that a method which

determines POD directly from lateral range curves, navigation error

distribution, and target distribution be implemented. POD predic-

tions generated from the GASP model can then be compiled in a form

suitable for a manual/calculator method as a backup to GASP.

b. Because navigation limitations are expected to influence POD, a

method for quantifying these effects should be developed and vali-

dated.

c. In the future, alternative search tactics should be evaluated. To

support this evaluation, further analysis of empirical POD results

should be conducted to compare predictions and empirical data for

targets between adjacent tracks to predictions and empirical data

for those targets near search area borders.

3

Chapter II NTRODUC TION

1.1 SCOPE

This report compares the results of visual detection experiments per-

formed by the Coast Guard Research and Development (R&D) Center during 1978

and 1979 with the cumulative probability of detection (POD) versus coverage

factor (C) curves found in the National Search and Rescue Manual (Reference 1).

Recommendations for future efforts in developing a computer-aided POD model

and a backup manual POD prediction method are made. The curves are presently

used by search and rescue (SAR) planners to predict a search and rescue unit's

(SRU's) POD for a search under specific environmental conditions and for a

given track spacing (S).

1.2 BACKGROUND

The empirical data (966 target opportunities in 322 searches) used for

this report were gathered during a series of three controlled experiments con-

ducted by the Coast Guard R&D Center during 1978 and 1979. The major goals of

these experiments were: (a) the generation of revised sweep width tables for

the SAR Manual and lateral range curves for the Computer-Assisted Search Plan-

ning (CASP) model based upon the results of carefully monitored experiments,'

(b) evaluation of the current POD model, and (c) proposal of changes to the

POD model, where appropriate. This report deals with "he latter two goals.

In order to make these assessments, it is first necessary to examine the con-

cepts of sweep width and coverage factor as they relate to the prediction of

POD.

'For a detailed discussion of experimental results as they relate to the up-

dating of sweep width tables, see Reference 2.

1.2.1 Sweep Midth

Effective SAR operations recuire effiuient use of limited SAR rescurces.

Optimal use of available SRUs benefits the Coast Guard by conserving human

and material resources while providing the best chance for saving life and

proerty.

At present, SAR planners use information concerning on-the-scene

environmental conditions, target characteristics, and SRU type to calculate a

quantity known as sweep width (W) for a given search.

Sweep width is a performance measure for search units. It is a single-

number summation of a more complex range detection probability relationship.

Mathematically,

Sweep Width (W) P 2(x) dx (1)

where

x = lateral range or closest point of aoproach of targets of ocoor-

tunity (see Figure 1-1) and

P(x) z prooability of detection at lateral range x.

A AA TARGET

HORIZONTALIRANGE

I SV

/I /

A

LATERAL RANGE

Figure 1-1. Definition of Lateral Range

S!-2

K4

Figure 1-2 shows a typical P(x) curve as a function of lateral range

(Reference 1), where x is the lateral range of detection opportunities.

1.0'

TARGETS NOT SIGHTED

P(x) 0.5

TARGETS SIGHTED

0 -SEARCHER TRACK

LATERAL RANGE Wx

I MAXIMUMI LATERAL RANGEOF DETECTION (RD) I

Figure 1-2. Relationship of Targets Sighted to Targets Not Sighted

In concept, sweep width is the numerical value obtained by reducing the

maximum lateral range of detection (RD) of any given sweep so that scatteredtargets which may be detected beyond the limits of W are equal in number tothose which may be missed within those limits. Figure 1-3 graphically pre-

sents this concept of sweep width. The number of targets missed inside the

sweep width distance is indicated by the shaded portion near the top middle of

the rectangle (area A) while the number of targets sighted beyond the sweep

width distance is indicated by the shaded portion at each end of the rectangle

(area B). Referring only to the shaded areas, when the number of targets

missed equals the number of targets sighted (area A =area B), sweep width is

defined. A detailed mathematical development and explanation of sweep width

A can be found in Reference 3.

Sweep width is dependent on a variety of environmental factors as

*well as on the SRU type and on the target characteristics. A detailed

discussion of the effects which environmental parameter: have on sweep

width for various SRU and target types can be found in Reference 2.

1-3

A :RAPMIC qESNTATIGN OF SWEEP NIOTH:rAR G&TS NOT 39T ICTIO

rp r Ia% :,g vor

S 00% i )

I A-AVE&P "MOTM

~4~-.----rAAGMr OdTtcm AA

#rrEcaN

aursla •W I WIIIIM

" '"...-...OETW"--- " '°'"" 1;0T"h

3. OCaIAU DESETA IN O VCE'VIT

Figure 1-3. Graphic and Pictorial Presentation of Sweep WAidth

1.2.2 Coveraoe Factor

Once sweeD width has been deternined, a track spacing is assigned to each

SRU so that a desired POD is predicted for the search by the POD versus cover-

age factor curves of Reference 1. The track spacing is calculated by entering

these curves (shown in Figure 1-4) with the desired POD, and obtaining the

corresponding coverage factor required. Coverage factor (C) is simply the

ratio of sweep width to track spacing:J w 2W4C (2)

From the acove relationship, track spacing can be determined for a given

sweep ,vidth so that the desired coverage factor is obtained.

1-4

ij

SEIARCH

FORT

0 3Q / SEARCH

IAM

S0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1It .0

COVERAGE FACTOR

Figure 1-4. Probability of Detection Versus Coverage Factor Curve

1It is apparent from the above discussion that effective search planning

relies heavily upon the accuracy of the present POD versus coverage factor

model. If this model cannot be validated empirically, then it should be

changed appropriately to improve its effectiveness as a search-planning tool.

1.2.3 Inverse Cube Law of Detection

The inverse cube law of detection states that instantaneous glimpse

, probability of detection (y) for a target is given by:

h (3)r 3

~where:

, k a a constant which depends on target's area and intrinsic contrast

"4 with background,

h - altitude of searcher, and

r - horizontal range to target (see Figure 1-1).

'1-5

|#U

7- -,<-: ... .

-' I I I I I III I ' I Iil0

It can be Shown (see Reference 3) that, for continuous looking, the cumulative

probability of detection (POD) as a function of time is given by:

POD 1-e 0 r()d

From the above expressions it can ultimately be shown that, for a parallel

sweep search, the cumulative probability of detection equation is:

POD 2 f e -zdz - erf(z)(5

where z wtW

This equation is derived in detail in Reference 3, and forms the basis of the

curves presented in Figure 1-4.

Inherent in the use of equation 5 are several assumptions about target

location within the search area, the manner in which the area is searched,

and the geometry between searcher and target. These assum1ptions include:

a. The target is located randomly within the search area and remains

in the area for the entire search.

b. The area is covered by equally-spaced, parallel search tracks.

c. Constant environmental conditions (i.e., constant sweep width for

any SRU or target type) prevail over the duration of the searcni.

d. Constant search soeed is maintained.

e. A number of "passes' are miade on both sides of the target by

the searcher.

f. The instantaneous probability of oetecting (y) the target is

dependent on the surface area and intrinsic contrast of the target.

g. The altitude h of tne searcher is small comoared with the range

to the target.

Chapter 3 discusses the extent to which these assumptions were met

during the visual detection experiments and actual SAR missions. The extent

to which deviations from these assumptions may cause differences between

predicted and actual results is discussed in Chapter 4.

.. 2.4 Uniform Random Search

The uniform random search model represents the case where the least

"nformation is known about tne target and no systematic search plan is

jsea. :' can be shown that for this case POD = I - e" W/ S 1 1 - e " C (see

Reference 4 for a derivation of this relationship). Figure 1-5 shows the

-leationsnip between the predictions of the inverse cube law and uniform

ranaom searcn models. As would be expected, the inverse cube law provides

nigher predictions of POD for the same coverage factor than the random

search model because a systematic means of searching is assumed.

'j. 1-7

3.9- INVIRSS CUse LAW

3.5-

2

1. 3..- 13 Z1 3 . .4 I .

COEAEFC AilAur -maia fIvre-,o .wadUiomRno ernMdl

Chapter 2

CONDUCT OF EXPERIMENTS

2.1 INTRODUCTION

The following sections describe the controlled experiments which pro-

vided the basis for this report.

2.2 GENERAL DESCRIPTION OF EXPERIMENTS

2.2.1 Time Frame

Data were gathered during experiments conducted from 11 September to 6

October 1978, 16 April to 22 May 1979, and 17 September to 25 October 1979.

These time frames provided a reasonable mix of weather conditions while

avoiding interference from summ~er air and surface traffic.

2.2.2 Location

All three experiments were conducted in Block Island Sound (Figure 2-1)

in areas ranging from 60 to 300 square nautical miles depending upon SRUs

involved and prevailing environmental conditions.

Whenever possible, searches were conducted in the same manner as actual

SAR missions. Twenty-four hours prior to each search, the Coast Guard R&D

Center released a SAR exercise (SAREX) message to all SRUs involved providing

detailed information necessary to conduct the desired visual searches. These

SAREX messages defined the search area, assigned search patterns and track

spacing, and provided other information that would be essential during actual

SAR missions.

2-1

x z

66LU <j jZ % -

Uohu.j~

0 On

Co

ILI~LRA LA. cn0Q

zz

4

I- c

2-2.

2.2.3 Participants

Numerous surface vessels and aircraft participated in the visual detec-

tion experiments. A brief description of the characteristics of each type of

SRU and a list of the individual participants are given in Taoles 2-1 and 2-2.

Table 2-1. Search Unit Characteristics

Crew Maximum Height of Eye

SRU rype Ce Speed Navigation Equipment (ft)Size (knots)

SAR boats

41 ft 3 20 OF'' 2 , Radar, Fathometer 10

44 ft 3 10 OF'' 2 , Radar, Fathometer 10

Cutters

82 ft 8 18 LORAN A or C, Radar, OF1'2, 25Fathometer

95 ft 12 15 LORAN A or C, Radar, OFL2, 20

1 Fathometer

Helicopters

HH-52A 3 90 TACAN, LORAN C2 --

HH-3F 4 115 TACAN, LORAN A, Doppler --

Computer, Radar

Fixed wingaircraftI

HU-16E 5 145 TACAN, Radar, LORAN A or C --

HC-120 I 9 300 TACAN, Radar, LORAN A, --IINS'

'Direction Finder.24ot used in experiments.

'inertial Navigation System.

2-3

vi

Table 2-2. Participating Units/Facilities

CG Light Station Montauk, NY

CG Light Station Race Rock, New London, CT

CG Light Station Watch Hill, RI

Naval Underwater Systems Center (NUSC) FORACS Facility,

Fishers Island, NY

CG Air Station Brooklyn, NY: CG 1442, CG 1368, CG 1424, CG 1391,

CG 1410, CG 1388, CG 1384 (HH 52A)

CG Air Station Cape Cod, Otis AF3, MA: CG 1473, CG 1479, CG 1484

(HH 3F); CG 7254, CU 7250, CG 1293, CG 7213, CG 7214, CG 1016

(HU-16E)

CG Air Station Clearwater, FL: CG 1351, CG 1340 (HC-130B)

CS Air Station Elizabeth City, NC: CG 1340, CG 1347, COG 134a,

CG 1346, CG 1341 (HC-130B); CG 1504 (HC-130H)

CGC Cape Fairweather (WPB 95314), New London, CT

CSC Cace George (WPB 95306), Falmouth, MA

CGC Cape Horn (WPB 95322), Woods Hole, MA

CGC Point Bonita (WPB 92347), Falmouth, MA

CGC Point Jackson (WPB 32378), Woods Hole, MA

* CGC Point Knoll (WPB 32367), New London, CT

, CGC Point Turner (WP8 82365), Newport, RI

CGC Point Wells (WPB 32343), Montauk, NY

CG Station Block Island, RI: CS 41441, CG 4A349

CG Station Montauk, NY: CG 41342, CG 44348

CG Station New London, CT: CG 11413, CG 41337, CG 41350

CG Station Point Judith, Narragansett, RI: CG 41385, CG 44352,

CS 4,321, CS 44349

.] 2-4

*1r

2.2.4 Navigation

Navigation techniques used to execute assigned search tracks varied for

each SRU type. The HU-16 aircraft generally used LORAN A or C, HC-130 air-

craft used an inertial navigation system (INS), and helicopters used TACAN

when altitude and conditions permitted. In addition to TACAN, HH-3 heli-

copters used a navigational computer which input LORAN A, Doppler and

TACAN during their searches. Cutters generally used LORAN C navigation while

41- and 44-foot boats were usually limited to dead-reckoning with periodic

visual and radar fixes. Table 2-1 summnarizes the navigational equipment

available to each type of SRU during the experiments.

A qualitative review of representative SRLI tracks indicated that cutters

with LORAN C navigated most accurately while aircraft relying upon TACAN or

LORAN A had the most difficulty in navigating the search area. Figure 2-2

shows an example of an HU-16E fixed wing aircraft search using LORAN A.

2.2.5 Search Tracks

Search unit tracks were laid out in the same manner as actual SAR mis-

sions. Two basic search patterns were used: parallel (Sketch 1) and creeping

line (Sketch 2) (Reference 1). To make best use of onboard navigational

equipment, some units slightly altered the basic patterns (Sketches 3 and 4).

a. Parallel Search (PS). Search legs were parallel to the direction

of the major axis of the search area and were separated by a

specified track spacing. Commence search points (CSPs) and outer

search legs were one-half the track spacing (S) inside the search

area perimeter.

2-5

IMa

1223 -_ 1228

I !

X1

1218 X4

- - -- - - -- -- -- -- -1213

1223 1203m m-1 i

IN

|XX2l

IN

-, |1158 .- - .

START 1208SEARCH X3

3 5 AREA SIZE: 20 x 12 nm01 234L INTENDED SEARCH TRACK

SCALE (nm) ... ACTUAL SEARCH TRACK

NOTES:1. HU-16E SEARCHED FOR 16-1t BOAT TARGETS (1.2.3.4) AT 1000.ft ALTITUOE AND 120-knot SEARCH

SPEED USING 4.nm TRACK SPACING.

2. ENVIRONMENTAL CONDITIONS: VISIBILITY A nm, WIND SPEED 6 knots. CLOUD COVER 100%. SWELLHEIGHT 0 ft.

3. ACTUAL POD OF SEARCH WAS 50%: TARGETS 1 AND 4WERE SIGHTED. 2 AND 3 WERE MISSED.

(iJ) Figure 2-2. Actual and Intended LORAN A Search Track for HU-16E(Fixed Wing Aircraft)

2

, 2-6

I iA r'r

c p

ISCENTERI j . j

7 MAJOR AXIS

I MINOR AXIS

L'

Sketch 1. Parallel Search Pattern

b. Creeoino Line Search (CS). Search legs were perpendicular to

the direction of the major axis c¢ the search area and were separated

by a specified track spacing. Start points and outer search

legs were one-half the track spacing inside the perimeter of

the search area.

fS S

Sketch 2. Creeping Line Search Pattern

c. Cutters with LORAN C (HU-16E with LORAN A). The two basic search

patterns were skewed with respect to the major axis so that the4cutters could follow LORAN C lines, and the HU-16E could follow

LORAN A lines. The basic search patterns assigned were parallel

search LORAN (PSL) and creeping line search LORAN (CSL).

2

, 2-7

LORAN LINE

- - -~-LORAN LINE

{LIN4 - LORAN LINE

"' a o . 4 LORAN LINE

Sketch 3. Cutter and Fixed Wing Aircraft Search ?atzern

d. Helicooters with TACAN. The two basic search patterns were sKewed

so that the helicopter could navigate along arcs of constant range

from the Norwich TACAN station (modified parallel searcn) and

from the Hamoton 7ACAN stacion 1modified Creeping line searcnil.

TACAN is a distance measuring navigation net ano was one of the

means of navigation availaole for helicooer searcn.

iS

Sketch 4. Helicooter Searcn Pattern

2-8

r

2.2.6 Taroets

Target types varied from experiment to experiment and included 16-foot

blue or white boats and seven-man life rafts (black without canopy or orange

with~ and without canopy). Either boats or rafts were anchored at predeter-

mined locations for each search (see Section 3.4.1). At the beginning and end

of each day or whenever targets were relocated, a microwave rangi ng system

(MRS) was used to accurately mark the position of each target.

2.3 MICROWAVE RANGING SYSTEM

In all three experiments, a microwave ranging system (MRS) monitored the

positions of SRUs and targets. During the fall 1978 experiment, a master

transmitting unit was located at Race Rock Light Station with a secondary

transmitter at Montauk Point Light Station. To provide coverage over a larger

area and a more consistent performance under poor transmitting conditions,

the master station was -moved to Fishers Island and another secondary trans-

mitter was added at W~atch Hill Light Station for the 1979 experiments. The

geometries for both MRS configurations are shown in Figure 2-1. Each SRU

(except HC-130 aircraft) was equipped with a transponder which allowed the

system to track it. The on-scene commander (OSC) vessel was also equipped

with a transponder so that target positions could be marked by the system.

To provide data for track reconstruction, the position of surface SRUs was

recorded every 3 to 5 minutes and the position of aircraft SRUs was recorded

every minute.

For a more detailed discussion of MRS operation, see Reference 2.

2.4 DATA ACQUISITION

AEach day, an observer was aboard each SRU to collect data and monitor

crew search procedures. As the SRU swept through the area along the

assigned track, lookouts reported all target sightings to the observers

along with information which would facilitate post-exercise validation

of the sightings (search units did not divert from track to identify sight-

ings). To verify sightings, the following information was recorded:

2-9

- -_______r____,

____,

______

a. rime target .as sighted,

b. Aoproximate range and relative bearing to target,

c. Relative bearing of sun,

d. Searcher course, speed, and altitude,

e. Target description, and

f. Lookout position.

:n addition, the OSC and observers periodically collected environmental

iata, including visibility, wind speed, swell height, sun elevation, and

cloud cover. Time on task (search time) was also recorded by the observers

on each SRU.

2.3 DESCRIPTION OF EXPERIMEN1T CONDITIONS

2.5.1 Sunmnarv of Detection Oooortunities

Table 2-3 provides a summary of the total SRU resources dedicated

to the evaluation of the current PO model in terms of searcn time expended

and number of searches conducted. Search time is defined as the cumulative

, number of hours each SRU type spent searching 2 during the experiments.

The number of searches represents the total number of complete searches

conducted by each SRU type. The breakdown of the 966 target detection

opportunities is aso given for each type of search unit.

A.1

2-10

', 4

Table 2-3. Summary of SRU Resources

Total Search Total No. ofSRU Type Target Type Time No. of Searches Detection

(hr) Opportunities

41/44' boats Boats 101.6 33 138

41/44' boats Rafts 73.7 29 106

92/95' cutters Boats 123.2 37 133

32/95' cutters Rafts 87.8 37 128

Fixed wing Boats 37.0 41 114aircraft

Fixed wing Rafts 30.4 46 105aircraft

Helicopters Boats 44.6 56 145

Helicopters Rafts 28.5 43 97

2.5.2 RanQe of Environmental Parameters

An effort was made to conduct these experiments under conditions reore-

sentative of those experienced during actual SAR missions. Table 2-4 showsthe range of environmental conditions that existed during these experiments

and the percentage of FY 1979 SAR missions that are represented by these con-

ditions. In general, the environmental conditions not represented in these

experiments are the poorer conditions (visibility < 5 nautical miles, wind

speeds > 20 knots and swell height > 4 feet). These conditions are not repre-

sented in the data base for two reasons:

a. Conditions in the search area at these times of year infrequentlyreach these extremes and

.4

b. Degradation of conditions much beyond the values above would

cause cancellation of the experiment for safety reasons and/or

to prevent loss of or damage to the targets.

2-11'.4a

Table 2-4. Range of Experiment Environmental Conditions

I

SRU Target Range of Environmental Conditions*

Type Type Visibility Wind Speed Swell Height

(nm) (knots) (ft)

Surface Boats 3-20(91) 0-22(98) 0-4(93)

craft Life rafts 3-13(91) 0-17(93) 0-2(77)

Boats 5-15(83) 0-20(97) 0-3(87)A ircr aft

Life rafts 5-15(83) 0-20(97) 0-3(37)

*Numbers in parentheses indicate the percentage of FY 1979

SAR cases involving 16- to 25-foot targets that are represented

by the range of environmental conditions experienced during the

experiments.

2

2-1

• -1

Chapter 3

ANALYSIS APPROACH

3..1 INTRODUCTION

The following sections describe the methods used to reduce the experi-

mental data, and compare the data to theoretical predictions.

3.2 ORGANIZATION OF RAW DATA

Raw data from these experiments consisted of the following:

a. Plots of the actual track followed by each SRU on each search it per-

formed (generated from MRS data). These plots include the locations

of all targets set for each search.

b. Plots of the assigned track for each search. These plots were gener-

ated using instructions given to each unit in the daily SAREX mes-

sages. Instructions included search-area center point and major

axis, assigned track spacing, area size, CSP, and search pattern.

c. Data sheets compiled by observers aboard each SRU and the OSC's log

for each day. These materials were described in Section 2.4. A more

* detailed description can be found in Reference 2.

3.2.1 Determining Opportunities and Detections

To be considered a valid opportunity for the POD analysis, a target had

to meet the following criteria:

a. It must have been located within the boundaries of the search area.

b. It must have remained in the search area for the full duration of the

search.

"4 3-1

Any target 4hich net these conditions and was sighted at least once during the

search wMas considered to be a valid detection. The POD for each search was

the ratio of detections to opportunities:

POO = numoer of valid detections

numoer of valid opportunities

3.2.2 Determining Nature of Tarcet Distribution

To evaluate target distribution, each plot of an assigned search

track was divided into a normalized 5 X 5 coordinate grid as shown in

Figure 3-1. Both the horizontal and vertical dimensions of the search area

were divided into five equal parts. This procedure set up a normalized coor-

dinate system which could be used to compare target locations in a consistentmanner over all searches. The distribution of targets with respect to this

grid is discussed in Section 4.3.

__r

II7--7

Figure 3-1. Search Area with Normalized 5 X 5 Coordinate Grid

1 3-2

3.2.3 Sortina Data

Raw data was sorted according to SRU type, target type and environmental

conditions so that POD versus coverage factor curves could be constructed for

each data base. A coverage factor for each search was calculated using the

assigned track spacing for the search and a sweep width estimate. These

estimates of sweep width were generated based upon the methods used in

Reference 2. These sweep width estimates considered all factors previously

found to influence lateral range distributions, search unit type, target

type, and applicable environmental parameters such as wind speed, swell

height, visibility, etc. It is noted that many of these sweep width estimates

differ substantially from those in Reference 1 presently used by operating

forces. Raw data files containing target detections and misses, coverage

'actor, environmental conditions, and track spacing for each search can be

found in Appendix A.

3.3 EM1PIRICAL POO VERSUS COVERAGE FACTOR CURVES

The raw POD versus C data for each SRU type/target type combination was

plotted using a computer binning routine which sorts data with the assumption

that the dependent variable (POD) is a nondecreasing function of the indepen-

dent variable (C). This technique, described in Reference 5, is consistent

with the expected POD versus C relationship and provides a smooth data set for

curve fitting.

Curves were fit to the binned empirical data using a weighted least-

squares regression computer routine. Fitting functions evaluated include

POO I-e- KC and POD z Tanh(KC). These functions were selected for the regres-

sion because they exhibit characteristics that search theory predicts for the

POD versus C relationship. Characteristics which the fitting function should

exhibit include:

a. PODO 0.0 at C =0.0.

b. POO is asymptotic to 1.0 as C becomes very large.

3-3

c. POD is a monotonically increasing function of C.

d. Slope 1.0 at C - 0.0.

e. Curve is concave downward.

3oth fitting functions conform to these characteristics when K equals 1. in

cases where K does not equal 1, the slope (item d) oecomes equal to K at 0

coverage factor. (Note: POO = i-e"C is the Random Search Curve function.)

The coefficients of determination (R2 ) of each function's fit mere compared

for each of these models and also for the inverse cube law and random searcn

curves. These results are discussed in Section 4.1.

3.4 COMPARISON OF EXPERIMENT CONDUCT WITH THEORETLCAL ASSUMPTIONS

One of the more important considerations in ccmparing experimental

detection results with the present detection model was to determine the simil-

arities and differences between the assumptions of the model and the manner in

which the experiments were conducted. The experiments were not designed to

ensure consistency with POD model assumptions but rather were designed to be

conducted like actual SAR missions. Table 3-1 snows the assumptions associ-

ated with the POD inverse cube law model, and whether the R&D Center exoeri-

ments and actual SAR missions were consistent with these assumptions. The

* following sections discuss these individual differences and their potential

influence on P00.

3.4.1 Target Location

The current POD model assumes a random placement of targets within

the search area. For the initial experiment, the target distribution was not

uniform; there was a greater target density in the center of the search area

than on the perimeter. For the two subsequent experiments, an essentially

uniform target distribution Mas develooed over the course of the experiments.

:t is postulated that the target distribution for actual SAR missions is not

,* uniform, but rather, as was the case during the initial experiment, typical

3-a

L -6A

ea A

V W fa

0 - I

I ~ >C.r %.

LM L

oX A

e1 00 cu

I0 >

cc 8 U W => I10 6nWi'Avi -

L. kA

C a- 0 J. 4.

0 a, =1e U0 & Lvi j J S- Q)m L

1(a .- J ea.cu tI vi .w

LW LW ~3-5

targets are most likely to be near the center of the area, with the probabil-

ity decreasing as one moves toward the area boundary. In general, a most-

probable position is determined based upon the last reported position of the

target, drift, and leeway. Based upon the uncertainties associated with the

estimates of the parameters, an area around this datum point is then defined.

3.4.2 Search Methods

The current POD model assumes that the search area is covered by equally

soaced parallel sweeps and that a number of passes are made on both sides of

the target by the searcher. For the following reascns, deviations from this

assumption occur both during the experiments and for actual SAR missions:

a. For those areas farther than S/2 from the search area perimeter, the

intended SRU tracks result in each target being between two equally

spaced parallel sweeps; however, those targets within S/2 of the

perimeter are not covered by equally soaced parallel sweeps. There-

fore, these perimeter areas are searched less thoroughly than pre-

dicted by the current ?OD -nodel. However, note that the effect of

nonuniformity of this search coverage is reduced if targets are more

likely to be located near the center of the area.

n. Probably more imoortant is that SRUs were not able to orecisely fol-

low their intended tracks; therefore, the spacing between tracks was

not uniform, resulting in some areas :eing searcnea either more

thoroughly or less thoroughly than intended.

3.4.3 Constant Sweeo Width

The current PO0 model assumes tnat sweep width remains zonstant throuch-

ou; the search. Due to varying environmental conditions, sweep NIth is a

continually changing function. Vernaos more 7moortant, sweeo widtn ias been

found to continually lecrease with ,lcreases in time on tasK. Therefore, evenif a!!l other parameters remain constant, sweep mioth Nill decrease 'nrougnout

a search resulting in a reduction 'n :he oredicted ?CD.

3-6

I

Chapter 4

RESULTS, CONCLUSIONS AND RECOMMENDATIONS

4.1 GENERAL

Figure 4-1 shows aggregate PO data plotted versus coverage fact:r. Also

plotted are the SAR Manual inverse cube law model, the random search model,

and the two fitting functions described in Section 3.3. The most important

characteristic of the data is that for coverage factors of 0.8 or less, the

inverse cube law orovides an excellent representation of the data, while for

coverage factcrs greater than 0.3 the data falls between the inverse cube and

random search curves. Overall, either of the two fitting functions used fits

the data better than the SAR Manual curve or the random search curve. There-

fore, for making comparisons between data sets in later sections, the fitting

function that best represents the data will be used. The optimum value of K

for each fitting function was determined oy maximizing the coefficient of

determination (R2) using a weighted least-squares regression computer routine.

INVERSE CUBE LAW 9M) (5 5

09-

2 00 )R". EAI7 UV

0 24Z 16-oi-j

,5- ' 2°" 1*4)

" 3- 1 /110 051

- - - POD - Tann 10.943C)

- POO -=1 - .3OC

1012) ( DETECTIONSIOPPORTUNITIES RATIO

. 0 08 0.8 * 2 ' 1.6 1 3 . Z2 Z4 2. 2.

COVERAGE FACTOR iC

0, Figure 4-1. Comparison of Experimental Results to POD Models (All SRUs)

4-1

tI

.F , -

L -

Table 4-1 presents the R2 "goodness-of-fit" measure for both fitting func-

tions and R for the inverse cube law and random search curve. The results are

arranged from best fit to poorest fit. These results seem reasonable if one

considers the assumptions about the search patterns associated with each

model:

inverse Cube Law Random Search

Equidistant parallel Search is conducted withsweeps of the search area no systematic plan orare conducted, beginning method.at one search area bound-ary and ending at theopposite boundary.

Table 4-1. Goodness of Fit of Models Evaluated

Model Coefficient of Determination (R2)

pOD i1-e"(L"30C) 0.940

POD Tanh(O.943C) 0.891

POD = TanhC 0.36i

POO =-e "C 0.482rRandom Search Curve)

ROD - erf(-VI- C )

(. 0.352(inverse Cube Law)

NOTE: C is coverage factor.

in practice, the SRUs conducted their searches based uoon a systematic

plan; therefore, their performance would be expected to be better than the

random search predic:ion. On the other hand, the search patterns assigned

did not orovide for coverage of oermeter areas (within 3/2 from the search

area boundaries) itn the thoroughness assumed by the inverse cube law. Also,

Clue to navigation limitations, SRUs Mere not able to implement these patterns

exactly as soecified by the OSC. Thus, performance lower than that oredicted

by the inverse cube law is a reasonable result.

4-2

I

..4~- t -

The fact that the empirical data closely followed the inverse cube law

for coverage factors less than 0.3 and fell between the inverse cube and ran-

dom search curves for higher coverage factors also appears to be a reasonable

result. For low coverage factors the overlap between lateral range curves on

aojacent legs is less than for higher coverage factors, and thus similar devi-

ations from the intended search track due to navigational limitations will

have less effect on POD as shown in Figure 4-2. On the left is an example of a

lower coverage factor (S = 2W), on the right, an examole with a higher cover-

age factor (S = W). Tracks A and B are intended tracks, while Track B is the

track an SRU might follow due tn navigation inaccuracies, which would cause

the track spacing to increase oy an amount X. As these figures illustrate, anincrease in the track soacing by X would have a greater effect on the expected

value of POD for Case S than Case A (calculated by summing the POD contribu-

tion from each track assuming independence). For Case A, the probability of

detection at the intersection between the lateral range curves is relatively

low even if the intended tracks are followed, so that the incremental reduc-

tion in POD caused by the increase in the track spacing by an amount X is

small. in contrast for Case 3, POD at the intersection of the lateral range

curves is relatively high and the point where the curves intersect is the

region of maximum slope. Therefore, an increase in track SDacing by the same

amount X would cause a greater reduction in POD than would be expected fo-

Case A. Note that the magnitude of this effect is sensitive to the slope of

the lateral range curve. Sensitivity to lateral range curve shape will be

discussed in Section 4.4.

TRACK TRACK TRACK TRACK

A B A 8

TRACK TRACK

A

A / I

2W X W X

LOWER COVERAGE FACTOR HIGHER COVERAGE FACTOR(CASE A) (CASE B)

Figure 4-2. Effect of Deviations from Track on POD

4-3

U,

4 C SON OF ;eSULTS BY SRU 7"YE

SRU types biffer with respect to navigational capabilities, and the

shaoe of their lateral range curves for a specific set of search conditions.

Because of these differences, it is postulated that a unicue POD model fit

is approoriate for each SRU type. This hypothesis was tested in a preliminary

manner by comoaring the best-fit 00 model for each SRU type (see Table a-2).

As shown in Figure 4-3, all models fall between the random search and inverse

cube models. 'Ahile differences exist between tne best-fit models for each

SRU type, these differences are not statistically significant at the

90-percent confidence level. Further investigation into POD differences for

various SRU and navigational equipment combinations should be conducted to

better quantify the effect of SRU type on detection performance. Suggestions

,or this investioation are made in 3ection 4.5.

Table 4-2. PO0 Model Fits by SRU Tyoe

PO0O Tanh(K PCO KCSRU Type 2 _ ___K R2 R -

41V'/4 boats i.C55 0.96 !.167 C. _

32 '/95 cutters ).984 3.379 .397 3.333

Fixed wina Ia 1.006 0.315 1.46 1.759aircraft

Helicooters 0.934 0.421 1.238 3.712

4.3 TARGET 31TR.3UTTON zFFECT3 3N POD

3oth the inverse cube law and randcm searcn curve assume a uniform

A prcoaoility 1istribution of target location 4ithin the search area. ow-

ever, the target :istribut4on Nthin an actual SAR search area is iuite likely

not to .e -andom; the most probaole target oosit4on may Neil te in the center

of tne area qitn 'he or baoiiity of a target teing at a oarticular location

<1-

4k

INVER0E CUBE LAW

10-

0.8-

z Q.79

. 0.6-

U.

oos

. 0.3-

- - - -41 '144' BOATS: POD -Tanh (1.055C)

0.2 CUTTERS: POD = Tanh (0.984C)

... ......... FIXED WING AIRCRAFT: POD =Tanh (1.006C)

HELICOPTERS: POD = 1 ac

0 0.2 0.4 0.6 0. 1.0 .2 1 4 1 6 1.8 2.0 22 2.4 2.6 2.8

COVERAGE FACTOR (C)

Figure 4-3. Comparison of Best-Fit POD Models for SRU Types

decreasing as the location approaches any of the search area boundaries. For

the types of searches (PS or CS) conducted in these exoeriments, the perimeter

areas (areas within S/2 of the area boundaries) are searched less thoroughly

than interior areas (i.e., no overlapping of lateral range curves on adjacent

search legs). For a uniform target distribution, a greater percentage of

taroets is in the=: Perimeter areas than for a distribution with more targets

M in central areas. Therefore, the effect of a less thorough search of theAperimeter areas snould ne greater for a uniform distribution than the distri-

bution with more targets in the central areas.

During the first visual detection experiment (fall 1978), no attemot was

made to maintain a jniform target distribution and, as a result, the target

distribution was biased toward more targets in central areas (see Table 4-3).

P .4

Table 4-3. Distribution of Targets

Fall 1978

Location Within Percentage of Experiment 1979 Experiments

Search Area Search Area* Expected" Actual Expected " Actual

Number Number Number Number

Central areas 36 (9/25) 58 290(160 X 0.36) 9 306 X 0.36) 336

Corners 16 (4/25) 26 129(160 X 0.16) (306 X 0.16 91

Borders except 18 (12/25) 77 6 387for corners (160 X 0.48) 6 (06 X O.a8) 379

*All search areas were divided into 5 X 5 grids to normalize the results.

"*Expected number of targets in this area if the distribution were uniformover the search area.

During the two subsequent experiments conducted in Block Island Sound

(spring 1979 and fall 1979), an attempt was made to achieve a more uniform

distribution (see Table 4-3). Therefore, by comparing the POD results

between these experiments, the effects of target distribution on the POD

model can be investigated in a cursory fashion.

The results of this comparison are shown in Fioure J4. The best-fit

curve for the 1973 data lies above that of the 1979 data. The two curves

differ sufficiently to reject, at the go-percent confidence level, the

hyoothesis they represent a single model fit. Thus, preliminary indications

are that target distribution differences may have an effect on POD. A more

rigorous investigation into detection performance for various target lcca-

tions qithin the search area should be conducted in the future to better

Squantify the effect of target location on POD. Suggestions for this investi-

gation are made in Section a.5.

4-6

INVERSE CUBE LAW

...... .. .. .. ..

• '......RANDOM SEARCH CURVE

0.8-

Z 0.7-

o oo."0.6- i

0.4-

o - 1978 DATA BEST FIT CURVE:

S.3- POD a Tanh (1.037C)

90016 CONFIDENCE BOUNDSFOR 1978 DATA

.2- - 1979 DATA BEST FIT CURVE:

POD - -- 1.27C

......... 90% CONFIDENCE BOUNDSFOR 1979 DATA

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 '.8 2.0 2.2 2.4 2.6 2.8

COVERAGE FACTOR (C)

Figure 4-4. Comparison of POO Models for 197S and 1979 Experiments

d.4 INFLUENCE OF LATERAL RANGE CURVE SHAPE ON POD MODEL

Another question of interest with respect to a POO model is its applica-

bility over a wide range of environmental conditions, since the shape of the

lateral range curves as determined from empirical results of Reference 2

changed dramatically when environmental conditions deteriorated (see

Figure 4-5).

To investigate this question, the empirical data which yieled "peaked"

lateral range curves (which were generally typical of good to excellent

environmental conditions) were compared to data which yielded "flat" lateral

range curves (typical of poor environmental conditions) (see Figure 4-5).

These data were binned on C and fitted with the functions described in

Section 3.3. The results with highest R2 are shown in Figure 4-6. Note that

the peaked lateral range curve yields slightly higher POO predictions than the

4-7

hL

0.9

0.8, /PEAKED LATERAL RANGE CURVE/(EXCELLENT SEARCH CONDITiONS)

0.6-

P(X) 0.5

0.4. FLAT LATERAL RANGE CURVE

03- !(POOR SEARCH CONDITIONS)

0.2-

0.0 0.5 1.0 5 2.0

MULTIPLE OF SWEEP WIDTH

Figure 4-5. Examole of Peaked and Flat Lateral Range Curves

flat lateral range curve. 'Ahile the difference between the two fits is not

statistically significant at the 90-percert confidence level, it suggests

that further investigation into the influence of lateral range curve shape on

PO should be made.

a.5 CONCLUSIONS

Th following conclusions are drawn on the basis of these ana'Yses:

a. he present SAR Manual POO versus coverage factor model overesti-

mates POO, Particularly at higher coverage factors. Other mathe-

matical models that predict performance between that which the cur-

rent PO versus C model and the random search curve Predict provide

much better fits to the empirical data. Therefore, there is a clear

need to revise the SAR Manual POD versus C model.

4-8

-. 1

INVERSE CUBE LAW

0.8-.'" / .......... RANDOM SEARCH CURVE

1.0,

0.8-

0 0.7-2

06

04- BEST FIT TO PEAKED LATERAL

0 ~RANGE CURVE DATA,ccPOD2 -

0.90% CONFIDENCE BOUNDS FORPEAKED LATERAL RANGE CURVE

-.-- ""BEST FIT TO FLAT LATEFAL RANGECURVE DATA: PO0 v 1 - 6 .2?C

.......... 90% CONFIDENCE BOUNDS FORFLAT LATERAL RANGE CURVE

0 0.2 0.4 0.6 0.8 1.0 1.2 .4 1.6 1.8 2.0 2.2 2.4 2.5 2.8

COVERAGE FACTOR (C)

Figure 4-5. Comparison of POD for Peaked and Flat Lateral Range Curves

b. With the development of empirically derived lateral range curves,

the standard method of determining POD should use these lateral

range curves by "driving" them through the search area and deter-mining the expected value of POD directly. The Computer-

Assisted Search Planning (CASP) system should be used to accomplish

this and a manual version of the CASP model should replace the

present POD prediction method in the National SAR Manual. It is

necessary that any manual "backup" POD model orovide predictions

consistent with the standard computerized technique.

' '4 4-9

4.6 RECOMMENDATIONS

The following recommendations are provided on the basis of these

analyses:

a. For use with the CASP model, it is recommended that a method that

determines POD directly from lateral range curves, navigation error

distribution, and probability distribution of target location be

implemented. POD predictions generated from the CASP model can

then be compiled in a form suitable for manual/calculator methods.

b. Because navigation limitations are expected to influence POD, a

method for quantifying these effects should be developed. This

development should include:

(1) Analysis of navigation errors experienced by each SRU type

during R&D Center detection experiments. This should include

development of representative navigation error distributions for

each SRU and navigation equipment combination.

(2) Convolution of lateral range curves with navigation error

distributions through the use of either mathematical expressions or

Monte Carlo computer simulation (depending ucon the nature of the

navigation errors). These convoluted lateral range curves would be

used in the CASP POD calculation.

(3) Comparison of CASP POD predictions with empirically derived

POD data to validate the CASP method.

c. Qetermination of POD directly from lateral range curves provides

4the opportunity to evaluate alternative search tactics. Using this

method, the best tactic(s) for available SRU resources and a given

probability distribution of target location can be determined.

Therefore, further analysis should be conducted to compare

empirical POD results with model predictions for targets located

4-10

both between adjacent search tracks and near the search area

borders (within S/2 of only one search leg). The results of this

analysis should be used to establish and validate a method for

treating different probability distributions of target location in

the GASP model.

4-11

REFERENCES

1. U.S. Coast Guard. National Search and Rescue Manual. CG-308, Superin-

tendent of Docunents. Washington, D. C.: Government Printing Office,

July 1973, with amendments 1-2.

2 Edwards, N.C.; Osmer, S. R.; Mazour, T. J.; and Hover, G. L. Analysis of

Visual Detection Performance for 16-Foot Boat and Life Raft Targets.

U. S. Coast Guard Research and Development Center and Analysis &

Technology, Inc., February 1980.

3. Koopman, B. 0. Search and Screening. U. S. Navy OEG Report No. 56.

Washington, D. C., 1946.

4. U.S. Naval Academy. Naval Operations Analysis. Annapolis, Maryland:

Naval Institute Press, 1968.

5. Richardson, H. R., et al. The Contribution of Operations Analysis to the

1974 Clearance of Unexoloded Ordnance from the Suez Canal.

Daniel H. Wagner, Associates.

I

Apoendix A

RAW JATA

This aooenaix contains raw data files ccmvolea curing :he :nree s;

detection exoeriments for each SRU tYoe. Each row of ,a-a eoresents suc-

cesses cr failures on an individual search. These data were used to oeveloo

the empirical P0C versus coverage factor curves presented in this reoort. The

foliowina is a key to :ne format of the raw data files:

Column 1: Search Numoer

Column 2: Numoer of targets detected :r missed turina the search 'ietec:ln or

miss indica:ed in column 1".

olumn 3: detection or Miss 'I : te:-cn C %.ss'

Column C: Coverage Factor

Column 5: Tar:e: Tove ', I =5-foct 3ca: 2 Raf

Column 5: Visibili:y (nautical miles)

Column 7: wind Speed* (knots)

Column 3: Swell! eich: (feet)

Column 9: SRU Tyoe 1 :/44 Boats, 2 : :ers, 3 ' xea 4 ng Aircraft,

4 -elicooters,

Column 10: Track Soacing ,nautical .miles)

'rime-weigned average for the search.

"4-

L K

'I- ,J

" ./ .- u 1 .uj .o, .. u I.Uu -. V

u .. .v iuu e .o j ..' i.v,'

6~ .J 1..') u*v v I v~

" . I u .j u 1 , I o u o *u v

J. I L.LJ ni.ju d . v I. u .

u if I.S v, v , L~~U.#v .vu a v . Vu

. ../ .Uj . Iu 3 . u .uu Q v vv0 * .t I.U * j I .* U 'I I. e l ., u .* U "-. UU

v .v~ v u~v 0 j. v -u m .0)u .v v o u v

" 1 u . .QU I U . Iu .v) . Q 4vv If.vv U.05 .uu lU.Iu b.lk l.l) I.U ".')1 0 .I .(u I ) .,u )o.1 U .u .I. 5J v .uu

' .I v ..u I o. U 4 .u ,j l. . ) 1 .o u vI v .a U. I .u L .iu vo.u u./u i.".

Ij I v~h Iu uIvvId I/u Iu vvS I u.7o .U I9.'uI e 1. 5uvv I.Uv

. e Q u u . v I. v d . ')) ) k.UU .Uv

' sn. (.U) Ij.4 J.U ) .'u . uj

,J I I .,Ju v.u .. u e.Vu I.v-

Io e 1j u.4d I°J l.Uu I.I)U ¢.'.I v .UU L~u .)4

v , u 3 i U o. v I)u ' U v .vu-' I u° t .UJ IU°. ,J l.SU I*v(u i.uu .u

1. *U'J.I U o U .zv, u . * u.

e f u u. I i .u i.,u I . v vu I4 .uu U

. V € . u e. .,U o U." v .v, U i ,, Q.v

1.u v u u uu vu

' 'J ) * ,i *.u * j uo v ' . oU *,IU

It .u I. I' .u 4

. 'q . UV .u v

10 - i .vu I.JU I 3.uU iio o., .uu -. uv

,. ~~~j V V 4 v.)'

.v .U .F . u". v v u

-.3 4 u ').' I.Uu l:.u u .. T4 i .UU -. Uu

- I I. - u v 0.UU o.'4 u.I, . u.UU

1Q u

I~ u u.4 U 0 .u I.' *' o

u ' . .u v u i . '.p' e.oU ..

, I o e.'Jv -'v. u .0 e ) U Q

3 1-' 'd u . 11 ~**0) 7u * 0 u . U

j I u i .. ) u I'z.j .. v . ,) .vv U. u

I I ', . ~ .0J I .'J I.Uj 'J.lU 'a *-.v* v .a u .0 IJ'~ u Uo eu . ~ U J

3u~~~~ o, Q u .vu V.~ .. "D uVi u v' ,U

5. ,. .. . . v -35 , '

.... 5 II~ |111u

- 1 i I .UO €°)'j 1 °'j 1° U ° i°V J~

Se e1v u 1j v . v.'i Vu.5 j c v ,V Q!..,u o. Vj v.Z 1.uv v

u .13 (f u kI (i nJo --j. (Ic 1i U I V3I e d UV .o13 C, u 'D.c v, iVl 0.v D .iuVVv

5. 1 v v.O d .Vu -f. -.. u a .i 1 .uv3. v . 33 2.u l c. i vu o.c: e.uJ o.uv v v

j u v .3. e.vVj C 1 .1V 1. ) d .uU V.IJy i.lJ4J

So 1 v 43 c. u V zuu 1VJ .':I .v u V V3o d v. 0: e .uv "2~') 0u. u V u." .) IV l37 . 1 .0 e .u 0 1 .uu. U 1.v' o -. Q.3 v j . e.I d I..Jv '.ov V.1 ).u ).VQ

31 1 v . )j ,.u I .,dV -. I.v .uv v3c v U.o ,.uu li.e'j /.OU I.Vo I.VV 4.VV3Q -4 1 v .. uu..5 euu I .u . u D.vu v.)v

,4 1 . 2.JJ Q.VuV .I,) L u .uu u.v4U "v .7.) d. v v ed.u v .VU . vu '., 0 j )

41 .t) u.U U lu.vu . 4.. Q .uj ,VV

'A 3 u .u.u vu 1 .vu .. U .5 .u u v U

Q L 1 u.v 2.uv v .uJ V.uv u.')u . v . u

4 I u 1.lV .u) l .vu .UU v. u . v UV

5 j.o1 1.uu .1Uv v . , Q.uv kive u .o e.uu U ) . U .U .uV uVe I vo ::iu . I ,uV .5 V L v i~ V V vv

- I.l Iu1 u v 1 .U -. vu V .)U Vv -4.VU

C4 I u 7.Tv .vU I .u 4.o 9 U % .') 0 tv . v V'4 1 v .3u .uu v .. U u.v Uvv .V u.VV"'3 1 1 u vu D vo '~u 1. u V

o 1 1.935 1.UU 1 .vUJ U. . ) 1.uU "u

,4 1 u 1.U 1.UJ 0.U r. vJ v.vv "IV

wI u 1 U 1C.u5 u.u u i. U V. I. u .47 v u u.O lv UO i vu u ov lvuV ,u V

"o 3 1 . v .Uu 1 .vU e.u0 v.tv i. vu j. Q'0 1 o.UO i.Uv u.VU .U .) .uu - v(44 1 u.' u . ., o.u v.5u .uv u., V

S .5 U U.14V .V ( .. UJ o.00 Q.)U . u -. UVtu 1 u u.52 I.u U .Uu .U :.3 I.uQJ r.v

I t.n? 7 .u v . A: C1vV I.Vu Q. y51 d u v . I uO l<b V.g t)o1u I. u V \;

59 e I I .1 2.UV 1 .5u )iO v.v . v VVe- ! U D.I/ ..uU .oU d. Iu V. v vv i.v

53 .5 1 U . 2..u 1 3 u 4.iu .oU 1.VJ .VVV u .6v .u V 13.5 1./u v.ou l )v .u v

5 1 1 U.1U , .vi) u v. I'. v .iV .v5'" 1 u . .uV l.V 1.7.'A v. u v u.vv

-4 1 4.4v e. 0 v).Iu ,0.. V u 7.9u V.u1S 1 v.u e.v 'D.u U v.l A .)v V

3/ 1 '.7V .Ov a.uv .uU l.vv u V v

av 1 1 j. 0 u k.U u V'ov c.O.5 2 v k.l 0 .vV u 1). .~v u~v ,u 0 v

" U C o u.ou @.U 13.uu 1-'.OV 1,/,' i.uu .v

e- 1 v C v v v o .du u~ . 3 oOC 1 d,.4U C.LU) .1U .0U L.vj v. v

04 A-31 ,

I II mlll kw

I i du I)o 1.U eu~ j v .u u e vuQv u.3 1~ u d V. u ) :) v 1.) v ) t v;.

e ~ 1. J 1e.v)u ju.4j e.vf) jv 4V

v3 o 1 .~u u v u , .vV d 1

"Duu .I .., 1.JU e .v~u d uj

v 1 * u I .i 0 u a v / V *- jt) .103 0 1.u .,Iu u v.~ vv v I.j DU'i 1.~' u : UU

1o u v~l v.U vuJ d1 ~ 1. U() e. u v

1~ .'1 I.'J ~ u i.j 1. u VUi'

II ~ 1 1 * Q t) *u l.u Q ~ o 1 u~ u 41u 1j o ..' u.o L u . u u.~ Q u

ue 4 Q.. 1u v I. udmu ITv eu v

111 u I u U I 1I0 l.u o I.'.) e. -V 4

I~ tj u, v.. v I.)Q'.) 1~Ji e.59 d.., v .vu:~ u 1.S u.). Q~J 0.00 1 o - vAiu

uQ u u.3 u. 1. u v .u 9 u u u v v

I~~~ u u c:3 ..o v i.I U~O ~u~ u5 u ..' I~u i u (iv e u eu I

7 v 1 I v~i ..J' i .) idJ u.j V

3i u .1 u.~ .. ) ~ . uuI u o. u v d.Vv3o. 'J' I~ n.0 ~ . vvIv u 0.7 u 2.0 u .0v

.) I :)~ .. O* ;) d 1ou.) -J ,9t

3 i v I.' 1.oj 73* ~ .1 1*UL u e. t vjO.3 .,33 1v o ~ u.1 v.o'I .v

I u .3. UV 'JV o d.~5 I I d U 1.u u~~ I) tV e. v) w.0

64 1 0.4 1.0 0~J Q.7 -o e. ') ).4 .5 n .J 1 ov I j 1 rj v V v

e 3 0.0 1 ti4 4 U 1 1 u o r-~ ~ . VU I

3 c i 5 100 v d., 1.o 4*t v e J e v.5 1 ~ . .vu Q0.u I 1.,4) t lu li V

00 1 ,- J.o 1 u I.., 7' o mvv3 -'3 3 j ') v~-. I.O av e. u e ujj

3. .. 0,3 u Q U e. .~ ij i~ u u d*.,.,j

4e . I S.* v .d* .v

.5 . -)

3 i-1u u

.1 geS .2uv I~o vu 0 4 v' e u v3 1 1.3 S ~ 0 e ~ 0 u I u U 0*u :) C vv

13 1 .4o . 0 u .. 0 IS0 0.u e.0 u ".:u r0V0io ~ .55 v2o v ~ o I.o o*S u Z0( ) dvl

5~1 I.0 v.0 l .40 u 1C.I v d.0 vu.VV

.4uI-) v IO v1b /.1 1tC ~ .v '.S 1 1.03 2.ou 0.5 44.iuo 1.oo r-.oo .

u 0. I Q u 11.5u 7 .o 4 u u r-. vuto. 1 u v7 2.0 1 1 u '4U u zo e. u k 4

1~ 1 1 0b U) 2.00 tw 0u O. u e ~. uU u

u 0 3 . Q e .o 'Dv t .t u . !) c e. Qu.. L o. 43 U 2 .v P .5.oU lo 0. ") u e.O .v.00

L) b Q o.. It.0 .. o 10 v UL'u )( . v . u u

2 0 0.43 L,.o e. Vu o. 0ij u .2.uot)u41 1- 1 u . o 2 .0 u o .I .5 .So 1. u u u 0

40 u u u I w u.o uo u.2 1.,o cu ( .vo.40 v v 1.I u.u 1. v u u d .uuveuo

u2 1.o .. 0 1 .5 .u0 e.~ e.O v vSe 1 0 1.1 o.o I u d .. U.> eo v

5d 1 v.4o d. uu 1.u ts~o u d. 0U

11' 3 1 1.O d 2.0 11.2o0 .5 . u0 v .0 n,0 v e.00vU

1 4 3 1 ? U.o 1.0 1'.o 119 o. '- .vu e

5n 3 1 1.33 1.VO 1~.)u d.uo o.5o d.oo -4V51 '~ 1 1.0d 1.00 1'.0u d.o,0 U..OOi e-.0u 4

514 1 4 1.( 1.0 I vu I u 1 9 .u0 35.e2u c.v00 C.0 v 3 1 u Q 3 1.0 u'l u vU.00 su 1) e . vv Zf3o U 1 u U.- 1.o v .70u o.d 6 6u v 2 . u u - ,P)l 1 1 u . () 1.0 0.0 u ID. O evu d 2.0u0 .5. Qv

2 u 0.q1 1.00 0.00 0.00 -,..o d.oO J..vv

oc 1 1 1 . t)1 2. iu 1 o0 u tl50 '.)u c . Vv u .0oc I U 15 u.0 V"d 1.5 0.') .. 0 U 0v

.5 u o. 93 P2.0 u ~ o 10, I.0 1.o u.0 I .vr00

d v .- 93 2.00 t 1.o 10 Iu.0uu0 1vO u .0 e".00ud2 1 1.b0 2.00 10.2v 4~.3u 0.00 e.vv .5.00

~ I .50 2 . 0 1 f..du92uv .~ t.0) 2 .v0 v .00

I u .7 o 2 . u0 1 -10 1 1 .3u 1.v0 d e. vvvo u o.T 1 2e.00 1 610u 1 1..5u 1 .00u . v vk,o~1 .)~ 2 . '7 U o . 3 u 0.00) e 2. vv ~ . v

1, . v n.0 2.0 v l/Io o . 5 v. ( L0 2.00u r .o/ 1 1 U.73 2.00 es.00 44.9v 0.-') d.vU C.Q0

A dv 1) 1 4 2O I v. to v.. e . v r c. ')

1u u .0 1 .. 7 I 1. e.o .v0 Q C Vv .5 1 v. 2.o u v I u 5.0I 1oj e. UUu . 7 v 2.k v ~ ' v I u v .U ~ O

1./ 2vv .0.0a01 'j. cv,j

rm S FPAGi -Jr 48T qUAL1241 *ML

-5 00

-=y

I ~ ~~~~ u.5 Q. 1.1)i lUU .J u SV

e~~~~~ ~~~~~ - 1 1, 1U .'U .) .U*uU

~~~~~~~~~ t) 1 .*) i.j jl) Ul 1Q ~4~~~ ~ U~ u 1.e Q.u L.U ~ .U.~

- ~ ~ ~ ~ 1.1 .,) 1.U 1uU 1.u ')

') Q lvi 1.u 1 u u.I '3 ) u.v u

v1 . l.5/ lu k u 1 0 4L oU 1.) jU .U

0 u Q.' 4.I Q u.V) vZU u vV

1 .1 1 V .5.~ 5 uuu u/uv U5uV e i v

kj 1 15 i. u5~ uS/ -V. Qu ) u5.v

i .5 1 1 *vU I *u u 5.UU v5~ 0.J .±JU z.

1 / 1 V 1.33 ~ u 1 *V .. u U. 3.). U

I~ ' ~1.U .f.VJ 1.v ~ U .5'u

. 1 w.U 1U 5 U .UU ) Q 5V

v u5 o v*u 1.U Q.U .U Q. .5.

*~ j 1) 4* 3 I. j t~V W v UUz

iu I Q Q 1.U 1~U Iu u .' v u)

2 -n 1 u v ' u 1.3) I.U ~ U) 4 .51) 43.J

I 13!D I 1 nv 1 ) ~'D u 1 *UU ..

3u 1 '10.U 1.U u ~ ( 3o ~ j5U

A~ u1 1 1.~ u.j u~. lJii U.5 u

I 1. ~u Q.U 41) '.u 1U 5

3- .5*I. s I~*Ju *~U*' 5*i

3.5~ J5 u j.5 j.U IJ.s v.5 Q 5V

P3 Q~1j55 ~ s 15' .J.I .5 .U -

S' . *~,j ~ .jJ 1.5.UV V 5.( O 5~~

T11 ;- '5)

-Lxe YU.I.

u/U .03 e.vu 1 ).'ju :).Du U*OU 3.uj U

v~ U.O ( uU a.O u .~ v u u v

'. 1 1.u C vU

u vv . . u.o d~u v. ou u c.ou v I u - t) o v..i '.)v U . V'j C:

d. u w v.~ v.j k, u u U Qv

4'- e~ I 3 eu~ 1u .vu v U.vJUvv ~d~ 3 I v~~i d.U i~ vi u u u . vvv

u5f i u u.v v.U I v

u0 1 ) -1 u~o I.U 1~U ou v . v Zv u

'0 7 Iuu 1.uu 1 .Uu I1.6u e.VI .).Uv : v

- 1 1 .1,.uu u ~ IQ 1 .u u e.1 )vv C.ku 14)u 1 ', u~s 1.U Qu Cl/U .uU

51 1 v.1 I.u 'o. iu I e . v o *S e 3uw vvr- e 1.u 1u u I 'D . v -. u u 1.u o( u U v

.5 3 U v I.~ 1.u v 'I. u lu) u o .u u v.uve I e.~ v .u iv Iu 0 .' ;I ju vuu d . u

5s 1 u u I uU 0~1 o. Iu 1.-u ~ ,uuk -

v u .71 ? v u I u. uu 1.~u 1 v z ..u v v li50 u j)1 ?.ou ]v.uu Z'.uV) 1.uu .S.vv )uU5' 1 1 0.6u d.uu lv.uU 4.lu i.uu .vu ).Uu

'.) I u u.b r7.uUv u. vvu 1.U U I 1.u u u .uu u

'. 0 I e1 ~ . uu d cU .o U i vu() .Uu uk

nu 1 s.01 d.Qu '4.vu O.UU 1.uu -. u )Uol 1 . d .uu 1u.uu) O.lu 1.uu 3.uu .u

v v ,.O o 3 v ou~ u.u0 1 v 4) .3 .uv vUI U Q uu 1 0u l. uu I uu .~u v kui v . 0u u.u I. u u e.uu I uu 2.v k .u v

1L 1 . 1.- ~ U) u 0 V ~~ . u o . vu v C:.

03u 1 v v.u u u eu o .u v I U' .U,

1 14: u u~l .U U e I .& o u 4u .u v c.ot J ./ , 2.u 1.uu 10~.2u d.-uv -I.'uQ C

00 1 u.~ ~ u I.u u I e v vu c .. uu

.on . 1 .i.Su) 0uu j~us 1*J Cuu* eu .5. u e ,*1 r.U~ u u u /.u v u v u u

u 3 1 .liu .u .I u u I.0 u i.U 'Du u

/1 I .u d .u ) v. 1.uu 1 o.u uk7 u v.I 2. vu k). u -04 u I.U - ( ki L) d~u c.u u r'-...u ID.l v u c vU ~ .VQ

4 A-7

TMI: ?"i W IALT ~k k~,)V .- rfF i

e 4,

Lii

e I Z*UUv

10 1 1.1 eu li 1'4U0 Q ~ v r) Q u ) e.u

7, u a. 1) V 'I.O -\ kr4 . 1 IU v IA J I'Q V

, I I . 3'3 14 J = .J' .Q .lu 9 k)"\ ii(j 1 3j.( u o v V

L ~~~ ~ . 4'',4 v u)v

1 0) e uvvQuu I.Qi . \

u uuQ4

u )' UuuvU .

2 u u '0I~ 1.u d12 OkiU I)

u w 1l .u 1 .u 1 .5 u e'.v ) vV'

.! 1 b 1.~ ~ u 11.o k, 1 c vvv

1! u v.~ u 0 . .v u.u 11O *v v~ 1 1.7u 1) v~ui 2u 2. V

Q 1 1 v.J V.U v~j Q.8o 1 u 'n4 u.~ 1.U '.,3o 1.u o uiU

I U 1 u u .,u e,:.')u i . Uu v u ')/ 1 u Qi.ou 1.uu et.vu iD.vu d.vQ V OO

v 1 : 1 . u ).vIu 2(.u1 d . uk .uuv o.u vk,1 ~.64 1. uv ~. vU I .uU 2. Iv 4.v d.D

'I v U. di 1 .wu 0 . V I .4j e.ia 4. uIi 1 1 o 1.uoi O.uv h1.:)v 1.041i "&.Vv '.

lu 1 u v .00 1 .u Q 0..4id) 11I'. ) Q 1 .'4 n 4"*4)4) ).1I1 V.00 1.) 1).uv d1.74) 1.V)4 4.vv ,.IJ4

d1 u Ui.60 1.Uu lo.v4) o.io 1.4)4 '.Vvic 1 .2,4 1 .u 1~u u wv I v u ".u .u U

1I . ?- 1. 1 .u () v u v vu "v vvue- I i e I2 1. u v ~u u '3. uu I .,u '.V Uv

I~ .5 1.2e 1.Qu 1).Uu 2.10 1.4)4) 4*4)4) .V

I'D 5 1.3,4 1.4)4 1.u 1.ou 1.UU -. Vu D.VuI~ 1 u .59 1*( . u .4e)4) 1 v u e . vv L 4 .uj .e v1?7 1 1 1 .31 1 . uv 4)v .4) 'dU ~. v v (1 C 1 1 1 .n4 1.u 0 m . u u .v0 Q v 5 o 'A . V e '1-4 1 1 k). 7 1 1.4u v 0 .40 /*viu I . I i 4 vV .2v) e I y.4 1.uu 1kU.4)4 1U.4)4) 1.v.i4 '4.vu )

2 v 1 0 Q 0. 4) I . u v Ik). u o () . vu 1 . v ) ". v Vv21 2 1 0.03 1.uu lu.uu lv.4)u l.4)4 *'4.vu ".u21 1 v v.06( 1u Iv0uIu lv Qu I v o '. u vQ2~ c I I Iu o0 1u v 1 Z).u u o~o U v dv Q '.v u u

3 1u u .u u 1 .4u ou u I . ku 14 .4U d -. !:i

e I 1~u 244u v I.u j. v v I. 1 4 .uQ-.'.

2o I u 1.24 1.ou 1J.)4 Iv.vo 1.Vu .4.vv t'

2o 1 1 1l u I. u u 1.). v vu.. 1. vv 4 . V ,

2 U J14~ 4j .u)o 1 Q v4 I~.( 1. o) "v uve 2 I r 1 5 1u u 1 .4)4 u.4 0. U v .u v

20 e 1 1.1;7 1.4)u 1).u .4 Q.v4) '.Vv !v

2~ q 1 ..2 1.4o 1 '3. v o 3. Do 0.0 : .)v vu3 P u 1 . vu 1 10v) .u)4) u. -u 4U U

m 3v 1 u 1.o 1.vu 110.4() 11.0v4 4).)U 4.4vv

51 e 1 u.O :i4i. 5. 'D4 t. u u 0.4 Q Vu Z)v

1 v u .5. 7v ,.U U v.4) v .V VvSc e d I.i 1v k 5./0U v.4 4JU4 vJU 3.v

3- v u.) u.v ~.7 .4 v)v4 v V0 v 03" 1 L)2 o.v _5.U 0 Vu 0vu v

3,4 us u .0' 1.w v .4 3 .24uv) u u u.vj ou v

A-9

I "IS PAQ IS "Sr WtlJI P"OCLC.Or*~u ~x r~ I~DVo

?- Il U 1i .o v .0 1 .' . #V)' v'.0

e c~ Q 01) vu0I v U 0 u U0"00 00

e3 ju U v.o l*.J u~o J111 Q~1 Q

3o I v 1.2 .Q0 lJ.Q0 11.oo d*1o O .00

3,f 1 1 .3 U .0') u~o 1 1 ) Q 4 . Q Uv 0

3-" ' u U 3~o 10 '. 11) ~o ~j U t.0)0

,41 1 e U UOI 1.U 13U z10 10) .o

41 e i.o 1.vo j Uo 13.) Q)o U.3 4.0 .

-41 1 .0'.) u 0 1) 3.00 1 ) .5.o U *v ~e 1 v 1.0 U.1 3.0 U40 4.0 v.o v0v

,4) 4 U 1.0 1.00 v30 110 1.0' UU) v Q D 3. t

J" 0 .00 1.0v O3" .00 d.v0 '.00 :3.00

4 .20 1 .0 u U .v 4 .0 U U')Q .0 U .k)

1 14 1 .o v 0 U U L ~.1 0 0.3 ". u 3 v

4 *-4 I.oo 13.00 ',.0' 'J :)') 4.uu :)'P

1 ' I U0 .1 1~t 1.0 13.0 1oj o3 "vO v.Q

v 1 o I U, 1.0 1.,' j t u 1o 1.J' :)

133 u o'1 1 0 D 1'.o Q Q u1) j.3 u ~ o 30

51 1 u j.3 2.0.0 w)3' 1.'' '0

'.3 . v o. ~ ~ 2o U 3.0 .3U -o v.oQ

56 1 0 j.0 u.o U 3o 3.z u.t -'. v

So ~ ~ ~ ~ ~ ~ ~ ' 1) 4 0.0 2.'Q30 A3) .0 -0)

v 17 .o j U 0.' Dtp3 . v :.')

5 o 1 1 U .00 .00u 3.0 o.I *ju ou %U D 0

50 v' 1 o.00 u.0 13.)v '1j Q.3 I .oV

51 1 o0.0 7o0 j3u 110' -.0 u .0 v .o)U

v 2 1 )~4 2 .00 3.00 ju 1.0 .00 U Vu I

n I e u .' o.1 .' 130 i.) v .v L.0)

n d.2 t.0' 0.)' .' o 4*o U 3.oO 1

d4 q -4. u u v t* / ' '0 QI Q 00 1) * 00 u v

u~ .3 e ..1 v.o 1 d0 Q o I..' ?'0 e3.0IV0v

-iF .1 -j.u 2.0 1o i.0 U.3 .0 v U uv

k)0 l.o ?-I. t) 1.ou . v0 S.3 ki u0')

11).; 1.0. 2.Q 13u v.1 Q.3 4U0 )

U Ij I~3 V.J 1.) 3.. ~~'U .0 -

4 A-10C

1 S A"

v uuu v l v .vU 'u u a v

1 u '. '.I Q.0 u. v i 4

u ID i.L o.J' u~u v.e 1. .

0~. 1 v~e I 1 v~j u .c d IuAJV/o~1 1.U v.U I3.vu e./i u .~ v v

11 It 1 u' v . v lu, 1 1.3 u w4 qV .5'U

71 1 u U.', e.)( li.uu 'I.iii 1.v'j *'.Vv *.

17 1 0 u ./ 7 v k 1..v u .~ 1. u u u u %

7' I i, un? c.v v .n tj -).00o v.L u vv ,U

hU 1 o u.7 '.U 1u u 1u u Vuu

01 ui i,.6 .uu o . v 11.ju 1u~ u Uv

n v . I ev ) c .v lu./ IoIJ v 0 U k

1 1 v V.1 Q.v .'u u 1u. 6 I~u vuvC5 1 v ui1 dv 1 u u ..5 u u.' 'u v

u- % u u. u~J I~U ~ .L v.' u

~ .V . .'6 .U 1~vJ ioj 4(f u u.~ u 1 uv '

I ; 7 o ()v O./ O.O v Uu vuv

01 w, u.d 2.i, C5U '..U U.4( t)u0v01 1 1I d. ~ .vU ~ . - 4.73u I . u u 4 v

I~ d 1. u*~ e u 1.4() 1 .u 'I vu .v u c.'

d 14 %) v.L cf.U 1 U uO~ u 3.J0 14 u v2" 1) u/ e- b7 u1u -)( 4iU .vUU ~ .yv

U4 .iju e*Uu oD.Vu "'.1U 1 .Vv 6.uu ~C: 1 1. e .U vu u v o. UQ I. vv - v 4 .

d v 1) ki3 i0l ~ .u 4. 7t) u - .vu -.

v )v ( .uvv4

9 31 4 vu C

.' -

Appendix 3

METR:C CONVERSION FACTORS

Feet to Meters

I foot 0.3048 meters

Thus:

3 to a-foot swells I-meter swells,

a I6-foot boat " a 5-meter boat, and

an altituce of 500 feet ' a 150 meter altitude.

2. Nautical M iles to Kilometers

1 nautical mile (nm) = 1.352 kilometers (Kin)

Thus:

10 nm visibility 1 18.5 Km visibility, and

a 2 nm range " 3.7 Km range.

3. Knots to Meters/second and Kilometers oer Hour

1 knot = 0.514A meters per second

1 knot z 1.352 kilometers per hour

Thus:

a 10-knot wind soeed 5-meter ver seconc wind soeed, and

a 10-knot search soeea 18 Ki.omezer per nour search sDeed.

2-

., °