I",,

aO) Technical Document 1635U.) August 1989

Effect of MarineIAtmosphere on

Performance ofElectrooptical Systems

J. H. RichterH. G. HughesM. R. Paulson

DTICELECTEDEC13.1989

• i~lB U

Approved for pubdic release; ditrbution Is unlltf d.

89 12 12 0,4

NAVAL OCEAN SYSTEMS CENTERSan Diego, California 92152-5000

E. G. SCHWEIZER, CAPT, USN R. M. HILLYERCommander Technical Director

ADMINISTRATIVE INFORMATION

This work was performed from 30 September 1988 to 30 September 1989 for the Office ofNavy Technology, Code 214.

Released by Under authority ofH. V. Hitney, Head J. H. Richter, HeadTropospheric Branch Ocean and Atmospheric

Sciences Division

SM

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE

REPORT DOCUMENTATION PAGEla. REPORT SECURITY CLASSIFICATION lb. RESTRICTIVE MARKINGS

UNCLASSIFIED2a. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION/AVAILABLITY OF REPORT

Approved for public release; distribution is unlimited.2b. DECLASSIFICATION/DOWNGRADING SCHEDULE

4. PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)

NOSC TD 1635

6a. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7. NAME OF MONITORING ORGANIZATION

Naval Ocean Systems Center

6c. ADDRESS (Ct. S1'ma&&7P os*) 7b. ADDRESS JC, SaWZIP Code)

San Diego, CA 92152-50008a. NAME OF FUNDING/SPONSORING ORGANIZATION 8b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT DENTIFICATION NUMBER

Office of Navy Technology Code 2148c. ADDRESS (CSmd27P4) 10. SOURCE OF FUNDING NUMBERS

PROGRAM ELEMENT NO. PROJECT NO. TASK NO. AGENCYACCESSION NO.

Arlington, VA 62435N RM35G80 N01A 540SXB3

11. TITLE (wrkjd S*c"W f adan)

EFFECT OF MARINE ATMOSPHERE ON PERFORMANCE OF ELECTROOPTICAL SYSTEMS

12. PERSONAL AUTHOR(S)

J. H. Richter, H. G. Hughes, M. R. Paulson13a. TYPE OF REPORT 13b. TIME COVERED 14. DATE OF REPORT (YbwUaDq) 16. PAGE COUNT

Final FROM 30 Sep 88 TO 30 Sep 89 August 1989 4516. SUPPLEMENTARY NOTATION

17. COSATI CODES 18. SUBJECT TERMS (CaW n warm fm7dm tn )

FIELD GROUP SUB-GROUP ;,optical propagation

IR radianceaerosols

19. ABSTRACT (Covtnwx ai e ina sswy ai let*4 hybxk ' vtr)

This report reviews the exploratory development effort in the US Navy to provide an atmospheric effects assessment capability forexisting and planned electrooptical (EO) systems. For many EO system applications, it is necessary to have an accurate knowledge ofmarine background radiances and to consider the effects of the intervening atmosphere. Accordingly, a capability was developed toestimate the apparent sea surface radiance for different sea states and meteorological conditions. Also, an empirical relationship wasdeveloped which directly relates apparent mean sea temperature to calculated mean sky temperature.

A careful investigation was conducted of lidar (light detection and ranging) techniques. It was concluded that single-ended,singie-frequency lijars cannot be used to infer slant-path extinction with an accuracy necessary to make meaningful performanceassessments. Other idar configurations may find limited application in model validation and research efforts. No technique hasemerged yet which could be considered ready tor shipboard implementation.

A shipboard real-time performance assessment system was developed and named PREOS (Performance and Range for EQSystems). PREOS has been incorporated into the Navy's Tactical Environmental Support System (TESS). The present versions ofPREOS is a first step in accomplishing the co.mplex task or real-time systems performance assessment. Improved target and backgroundmodels are under development and will be incorporated into TESS when tested and validated.

20. DISTRIBUTION/AVAILABILITY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION

[] UNCLASSIFIED/UNLIMITED ] SAME AS RPT DTIC USERS UNCLASSIFIED

22a. NAME OF RESPONSIBLE PERSON 22b. TELEPHONE (aoAmC00) 22c. OFFICE SYMBOL

J. H. Pichter (619) 553-3053 Code 54

DD-FORM 1473, 84 JAN 83 APR EDITION MAY BE USED UNTIL EXHAUSTED UNCLASSIFIEDALL OTHER EDITIONS ARE OBSOLlETE SECURITY CLASSFICATION OF IS PAGE

UNCLASSIFIED

SECURITY CLASSIFICATION OF~ THIS PAGE (Whn Do EbimW)

DD FORM 1473, 84 JAN UNCLASSIFIEDSECURITY CLA8SIEICATION OF~ THIS PAGE (WhwDWEAWu)

CONTENTS

IN -I R O I) t 'C I IO N ..... . . . . .. . . . . . . ......................................... I

\ .-\ R IN : A F R )S ( I. M ODIA N G ........................................ .................. 2

\I.\R IN F IN -R .\R [) B A( K G RO IND S ........ ........... ............................. II

\-RO S () S .N SIN G \V III+H [ ID A R S .....................................................

\SSFSS\I1FN I ()- S') SI F\ I'ERFO R\ .1 \NCF .......................................... 28

('() NC I t. S10 NS -\ND RFC( M M [ \ ),,\ I IO NS ...................... ................ 36

R I-I R E N C F S ... 3..7.. . ... ..... ........................................... ........... 3 7

Aooeessilao f

INTIS GWA&I

DTIC TABUmWzaounoed 0

By

Diatribut I on/

Availability CodesAvail and7/op

LA% Spec ibA

iii

INTRODUCTION

The V S Na\ currently uses or is planning to use a number of electrcoptical (EO)s\stens. [or example, forwkard-looking infrared (FI.IR) thermal imaging systems areCenrrcnt lv installed on many aircraft as an integral part of their weapon systems. Infraredscarch and track (IRSI ) systems are currently in a research and development stage. Heat-,CCINng or laser-guided w apons ha\e been used for a long time and new systems are underd' ptclopment. lo optimi/e the performa,,:ce of these EO systems in tactical situations, a wayto assess their performance is needed. For surxeillance systems. this means a way to predictthe s\stcm's ability to detect and correctly identify a target of interest: for weapon systems,aI \w.a to predict the system's ability to track and destroy the target.

An assessment of surxeillance and tracking equipment naturally requires a knowl-edgce of the limitations and capabilities of the operator and sensor hardware, but it alsorequires a knowvledge of the thermal characteristics of the target and its natural back-ground. and the transmissivit\ between the target and sensor. These latter parameters arerequired as inputs to computer algorithms w hich determine the ranges at which the equiv-alent blackhody temperature difference between the target at icro range and its back-gutind is degraded by the atmospheric transmittance to the minimum detectable or resolv-able ternperatures of the sensor.

Both target and background infrared signatures are controlled by the interveningatmosphere. For a surface ship, heating from the sun and internal sources, and cooling bythe ambient winds and ship's motion (as well as solar insulation effects) cause the ship'stemperature to \ary. The apparent temperature difference between the ship and its back-ground. as viewed by the sensor, will then depend upon the infrared transmittance andemission along the propagation path. To be of practical use in tactical situations, airborneI-l.IR s\stems must be able to detect and identify surface targets at ranges of several tens ofkilometers. This requirement limits the depression viewing angles to within a few degrees ofthe hori/ontal. As an example, the range to a target as viewed from an altitude of 1 kmk Ith a I-degree depression angle is 80 km. Depressing the viewing angle to 3 degrees placesthe target only 20 km away. For larger viewing angles, the infrared radiance of the s;easurtace is the primary contributor to the background scene. Closer to the horizon, however,the sky radiance reflections and emissions by the intervening atmosphere must be takeninto account. While a smooth sea is a poor emitter of infrared radiation at the small view-ing angles. emissions from the individual wave facets of a wind-ruffled sea contribute to thebackground scene.

[ he dexclopment of suitable target and background models to be used in computeralgorithms of system prtormance presents a formidable task and is needed by both usersand desiuers of [O systems. In 1978. the iuider Secretary of Defense for Research andLnugilcering tormulated goals to be followxed by the three services to address the primaryjactosr (e.c., aerosol extinction, gaseous absorption. refraction, and turbulence) affectinghe transmission and emission characteristics of the atmosphere which determine system

pertormance. Baicall\. these goals (as updated in 1982) are

,\'-,rzo clx model propagation environments of naturally occurring and man-nade aerosols and gases, including the effects of turbulence and multiple scat-tering. Relate the particulate constituents to measurable and predictable meteo-rological parameters.

Sl)e\clop atmospheric sensor systems for model xalidation. test, and operationalsupport. Special emphasis should be given to studies of the applicability of sens-ing bx light detecting and ranging (lidar).

0 Relate atmospheric effects to the performance of military systems, in opera-tional as well as in research and development activities, to include the develop-ment of tactical decision aids. This includes the consideration of transmission,background, and target factors which are based on climatologies or real-timeen ironmental inputs.

Since their inception, considerable progress has been made toward meeting thesegoals. The effects of gaseous absoi tions on radiance and transmittance have beenadequately addressed in the LOWTRAN 7 code developed at the Air force Geophysicall.ahorator. (Knci.s et al.. 1988). Since multiple scattering by aerosols has recently beenincluded in I.OW-IRAN 7 and turbulence has been addressed in the Army's Electro-Optical, sterris Atmospheric Effects library (EOSEAL) (Miller and Ricklin, 1987), they will nothc co\ ered here. In the following sections, examples will be presented of advances madeto\Nard meeting each of the goals, with primary emphasis on the measurement and model-ing of absorption and scattering by aerosols and :heir effects on radiance and transmit-tance, which determine system performance. It is the purpose of this document to summa-rn,,c past accomplishments, discuss present deficiencies, and define future directions for thea~scsmcnt of Navy EO system performance.

MARINE AEROSOL MODELING

Selectable aerosol size distribution models are now available in LOWTRAN 7 forcIlculating their scattering and absorption properties. Of primary interest here is the NavyMaritime Aerosol Model (NAM) (Gathman, 1983). This model was developed by using anextcnsive set of measurements made in the Eastern Pacific (San Nicolas Island) during1978. 1979. and 1980. These measurements were of broadband (visible through the farintrared) atmospheric transmission and meteorological parameters, aerosol size distribu-tions, and radon concentrations. This was a cooperative effort between Navv Laboratories(Naval Ocean Systems Center, Naval Research Laboratory, Naval Surface Warfare Center,and Naval Weapons Center). the Pacific Missile Test Center, and the Naval PostgraduateSchool. Shipboard measurements in the Atlantic Ocean (USNS Hayes, 1977, U'NS L.vnch,1983) of aerosol size distributions and meteorological parameters were also included in themnodel's dcevelopment.

Key to the development of NAM was the recognition of a trimodal structure injter/)sol size distributions plotted from the aforementioned field experiments. This trimodaldistribution ,Aas interpreted as a superposition of three aerosol distributions: the smallestdue to continental aerosols, an intermediate size dependent on the average wind speed, andthc larg&est goxerned by the instantaneous wind. Accordingly, a size distribution model (atradius r) is the sum of three log-normal distributions given by

i-(r) I4, cxp (n (cm .' um I)

r:2

I

hcrc

A4 2t000(,M) (2)

4, 5,866( F 2.2) (3)

06,,I I) o... (4

i III i i I I I I I I i2

Component .4 1represents the contribution by continental aerosols. A At is an air

mass parameter that is alloked to range between integer %alucs of I for open ocean and 10lor coastal areas and Is gi en by

A4.A = /NT( Rn 4) 1 (5)

Mhere Rn Is the measured atmospheric radon content expressed in pCi m . In the absenceof radon measurements. the air mass factor can be related to the elapsed time. T(da\s). forthe air mass to reach the point of observation:

-11: lN.T[9 exp( T 42 + I (6)

(omponcnts A i and -, represent equilibrium sea spray particles generated b\ the surface'\ind speed a eraged over 24 hours in m s) and the current surface v ind speed ( in

in si. respectively. In EL. I. r,. the modal radius for each component referenced to a rela-

ti\e humidity of 801 (r, = 0.03 p/m. r, = 0.24 gm. and r.= 2.0 pm), is allowed to gro ithrelati\e humidit\ (RH) according to the Fitzgerald ( 1975) formula:

FZ [(2 RH 100) 6(0 RH 100)]O (7)

The contribution to the total extinction or absorption b\ each component can then be\k ritten as

A) S F) {C, f Q,.(X.r.m) exp [ In -r r2 dr (8)r fr

%k here C,= 0.01-1 7/f)A,. The factor I in the expression for C, ensures a constant totalnumber of particles as the relative humidity increases. Q,.(A.r.m) is the cross-section foreither the extinction or absorption normalized to the geometrical cross-section of thespherical particle, and m is the complex refractive index, which is allowed to change fromthat of dry sea salt as the particle deliquesces with increasing humidity. LOWTRAN 6proides precalculated values in tabular form of the parameter oe.,A,) C, at discrete wAave-lengths for four relative humiditles (50"1, 85 *, 90"(. and 99c('). from which the averageextinction for a ,pecific \%avelength band and relatie humidity can be readily determinedbx interpolation. When an obserxed surface isibility (IIS,, is available as an input to themodel, the amplitudes, of the three components are adjusted by a scaling factor (SF) so thatthe calculated aerosol extinction coefficient. a,, at a xa\elength of 0.55 pm. is the same asthc obsercd extinction, a,, determined from the relationship

3.9 f21 S - (9)G, + a,

o.hcrc , is the Ra. leigh contribution to extinction at 0.55 Mm.

I his model has undergone extensive evaluation of the accuracy w ith w hich it canpredict infrared radiance and transmission. It should be noted that the current wind speedcomponent. A . is different from the value published in I.OW-FRAN 6 (Kneivs et al.,1983). xx hich was A 0.01527( V, 2.2). This modification was found to be necessary inorder to match published measurements of infrared (8 12 mm) sky radiances (Hughes.19X7) In that ,kork. it wxas determined that for Iov. wind speed, the measured and calcu-

!aicd radliances agreed "%ithin 2' at the optical horizon. Hoxxev er, during moderate wind

3

speed. the current xxind speed component of the aerosol model had to be lowered by factorsnear 0.05 for the calculated and measured radiances to agree. WheT' the model was used,the discrepancies betxaeen measured and calculated infrared radiances were found to beinsensitive to the 24-hour a.eraged wind component (Fig. I). Measurementj; by de Leeuw( 1986) of large particle size distributions (r > 5 microns) provided another method ofeCaluating the current wind speed component. In that work, size distributions weremeasured waith an impactor at different altitudes (0.2 to I I meters) above the sea surface. Inorder for the size distributions calculated with the model using measured surface meteoro-logical parameters to agree "ith the measured values, the current wind speed componenthad to be lowered by a factor near 0.07 (Fig. 2). Within the measurement accuracies ofboth the radiance and particle size comparisons, the reduction factors in the A3 component'.ere in reasonable agreement.

The accuracy to which the model can predict extinction coefficients has also beentested against the transmissometer and meteorological measurements at San Nicolas Island(Gathman and Ulfers. 1983). Good correlations were obtained between calculated andmeasured extinctions for near and midinfrared wavelengths. However, for the far infrared.the calculated extinctions were 20(' to 40C' greater than those measured by the transmis-someter. While these correlations may have been influenced by the uncorrected A compo-nent. they were shown to be sensitive to the selection of the air mass factor and to whetherthe visibility a'.as used as an i.pit.

The model was developed to be as representative as possible of different atmos-pheric conditions: ho'wever, it cannot be expected to exactly reproduce the optical proper-ties in a given location at any specific time. Neither of the techniques for selecting the airmass factor (radon count or air trajectory analysis) is currently available for shipboard use.Also the visibilities inferred from point-scattering devices onboard ship are most apt to becontaminated by ship effluences. A method is needed for selecting the input parameters sothat the model best represents a particular situation. A remote-sensing technique hasrecently been developed (Hughes and Jensen, 1988) whereby unique values of the air massfactor and visibility can be selected for different meteorological conditions. These valuesare inferred from [,OWTRAN calculations which agree with both the surface measure-ments of 8-12-pm horizon radiances and visible atmospheric optical depths determinedfrom satellite-detected upwelling solar radiances.

For this study, a Piper Navajo aircraft, equipped with Rosemount temperature andpressure probes and an EG&G dewpoint sensor, made vertical spirals over the ocean toobtain the profile of temperature. relative humidity, and pressure which are required inputsto the LOWTRAN 6 computer code for calculating the sky radiances. (A Barnes PRT-5radiation thermometer was also onboard the aircraft to measure sea surface temperaturesfrom low-level constant-altitude flights.) At the time the meteorological parameters kereobtained on 29 September and 25 November 1987. measurements of infrared (8-12 p.m)horizon radiances "ere also made with a calibrated thermal imaging system (AGA- HERMOVISION, model 780) having a 2.95-degree field-of-view lens. The thermal videoprocessor system (THER MOEKNIX) available with the AGA system displays the ther-mal scene on a computer screen in a format consisting of 128 pixel lines (0.023 deg pixelline). The response of the system is determined by placing a blackbody of known tempera-ture (t-O. I ' for temperatures <50'0 C) in front of the lens aperture. The digitized videosignal transfer function of the system then allows the blackbody temperature to be repro-duced to 'Aithin ±0.2 ('.

4

MEASURED RADIANCE (9 90.17 deg)3.0

.E V -= 2.2 m/s

LUo 2.5

E " .. V =6.Orm/s

Z

V 10 M/s

V 24-HR AVG WIND SPEED

2.0 I I2 4 6 8 10

V., CURRENT WIND SPEED (m/s)

Figure 1. Sensitivity of the LOWTRAN 6 NAM to surface wind speedparameters.

10-2

ORIGINALI MODEL

10- 3 _\

E a

E

10i-4 -\

ADJUSTED MODEL(SKY RADIANCE

MSMTS,SAN DIEGO, CA) "

-5 -

PARTICLE SIZEMSMTS, I

(NORTH ATLANTIC)

10-1.0 10 100

RADIUS (,um)

Figure 2. Comparisons ofmeasured particle size distri-butions with distributionscalculated with the originaland adjusted LOWTRAN 6NAM.

5

1-Or 0hsc the,,aSulrcmnHts, the scalllcr was located at an elevation of 33 meters onthe Point I oma peninsula in San )iego and was directed due west over the ocean so thata ppioxlinatelr hall the field-o-xiexk was above and half below the horizon. The dataprocessing softxkarc of the A( \ systcm also allows the the effectixe biackbodv temperatureo each pi\el in the scene to he determined. On each of the days. the pixel corresponding tothe Iaximurn temperature in the thermogram (20.4'C on 29 September and 16.5°C on 25\0\ ember) "as taken to coincide wkith the infrared horizon. From the current and 24-houra era gcd xi nd speeds ( 1 = 7.5 m s and F = 4.4 m s for 29 September. and V. = 2.5 m s,id 3 q m lot 25 Noxermberi measured on shore at the AGA site, and the verticalpitlilcs ot meteorological para meters. I ()\ I RAN 6 calculations were made to agree \kithtlk txnum pi\cl radiance in the scene b\ using nonunique combinations of air massaetn, a ,d \ sihilitic,.

In these calculations, the tnecorological profiles were di\ ided into 33 laers, asA!!,,%cd h% I (hx I R \N 6. 1ti lo\cr lay e, oft the profiles are also divided into sublaver,,

on'l1ta1nin1 the samc amonnt of absorhing ,nd scattering material and the same tempera-lUre a's the orieiial la.er. Ihis artificial la.cring has been found necessary (Wollenxxeber.S19,St ) to renioxc the anomalous dip ( II ughes. 1987) which occurs when aerosols are1neluded in the I ROVIAN 6 radiance calculations for icnith angles close to 90 degrees.

n c the \( -A scanner could not be accurately plumbed. the ,enith angle of the infraredir,,on tile pixel lile ci rresponding to the maximum radiance) was taken to be one-half

ol a pixcl line less than the angle at \xhlch the 1O WRAN calculations indicated thetcliacteo ra\ patl first struck the carth (() 9=0.179 degrees on 29 September and 90.174

deUrc (n 25 Noxenber).

In I-ig. 3, the solid lines represent the loci of points which allow the LOWTRANCaleIiations to match the Measured hori/on pixel radiances for different combinations ofair mass factor and \is hilit v. Ilhis featunrc of the calculations results from the visibility seal-ing factor of the si/c distribution remaining nearl constant for any appropriate combina-tion o! the t\\ 1) factors and from the relatixe inscnsitivity of the calculated extinction coeffi-cicnts tot the far infrared xxa\clcngths to the air mass factor. A\t the time of the IR radiancemcasurcmncrts, upxelling solar radiances "xere detected by Channel I (0.58 0.65 ,im) of the\\ I R R on hoard the NOAA.\-9 satellitc as it passed close to San Diego. These radiances.used to detcrmine the total atmospheric optical depths r,,). were themselves determined byusin Iic SAlt I(Science Applications International Corp.) satellite radiance computercde. \hich makes use of a direct linear relationship betxxcen upwelling radiance and thetotal atios phicric optical depth ((iriggs, 1975). Although the vertical structure of themctcorglogical parameters xx hich control aerosol growth xere different, the total opticaldepth, i t- the t xo dax s f x crc ricarl\ identical. I i a manner similar to the calculations ofltrlrcd radliane. calcatiiis of total atnospheric optical depth weic made to agree withthe naca,,urcmcnts niadc on both da\s. In calculating the optical depths, it was assumedthat all the acro,,ols x\crc confined to the mixed boundary laver as determined by theinicralt tlights. I lhc calculations of optical depths with different combinations of air massletor and isi hi lit are also sho. n in 1-ig. 3 as the dashed lines and are seen to bec,,trcnel ,ens iti c to the air nass factor. I he intersections of the solid and dashed linesthen determine the best coinbinations of air mass factor and visibil it\ to he used in thehaikgiournd calculations. I he inset in the figure shows the good agreement between theinterrcd siilities and those calculated h\ using aerosol si/c distributions measured at theInxxcst I:x els of the aircraft Ilights. I lhcsc \isihilitics seem reasonable. since the los Coro-fiados Islands. located belkcecn 25 and 30 ktii off the coast of San Diego. were not visible tothe naked cxc on either dax.

6

11/25/87 (T0 0.169) 9/29/87 fro 0.168)

I I - - - NOAA-9 Ch 1 (0.58-0.68 p~m)

9 HORIZON RADIANCE (8-12 pAm)

1 I I MET RANGE (kin)

~ IDATE AM INFERRED MES*UI I9294-5 26 25

on11/254-5 19 202 ~SIZE 0ISTRIBUTION MEASURED

3 9./29 (T74 , 204 C)

0 10 20 30 40 50 60 70

VISIBILITY (kin)

Figure 3. Loci of points of LOWTRAN 6 calculations with differentcombinations of air mass factors and visibilities which match measuredvalues of IR horizon radiances (solid lines) and satellite-detected visibleoptical depths (dashed lines) for 29 September and 25 November 1987.

\\hi A \I initial]\ developed by using surface measurements of aerosol~icdi~tuihtmnus and nicteoroloeical parameters, it is allowed to vary with altitude in

I (A) \ I RAN 0 accord inc to rclati\ e hum id ity Up LO an altitude of 2 km. at which point theIipwplieric \l (dcl is, Introduced. A dynamic mnodel using NA M as the surface kernel (the

\ax 0~(ccail \ Crtical Acrosol 1Model (NOVAMN/)) has recently been introduced (de L-eeuw

S9s'9 ) to het tCr account for thle vertical variations in site distribution. It was dev-el-pc[) ' 110IVI ou I acoo)perat i' effort betxwe-..! the Naval Research L aboratory, the Naval

l'ut ead ateSchool. and thle Naval Ocean Systenm, Center- (G(athmnan, 1989). The Phy-sic~s

,hlI lctronlics, laboratory% INM), [he Netherlands, recenitly joined in the evaluation anduc c ltort We I ccu\k ct aL. 19891. 1ie model is based onl the physical processes

fltcigthC pr-oduction. miintg, deposition. and(- si/c of' thle aerosols within the marine,ltwo~splcic. \crosols, with similar origins are represented by separate log-normal si/c

dtihutions. I lic rc optical effect p roduneed is the result of thle superposition of all theAc* tIo ,iouiip,. I o dcterm inte the aerosol si/c d ist ri but ion at any pa rticula r level, one of a

1 4 Ml\IL, protic mlodcls is tise' Vile ,,election process. indicated by the flo\x% chart inIi 4. 1,, dcici tnncd h\ thc meteorological Input data available and the wavelength at

khichl calculatlonl" arc to he madc. In addition to the general profiles of' temperature and2uukl\ I,, dctcru11111d hx radiosondes. tile Inputs, rcquired by NOVAM include the

., .ncst ol 'a rkace ohscr atilolls:

)m24 ' a \craged xNi rd speed (I m s)

' \I[ masactor (II... 10)

*( 1(0(1 a crc ( tenths)

* (Thud r" PC I . .. 1(11

7

START

DASHED LINES REFER ENTER LOCATION -- - - . I

TO INPUT DATA FLOW IDURING CALCULATION.

CALCULATE DEFAULT] NOTE THAT THE PROFILESOLID LINES REFER TO I SURFACE VALUES DATA FILE MUST CONTAINPROGRAM LOGIC FLOW I AT LEAST THE SURFACE

OBSERVATION

INPUT SURFACE . . . IOBSERVATIONS

MEASUREDPROFILE

IDATASTHERE YE

PRECS PITATION? I D Ai

+ ,C,. N'T SEMI-ISUBSTIT TE DEFAULT '"0 F N AUTOMATIC

PRMOFILE ROIL

VALUES FOR t MISSING I\" RFL

SURFACE MEAS-UREMENTSI CASE ANALYSIS-EFUL SRAUSNGRAIN

CALCULATE SURFACE rCONCENTRATION I

WITH NAM ANIS

/GENERATE DEFAULT --- YE \F. I FLL RH PROF"LER POFILE DATA ES RH PROFILE ST W DATA ATI FOR EVERY 100 FT I NO ALL LEEL ,SE TOF

DFigue STABLE MODEL NOT oNIS BLFILE SUPPORTED YET STABLE?

( DEEP CONVECTION MODEL IS BIL?

DEFAULT STRATUS NOTYCONC. SUPPORTED YET WS> =5 m/s|

PROFILEGENERATOR / CLU

WS < 5 m/s/ .. COVER (CC)( ~ YES SWALEENCT (W}I > 0 SELECT PATH CC<02

WL< I OH WL> I1 I WIND ONBSSOSPEEDCLOUD COVER+NO

SSELECT ON NONSTRATUS CLOUDS

EXTINCTION AND CODTPABSORPTION

CALCULATIONS [ WA .DONE FOR THE WEKONLY ONE YES

AEROSOL SIZE DIST I COVCTO INVERSION /

AT EACH LEVEL MOELDEELEDIN PROFILE '

EMD

-EIT"

Figure 4. Flow chart for NOVAM.

• .. a I I I I 8

• 10.6-pm extinction ,:ocfficiCnt i!:m I)

* Present ,%eather in standard code (0... 99)

* Height of lowest clouds (tm)

" Zonal scas,-,nal categor (I ... 6)

* Sea surface tempcrature (°C)

• Air temperature (OC)

" RelatiNe humidit\ (K')

" Optical \isibilit\ (kin)

* Local wind speed (m s)

I he product of NOVAM is primarilk a file of the extinction and absorption coeffi-cen,, at \ arious lexcls in the marine atmosphere. While NAM has undergone numerousCx a luatioM, and considerable improvement through the use of transmission and skyradiance measurements. NOVAM has not \et received a critical evaluation. Such evalua-tion under xxell-defined environmenta conditions is needed before it can be incorporatedinto LOI,'TRAN. Data for preliminary evaluations of NOVAM were obtained frommeasurements taken on the upwAind side of San Nicolas Island on July 18. 1987, and havebe.:n presented elsewkhere by de leeuw et al. (1989). Figure 5 shows examples of extinctionpr 4iles at both visiblc and infrared %\avelengths bands. The mceorological input data wereobaimcd frorn a tethered balloon platform on which a nephelometer and a particle measur-Ing system (PM S) particle spectrometer (Knollenberg) were located. Extinction at 0.55 Mmi, obtained directly from the nephelometer, whereas extinction at different wavelengthsmax be calculated from the si/c distributions obtained from the PMS. Figure 5a is theextinction piofile for 0.55 pm measured with the nephelometer and calculated from themcaurcd size distributions and from the NOVAM model. The data show considerable,catter in the measured extinctions with altitude. The NOVAM prediction is within theenxelope of the scatter better than 7511 of the time for this particular case.

Figure 5bsho\%s the comparisons between the extinction coefficients for the 3 5-pmhand calculted by u sing the model and the measured size distributions. A similar compari-,on lfr the X I 2-Lni band calculated from the measured size distributions and the NOVAMprediction it 10.6 pni is shown in Fig. 5c. The present version of NOVAM underestimatesthe extinction in the infrared hands in the region above the mixed boundary layer. Thisma\ he a resulIt of larger particles from the sea surface being mixed into the atmosphereaho\c the apparent temperature inversion through the process of entrainment.

Additiinal ckaILua tiOns of NOVAM hake also been undertaken by using airbornedata acqluircd in the Southern California area. lie results indicate that predicted andlcasur.. extinction coefficient profiles have similar structure but may differ in absolute<l. I lie agereements are stronglv dependent upon the proper choices of air mass factors

a rd the scaling to xisibilit\. Dcfault profiles of humidit\ and extinction do not in generalIrp rcelc l lhC le.'aslurcd \alucs.

9

1000 1I00# NEPHELOMETER DATA 0 KNOLLENBERG DATA

* KNOLLENBERG DATA

800 -- NOVAM PREDICTION 800 - NOVAM PREDICTION

*10 90

S600

0 0 0SE- 6oo - #

S. 400400 -

200 200

FLIGHT 1814 * FLIGHT 1814

0 I 0 1 - I

-3 -2 -1 010 - 10- 10- 3 10- 10- 1 10LOG 10 EXTINCTION (AT 0.55 pm) EXTINCTION AT 3-5 .m (1/kin)

(a) (b)

1000

0 KNOLLENBERG DATA

800 -- NOVAM PREDICTION

400 0

200 -0 0

FLIGHT 1814 •0 1

10- 5

10- 4

10- 3

10- 2

10- 1

1 10

EXTINCTION AT 8-12,um (1/km)

M

Figure 5. Comparison between extinction profiles predicted by NOVAM and experimentalextinction profiles derived from nephelometer data or from particle size distributions usingMie theory (Knollenberg data). The measurements were made from a tethered balloon.(a) Ccmparison of the NOVAM prediction for a wavelength of 0.55 Urn with a Mie calculatedprofile at 0.55 tim and the extinction profile derived from the nephelometer. (b) Comparisonof the NOVAM prediction for the 3-5-tim band. (c) Comparison of the extinction profile aspredicted by OVAM for 10.6/prm with the Mie calculated average extinction profile in the8-12-p~m band.

10

. . .. . m • 200

MARINE INFRARED BACKCROUNDS

V, as demon1t ratedI earl1ier, theC IOW] RAN codes have proven to he a \ ersatileon II pied let ng atrimosphecric radianrce a hove the hori.'on. 'Ihey do not vet, however.nclude the eapahiit- ofesilmating the apparent radiance of the sea surface for different

s'taC1 aC id nicteorolooical cond ition,,. A model has recent lv been developed of- the effec-tt~ r adia nce of thle ,ea suirface as a function of the \ ewing /enith angle and sensor height\\ ollenx eher, 198b) arid has been Incorporated into LOW] RAN 0 for processing on an

H111-90I2I0 computcr. IiI addition to the path emissions. this version uses the Gaussian-d I rtihuted v. a~ slope model of Cox and Mu Lnk (1)954). which is based on surface winmdsIpeed ifnd dkller:Cion reL:;iti'c to the look angle. Cuitctitly limited to NA.M. the model basi-call \ en letlates thle total contribhut ions to radiance at the sensor from the emissions of the

I ner\cringatnos.plic arid tilie sky reflections and emissions froni the w\ave slope surfaces.

icrs6 and 7 slioX thle coripari'ons of the measured and calculated infraredr adiances for ,eith angles \% Ithin about I dlegree below, the hori/on for the air mass factorsa id x isihilit ics I-or the 2 days In Fig. 3 (Hughes. 1989a). InI boith cases, the ma jor contribu-or to thle total rad ianrce just helo%% the hori/on is the path em issioin (&), While the

recflected sky radiance .\~ ~ anid (the surface emission .V&)are small, their contributionto the total radiance N(O(),, at this loxN level (33 m) of observation cannot be neglected. ItI" Iriteresi Ig to niote onl\ a small re\ ersal oif the relative nimagriitudes of' the reflected sky

AA. = 8-12 pm H = 33 m

11/25/87

HORIZON V0 2.5 m/s, V =3.9 rn/s

* MEASUREMENTS

3 0

E

W N(O

1N(6)1sk -

I-90.0 90.5 91.0

ZENITH ANGLE (deq)

Figure 6. Comparison of the measured andcalculated IR radiances for zenith angles aboutI degree below the horizon by using the airmass factors and visibilities determined for 25November 1987.

4

HORIZON A = 8-12 mn H= 33m

MEASUREMENTS

3 N(O)t

EN()~

2Uw0

9/29/87

0

5V c = 7.5 m/s, V= 4.4 m/s

. -.- .- -N- - -

0 -"¢

90.0 90.5 91.0ZENITH ANGLE (deg)

Figure 7. Comparison of the measured andcalculated IR radiances for zenith angles about1 degree below the horizon by using the airmass factors and visibilities determined for 29September 1987.

radiances and surface emissions between the two sets of calculations, which demonstratesthe small influence of the wave slopes for the moderate wind speeds on 29 September. Boththe calculated and measured total radiances on both days are in good agreement, whichplaces confidence in the model's usefulness in radiance calculations at other altitudes and/cnith angle!..

The selected aerosol models were used to calculate the contributions of the path,,ca. and reflected sky radiances to the total background radiance as a function of altitudeand ,enith angle. In Fig. 8 and 9. examples are presented of calculations for both days witha sensor altitude of 1000 meters. For zenith angles less than about 95 degrees, the majorcontribution to the background is the path emission, with the reflected sky radiance beingcvs than 101'r of the total in both cases. In Fig. 10. the resulting apparent blackbody

tcmperature of the sea versus zenith angle is compared for both days. Again, the higherapparent temperature for zenith angles near the horizon on 29 September results from thepath emission from the warmer elevated layers. The rapid fall-off of path emission withincreasing ,enith angle (i.e.. shorter slant paths to earth) is the cause of the decrease in

12

4.0H = 1000 m

11/25/873.5 -

3.0 SEA

,. 2.5 PATH

S- 2.0

z 1.5

1.0/

0.5 // fREFLECTED SKY

0.0 ......0 .0 , r I . . I ... . . .. I I I I I

80 90 100 110 120 130 140 150 160 170 180ZENITH ANGLE (deg)

Figure 8. Calculations of contributions of the path, sea, andreflected sky radiances to the total background radiance as afunction of zenith angle as viewed from an altitude of 1000meters on 25 November 1987.

4.0

35 - 9/29/87 H 1000 m

3.0 PATH SEA

-c-2.5

E2.0 - /

/LU/U 1.5 -z '< /

1.0 - I

0,5 II ." REFLECTED SKY

0.0 1 . ........ I . I I I i

80 90 100 110 120 130 140 150 160 170 180

ZENITH ANGLE (deg)

Figure 9. Calculations of contributions of the path, sea, andreflected sky radiances to the total background radiance as afunction of zenith angle as viewed from an altitude of 1000meters on 29 September 1987.

13

28H= 1000 m

26

24

22 9/29/87

w L 20 ,-ILLcr

18

w .........................................................a 16

- 14 - •- •" ./25/87

12

10

8 I I I I I I I I80 90 100 110 120 130 140 150 160 170 180

ZENITH ANGLE (degree)

Figure 10. Calculations of the apparent blackbody tempera-ture of the sea versus zenith angle as viewed from an altitudeof 1000 meters on 29 September and 25 November 1987.

apparent temperature on this day. In contrast, the temperature increase on 25 Novemberis the result of the increase in emission from the sea surface (which was warmer than theair) with increasing zenith angles. The relative contributions of the three components willof course change for other altitudes. As the nadir zenith angle is approached, the appar-ent temperatures at this altitude do not reach the measured sea surface temperaturesST,, = 15.2°C on 29 September and T,, = 17.6°C on 25 November) because of the pathemission contributions at this altitude.

In a practical sense, the sea radiance model needs to be improved to reducecomputer running time. In its present form, approximately 3 minutes of computationaltime on the HP-9020 is required to calculate the sea radiance at one zenith angle of viewingas compared to approximately 5 seconds for the sky radiance. Approximately 36 minutesof computer time is required to calculate the mean sea temperature within I degree of thehorion. TIhe major contributor to the computational time is the large number of skyradiance reflections and surface emission calculations that must be made to account for theradiance from all the wave facets. Computational time could be saved if the surface emis-,i itv and rcflectivity calculations were made at a single representative wavelength insteadot having to average over a wavelength band for each wave slope. Another way to reducethe computational time is to develop an empirical relationship to directly relate the appar-ent mean sea temperature to the calculated mean sky temperature.

14

To inxestigate this possibility, the I*H [ R MOIEKNIX system was used to deter-mine the mean eqtui.alent blackbod\ temperatures corresponding to an area I degree aboveand I degree bclok the hori/on in thermograms taken during different surface wind speedconditions ( Hughes, 1989b). For this study. 18 thermograms recorded during differentiniteorological and \ind speed conditions in 1987 and 1988 "~ere used to compare theInca n sk\ and sea temperatures,. In 1 2 Of the cases, tile actual sea surface temperaturereasured bv the aircraft radiation thermometer was available. For the full data set. themean temperatures were well correlated (coefficient of correlation. r 0.92), as showxn in

uig. I1, w, ith the sea tempciatuics being ,less than the sky temperatures. as indicated b\ theashed line for one-to-one correspondence. The mean temperatures differed the mostdi ng low xind speed conditions.-1 hese differences were found to decrease w, ith increasing%ind. Inherent in the observations are the actual sea surface tem peratures and wind speeds.It these parameters are included in a multiple regression analhsis for three independent%ariables. the follo\ ing relationship is obtained for the mean sea surface temperature.T-,(sca), in terms of the mean sky temperature. T , sk\ ). the current wind speed (I '). andthe acttual sea surtacC temperature ( T,, ):

T ,,,esa) 1.09 ,,(sl.y) + 0.37 1 + 0.24 T,, 10.5 (10)

[he correlation coefficient for the multiple regression analysis is 0.97. In this case. only12 samples \ ere used in the analysis (as compared to 18 samples in the linear regressionanalvsis shok n in Fig. I I ). x\hich indicates a definite improvement in the correlation whenthe actual sea surface temperat ores and ,kind speeds are included.

22.0

20.0 -

18.0 0

16.0

0,14.0 _

12.0 "

10.0 -

8.0 - 080 / (SEA) = 0.97 T., (SKY) - 1.94

60 (r 0.92)

4.0 1 1 1 " 1 140 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0

Tm (SKY), "C

Figure 11. Regression analysis comparison of the measuredmean sky temperatures, T, (sky), and mean sea temperatures,T, (sea). The dashed line corresponds to a one-to-one corre-spondence between the two parameters.

15

Earlier work (Hughes and Jensen, 1988) has demonstrated that the sky radiancesmeasured at individual pixel lines within about I degree of the horizon can be modeled%ery closely by using the modified LOWTRAN 6 code. Based on this, the replacement ofT",,(sky) by a calculated value of the mean sky temperature, T,.(sky), would appear justi-fied. This would allow the mean sea temperature to be directly determined from Eq. 10and measured profiles of meteorological parameters. Additional data are needed to verifythat such a relationship for predicting the mean apparent temperature of the sea is mean-ingful for other meteorological conditions sea states, and altitudes of observation. Sucha relationship would be a valuable inclusion into a system designer's handbook as a toolfor predicting infrared sea backgrounds in locales where only meteorological data areavailable.

AEROSOL SENSING WITH LIDARS

Measurements of slant-path visibilities are required as inputs to the LOWTRANcodes for scaling the selected size distributions with altitude. Lidar systems, which measurethe radiation backscattered to a receiver by aerosols at different ranges within the beam ofa laser pulse. have received considerable attention in recent years for this purpose. Thetechnique of inverting a single-ended lidar return to obtain range-dependc,.i atmosphericextinction coefficients requires an assumption concerning the relationship between thevolumetric backscatter and extinction coefficients. Where the integration is performed inthe forward direction (i.e., away from the receiver), the extinction coefficient a(r,) out tothe range where the transmitter and receiver field-of-view overlap, ro, must be known. Ifthe integration is performed in the reverse direction (toward the receiver), the extinctioncoefficient at the final range, a(rl), must be known. Ferguson and Stephens (1983) used anovel iterative scheme in an attempt to select the value of a(rf). The value of u(rf ) at aclose-in range (where the returned signal is well above the system noise) was varied untilthe calculated extinction coefficients as a function of range a(r) allowed the calculated andmeasured lidar returns with range to agree. The chosen value of a(rf) was then used as a(r o)to integrate out from the transmitter. This technique. however, is limited to situationswhere the ratio of backscatter to extinction is a known constant along the path.

Solutions to the single-scatter lidar equation have been presented for reverse(Klett. 1985) and forwNard (Bissonnette, 1986) integration where the relationship betweenthe backscatter and extinction coefficients is assumed to vary with range according to

0(r) = C(r)o(r) (II)

,A here k is a constant. For forward integration, the extinction coefficient as a function ofrange is given by

a(r) [I C(r)] exp [S(r)] (12)2 f f I C(r)] exp [S(r')]dr'

C tr,,) (r,, r,,

and for reverse integration by

[I C(r)] exp [S(r)] (13)o(r) : (13)

Cxp [A(r/)] + 2 f rj[I C(r)] exp [S(r')ldr'+2rC(r1)o(r1 ) r

16

here the constant k has been chosen to be unity. While these solutions allow for variablebackscatter and extinction coefficients, their usefulness requires an a priori knowledge ofthe ratio (Nr) as a function of range. Salemink et al. (1984) determined values of o and /3from hori/ontal lidar shots b\ using the slope method when the atmosphere appeared to behori/onrall\ homogeneous. lhev then presented a parameteri/ation between values of , oand relatieC humidity ( 7_- RH <= 87(' ). When the parameterization was used to invertv isiblc .a\clcngth lidar returns in the vertical direction, the derived extinction coefficientprofiles ( using radiosonde measurements of relative humidity) sometimes agreed reasonably\. ell 'ith those measured by aircraft mounted extinction meters. In contrast, de Leeuw etal. ( 196) using similar types of lidar measurements did not observe a distinct statisticalrelationship betw een backscatter and extinction ratios and relative humidity. Fitzgerald( 1984) pointed out that other factors, such as the aerosol properties. can strongly affect therelationship betw een /3 a and relative humidity and that the power law relationship ofEq. II is not necessarily valid for relative humidities less than about 80('. A unique rela-tionship bct ecen C(r) and relati\c humidity \, hich is dependent on the air mass characteris-tic. is 1, ct to be developed.

A lidar inversion algorithm using a double-ended lidar technique recently developedindependently by two researchers (Paulson, 1986: Kunz. 1987). In this technique. theassumption concerning the relationship between the backscatter and extinction coefficientsis eliminated by comparing the powers returned from a volume common to each of the twolidars located at opposite ends of the propagation path. However, the receiver gain of bothlidars must be accurately known.

The mathematical formulation of the double-ended technique is as follows: Considertwo lidars separated by a distance d. If we assign the origin of the propagation path to belocated at lidar I. the range-compensated power, S(r), received by lidar I from a volume atrange r is determined by the single-scatter lidar equation to be

S(r)1 = In K, + In [/3(r)] 2 r a(r')dr' (14)0

and that received by lidar 2 is

S(r), = In K, + In [O(r)] 2 f a(r')dr' (15)r

Where K, and K, are the instrumentation constants for each of the lidars, and a(r) and/3(r) are the x olumetric extinction and bacscatter coefficients, respectively. If the scatteringparticles are assumed to be spherical, the backscatter coefficients are eliminated bysubtracting Eq. 15 from Eq. 14. and we are left with

S(r)1 S(r), = In (K, K,) 2 f r(r')dr' + 2 fd (r')dr' (16)O) r

Since

f olr zf o(r')dr' -f a(r')dr' (17)r 0 o

17

Equation 16 becomes

S(r)I S(r)2 In ( K I K2 ) 4 f a(r')dr' + 2 f o(r')dr' (18)() r

Taking the deri'ative of Eq. 18 with respect to range, we obtain

I (Io(r) - - [S(r1 ) S(r2)] (19)

4 dr

The determined values of oa(r) can then be used with the system constants in either Eq. 14or Eq. 15 to determine the associated backscatter coefficients.

The double-ended technique has been used recently by Hughes and Paulson (1988)to examine the effects of spatial inhomogeneities on the single-ended lidar inversion algo-rithms. Measurements were made with two visioceilometer (Lindberg et al., 1984) lidarsover a 0.9825-km slant path near the ocean on the Point Loma Peninsula in San Diego,California. Lidar I and lidar 2 were located approximately 38 meters and 135 meters abovemean sea level, respectively. Each lidar was pointed about 3 or 4 meters away from theother's receiver, and the firings were offset in time by about I second to avoid amplifiersaturations and signal contaminations. The data presented here are samples of two returnstaken on 27 October 1986 during a period of reduced visibility and when conditions alongthe propagation path were observed to be varying both spatially and with time.

In Fig. 12a. the S(r) values calculated from the individual lidar returns are shown.In each case the ranges are referenced to the location of lidar 1. Similarities in the largeratmospheric irregularities are evident in each data set. However, there are differences in thefine structure of the individual S(r) curves which may be related to the slightly offsetsampling volumes and firing times of the two lidars, as well as to the different backgroundscenes viewed by each lidar. However, the background only adds a constant value to theobserved backscatter signals. This contribution was found to be small in all cases, since thesignal-to-noise ratio for both lidars exceeded 10 dB at a range of 800 meters. In any event,data-smoothing was necessary, and an I I-point (82.5-meter) running average of each S(r)curve was determined before taking the differences. The differences in the smoothed S(r)curves for each data set are then shown in Fig. 12b. The derivatives of the S(r) differencecurves were calculated with a running range interval of 15-points (112.5 meters), and thevalues of a(r), calculated from Eq. 19, are shown in Fig. 13a.

The corresponding backscatter coefficients, P(r), are shown in Fig. 13 b. They weredetermined by using the calculated values of a(r) with the appropriate S(r) values and systemconstants in either Eq. 14 or 15. The backscatter coefficients determined for lidar I andlidar 2 are in good agreement. The slight differences occurring within a range of 450 metersmay reflect the precision with which both systems could be calibrated. Interestingly, thehackscatter coefficients do not show the striking fluctuations of the extinction coefficients,which leads to the conclusion that C(r) was not constant along the path. The ratios of thecalculated extinction coefficient profiles and the backscatter coefficient profiles for lidar 2,((r, were determined (assuming the value of k in Eq. I I to be unity) and are presented inFig. 13c.

18

-2 ___________________________________

-3

LIDAR 2-4

-7

-8

-9-0.1 0 0.1 0.3 0.5 0.7 0.9 1.1

RANGE FROM LIDAR 1 (km)

(a)

2

co) -1

-2-

-3 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

0.1 0.3 0.5 0.7 0.9

RANGE FROM LIDAR 1 (kin)

(b)

Figure 12. Comparison of range-compensated power return, S(r), for(a) lidar 1 and 2 and (b) their differences in range from lidar 1.

19

2.0

Z 10

0

(a)

306-

LIDAR 1004

&0.02

(b)

02

Q-0.1

0______________________________________________0.1 0.3 0.5 0.7 0.9

RANGE FROM LIDAR 1 (kin)

(C)

Figure 13. Comparison of calculated (a) extinction coefficient ar(r), (b)

backscatter coefficient 0(r), and (c) backscatter/extinction ratio C(r) vs.range from lidar 1.

20

E-Xt nct ion c(CITficicn ts xocre calculated as l e i _ton of- ranp.l ior for~xard and

IrC\ crsI integratlonl h% using, the nrd% i'id ual I idar ret urns, and xk crc then compared wvith

hose dctc; iined bo, the: dlo uhb-ended tech niqL.C \ hen ( ) ats assumed to be cit her a

ost)Iitn itorII ilo \ cd t o \ ar\ \% ith range according to Fig. 1 3c. 1 he results of the double-1ne lidat techniq Ile pro ide thle close-in OF tlr-,a\~a boundar\ % alueCS for each lidar.

rcI 21. IiC 14 s t he cormpa rison,, for lor\x ard I ntegration determinedi %%hen C( r) isassumeIId too be constant Oscr- thce propagation path. I here is little or no agreement between

he ctI nekt iolns deter ined b\ cit her lidlar anrd t Iiose of thle dlou le-ended nicasuremnrt,,

ti LIotI) thle boundai\ \ aloeCs \wrC SpCCIliLd. Near at range of'600 meters. where the

01u ot0r) 'rierCaseC ( Fig. I 3d. the extinctions by liar I exhibit the well-known "insta-

buttIs'II in that tnle\ tenId to inIcreaSe \oithlout bound. Thle extinctions hy lidar 2 also exhibit

thle tisiahilitio:s h\ lending to /ero at the farther rant'es. I hiese data demonstrate tile sen. i-_

It:\ ot the Iinshbilitic' to thle rnlaeitLudcs I thle boundar\ values.

I cureI_ 14h Nho\\ s thle comparisons lor re\erse initegration Mihen (Or) is assumie(]he constan1t. Altitlilllh the soIlutions are "NitahleY there Is little agrcementt between tile

k10Iihle-entdAd meaCsurements arnd the extinctions deterineid b\ either Wiar.

It the ninesN of Or) are allos\Cd to %arv\ "xith ranige. as shox~ r in Fig. 1 3c, excellent

a eree m iiiae O bta ined bet xo cci the calculated extirnet ion eoefficierits oif lila rs I anrd 2.

1I lke extinction coeCtlcient profile" tor both the forw ardl Fig. 1 5a) arid reverse (Fig. 1 5h)n1tellfiti~ ions ar r1ial\ Identical v itli thle corresponding douible-ended measurements shox~ n

I heSe resltIs dem~lonstrate that If thle x a!Lu of' Cl r) varies with rane, but is assumedt)ho. t corista nt. rieither thle single-endled forx ard or reverse integration algorithms will

io range-dependenit extinction coeffieients to be determined with an\ assured degree of.icic ecrin it the initial bounrdary \alu es are speeified. If. however, the manner in which( r) a ries i specified., both the forw ard arid rceerse siritle-ended inversions for this data

'Ct repr-od nec, thle dituhle-crnded nieasuremerits remiarkably el Whether the same is truetIM Wthert OLailtsl iiiC(15 to be determined.

It the condliltionsl under M~liiel thle fork~ardf ii\e.,ion algorithm is stable can beestIa bushed, then a striLde-ended Wiar mx ers.,ion technique \\ tild be possible when augmented

ith itsetiriauetn of, extinction arnd measuremients to relate C(r) toi air mass

chiaaCteIsINt Ics ait d relaI iti tII Itv\.

%\ title the v' orks ol \1 ulders. 1 1984) arid de I ceu\N et al. I 1986) have concluded thati relitiship exit betx~en C~n arid reClati\e liuniiiditv. their nieasurenerits did not

atcemii t t ir e hiingrice i the aIir mass characteristics,. Whlet her or not Such a relationship can

o.IL ehob identtilied in) a practical sense Is vet to be deterined.

Ini sItililttorts MICher the different avers, of tle atnriosphere are hori/ontallv homno-

eenjiis ic need lIi knoko ingn thle relationship betkx~ccii tie backscatter and extinction

C wii~cit erirs cait be cli inate hcl b coiipa rinri tilie ral rice-cornperisated po-vkers received from

h a1:tiil t1inI2 t\o ()Ir mrel) dillerent clexatiori angles ( Rutssel arid Ii1\ iiigston. 1984.;

Pl'tuJ -i.N1sO \ssirny !lie ;trtosphecrc to he liori/orirallv homiogieneous \wih extinctioin

oil b~ie'eattercelicicrnts. nt0mt arid] 131). respcctixcl\. Milch \ar\ onr inl tli: xer-tieal

dilectItni. Irek po\kers rceised tronil an altitude hr, along, at slant path rl elexated 61 degrees,

tid ~Ie a lanit path r, elcexatcd c), degrees, I le. 10) are gixen h\

(/I, ~ll (~i/1

21

4

FORWARD INTEGRATION [C(r) = CONST /

3

LIDAR I

SDOUBLE-ENDEDE D

2

.-- LIDAR 2

1 -" , I ..

SI I

REVERSE INTERGFATION [C(r) CONST]

E ; : " LIDAR 2

2 DOUBLE-ENDED

P - "

(b)

4"

Fiur 14 Co p rio of "..t..,c..-n co fice t ".) vs.....ro

REES NEGAIN1(r) is asumdNoSTontat

2DDAR2

2,,.,, . DOUL-ENDED-'.-

0I I I

0.1 0.3 0.5 0.7 0.9

RANGE FROM 1 (kin)

(b)

Figure 14. Comparison of extinction coefficients a(r), vs. range fromlidar 1 calculated from lidars 1 and 2 by using (a) forward and (b)reverse integration and those using the double-ended technique whenC(r) is assumed to be constant.

22

3

FORWARD INTEGRATION [MEASURED C(r)]

EE . ':. "i LIDAR 2

-C

01(a)

3.

REVERSE INTEGRATION [MEASURED C(r)]

2

TE" ,_ ' LIDAR 1-

E , .. ... .- .. LIDAR 2

00.1 0.3 0.5 0.7 0.9

(b)

RANGE FROM LIDAR 1 (kin)

Figure 15. Comparison of extinction coefficients a (r) vs. range fromlidar 1 calculated from lidars 1 and 2 by using (a) forward and (b)reverse integration when C(r) is allowed to vary as measured.

h2 -3r

h1

Figure 16. Geometry for the two-elevation-angle lidar technique.

23

and

) K/3(h, exp 2 f hIa(h,)dh', sin (21)(h, sin 02)2 0

xhere K is the system constant. Dividing Eq. 20 by Eq. 21, we get

P~,( ,sin d) Ail

P(r) = exp -2 a(h')dh' (22)P(r2)(h t sin I2' sin 01 sin 0, f

or. taking the logarithm of both sides

In P(r,)(h, sin 0 )2 --2 1 1 hl(h')dh' (23)P(r2)(h sin 02)2 sin i sin 02(3

and

oh1 S(r1 ) S(r2)

f a(h')dh' (24)0 2(l 'sin 02 1/sin 0,)

where

S(r) = In[P(r)(h sin 0)2] (25)

Similarly

f h, S(r3 ) S(r4) (26)f-%(h')dh' (60 2(1 sin a, I sin 0k)

I hen

t7,h-, h

, (h')dh' f i(h')dh'- f ao(h')dh'" j0 0

[S(r3 ) S(r 4 )] - [S(r1 ) S(r2 )]2(l sin - I sin 0 1)

In principle, if the atmosphere were horizontally homogeneous, the lidar beamcould be swept in elevation and an incremented profile of extinction (and consequentlybackscatter) could be determined by using the returns from closely separated angles (Kunz.1988). The smaller angular separations, however, place stringent requirements on the,-,icracies to which the range-compensated powers must be measured. If we assume thatthe percentage errors, 6. in the S(r) values are equal, the percent error, AT,,, in the calcu-lated optical depths is

(, , - T ,) 7,j 100 (28)

24

k here r', is gixen by

[S(r1 )( 1±6)] [S(r4)(I ±6]} {[S(r 1)( 1±6)] (S(r2 )( 1±6)]) (29)2(0 sin 0, I sin 0)

W'here 7, " ,) is a maximum, the signs of 6 are such that the products of S(r)(±6) are allof the same sign: then

+[S(r3 ) + S(r4 ) + S(rj) + S(r,)]6 (30)T, 7-, -- 30

2(I sin i2 s I )

I o demonstra:-" :his worst-case condition, the extinction coelficient is allowed to vary line-arly from a xalue of 0.05 km I at the surf:ti.,: to 1.0 km I at a height of I km. Assumingthat the relationship between backs%,ttter and extinction were invariant with altitude (i.e.,C(r) = C). S(r) values were :aculated for heights of 0.1 km and I km for elevation angles of6, = 60 degrees and 1)2 = 30 degrees. The sensitivity of the optical depth to uncertainties inthe S(r) valu'- ;s shown in Fig. 17 for In(KC) = 0 and - 5.4, which demonstrates that theerrors trv are also dependent upon the relationship between backscatter and extinction. Inhoth cases, the sensitix ity of the optical depth to errors in S(r) measurements is readilyapparent.

200

175

- 150 -+

- in KC -5.4'-125

0 75

M

w00

25 "

0 L

0 5 10 15 20 25

ERROR 1N S(R) (+/- %)

Figure 17. Errors in optical depths as a function of errors inS(r) measurements for an extinction coefficient profile whichvaries linearly from 0.05 km - I at the surface to 1 km - 1 at aheight of I km for different products of system constant K andbackscatter-to-extinct ion ratio C.

25

a:

The extent to which the atmophere is horizontally homogeneous could bedetermined from a single horizontal lidar shot by examining the gradient of the range-compensated power return. Then, if the atmosphere appeared homogeneous, i.e., aconstant negative gradient in S(r). the optical depth between any two altitudes could bedeteri,,ir.'-! f :':,, . 27. ii,,,wc~ i, in,, -no-ph , hn the ccnveci-'ely -- ixed .m .rine

boundary layer rarely, if ever, has the degree of homogeneity required. Paulson (1989) hasrecently conducted an investigation as to the usefulness of the two-angle lidar techniquein a coastal region. In these studies. data were taken beneath a thin stratus cloud layer atabout 500 meters. Two visioceilometer lidars were operated side-by-side on the west sideof the Point Loma Peninsula and pointed west overlooking the Pacific Ocean. A series ofnearly simultaneous shots were made with this arrangement. Without changing the orien-tation, the elevation angle of lidar I was increased to 25 degrees and that of lidar 2 to50 degrees. A series of nearly simultaneous shots were also made with this configuration.

Examples of the horizontal range-compensated power returns, S(r), for a 5-pointrunning average for each lidar are shown in Fig. 18. The two lidar returns are in quite goodagreement. but the irregularities and increasing return with increasing range indicate aninhomogeneous condition. The S(r) data plotted as a function of altitude for the two-angleshots are shown in Fig. 19. The effects of the inhomogeneities are evidenced by the fluctua-tions in the S(r) curve for lidar I above and below that for lidar 2 at different ranges. Theoptical depths between different altitudes determined from Eq. 27 are shown in the follow-ing table:

Table 1. Optical depths calculated from differentaltitudes up to a maximum altitude of 475 meterson 17 May 1989.

Lower Altitude (m) F Optical depth

I00 0.811125 0.437150 0.584175 0.597200 0.647225 0.584250 0.688275 0.150300 0.260325 0.260350 0.342375 0.492

I lie optical depth bctween 275 and 475 meters is only 0.15, while that from 375 to 475meters is more that three times greater (0.49). If the data were representative of a horizon-tall\ homogeneous condition, the optical depth up to 475 meters should consistentlydecrease as hI increases.

26

-4

-5

2-6

LIDAR 1

-7 -LIDAR 2

-80 0.5 1

RANGE (km)

Figure 18. Examples of measured S(r) values obtained from parallelhorizontal lidar shots for lidars 1 and 2 with a 5-point running average.

05/17/89

0

2= 50 deg

-2

. 4 -6/S=25 deg

-80 0.1 0,2 0.3 0.4 0.5 0.6 0.7

ALTITUDE (km)

Figure 19. Examples of two-elevation-angle lidar shots (02 25 degreesand 02 = 50 degrees) with a 5-point running average.

27

ASSESSMENT OF SYSTEM PERFORMANCE

The computer code PREOS (Performance and Range for EO Systems) presentlyresides in the US Navy's Tactical Environmental Support System (TESS). The presentxersion of PREOS addresses the ability of FLIR systems on the S3-A, A-6, and P-3 opera-tional aircraft to perform range-dependent tasks such as the detection, classification,and identification of surface targets. The code utilizes polynomial fits calculated withI.OWTRAN 3B (Selby et al., 1976) of equivalent gaseous absorber amounts based onmeteorological parameters. The aerosol transmittance model in PREOS was developedfrom a model obtained from the work of B.A. Katz at the Naval Surface Warfare Center(see Hughes and Richter, 1979). This model employs height-dependent marine and conti-nental components. In a manner similar to the LOWTRAN 6 NAM, the marine contribu-tion is scaled to surface wind speed. and both contributions contain relative humidityparticle-size growth factors.

The agreement between surface target detection ranges as reported by operationalsquadrons and those predicted by the code has been disappointing. However, the wide,catter between observed and predicted ranges (Fig. 20) is probably more characteristic ofoperator capability and the quality of input data than the predictive capability of themodel. The meteorological inputs to the code were obtained in many instances from radio-sonde launches which may have occurred several hundred kilometers from the observationsites and may have been separated in time by several hours. Predictions of performancerange are also critically dependent upon the actual temperature of the ship and of its natu-ral background. The algorithms in PREOS are currently based on a fixed temperaturedifference between a rectangular target and its natural background. This approach neglectsthe effects of a wind-ruffled sea on sky reflections, emissions from the surface wave facets,and contributions of the intervening atmosphere to the total background radiance scene,which changes \kith viewing angle and the altitude of the sensor. Without knowing theship's temperature, which is dependent upon its history (course, speed, and surroundingmeteorological parameters), it is questionable that the accuracy of detection range againstan adversary can be much improved. However, a ship commander, when aware of his ownship's past and future courses, can use the prediction algorithms to determine the ranges atxshicn an auversary can detect or track the siip using passive IR sensors. These stand-offranges are of primary importance in estimating the time allowable for evasive actionsagainst guided weaponry launched at the ship or for the deployment of countermeasures.

Currently available computer codes, such as SIR EOS (Burns et al., 1980) (three-dimensional) and SIRS (Batlev. 1978) (two-dimensional), are capable of using severalhundred individual structural elements of a ship to model its composite IR signature. Thesecodes, however. are quite complex and require extensive running times, making themimpractical for shipboard use in a real-time prediction system. A modification to the SIRScomputet code has recently been developed at the Naval Surface Weapons Center. Thismidification. SHIPSIG (Ostrowski and Wilson, 1985), approximates the complex struc-ture of a ship with plane elements which represent the ship's temperature at zero-range onan average basis. For a given viewing direction, the simplest representation of the shipconsists of a single vertical and horizontal element, with the observer's orientation,ccounted for by appropriate area components. In the present model, the horizontal andvertical elements and ship-stack correction factors apply to a guided missile frigate-classship. The model requires as inputs the ship's course and speed as a function of time froma starting geographic latitude, the surface wind speed and direction, visibility, relativehumidity, air temperature, the ship's initial temperature, and the viewing angle.

28

70

6060t " -

z 4-i- * *. - * A

Z 40 - -AyT*0

Z.. 40y 47t A-- T

o -0 . - -.

,w I I I9 1 I

130 * ** 30 * 0

w * 4 *

0o20 -* * ***

10

0 10 20 30 40 50 60 70OBSERVED DETECTION RANGE (nm)

Figure 20. Comparison of observed FLIR detection ranges and those predictedby PREOS.

A FIIR detection algorithm has recently been developed (Hughes, 1989c) by using

the SHIPSIG model and the sea surface background model, which varies with sensor alti-

tude and viewing (/enith) angle. As neither model has undergone extensive validation, the

algorithm is not vet incorporated into the operational version of PREOS. To demonstrate

the use of the algorithm as a tactical decision aid to predict the vulnerability of a frigate-

class ship to detection by an airborne common module FLIR, a case study is presented

(using the actual course of a frigate operating off the coast of San Diego. California).

During a 5-hour period the ship's course changed. allowing solar heating of different sides

of the ship.

For this study, the Piper Navajo aircraft made a vertical spiral in the vicinity of

the ship to obtain the profile of temperature, relative humidity, and pressure, which are

required inputs to the modified 1.OWTRAN 6 computer code for calculating the sea and

skx radiances. A Barnes PRT-5 radiation thermometer was also onboard the aircraft to

measure sea surface temperature from low-level constant-altitude flights- the temperature

was determined to be 16.4°C. The vertical profiles of temperature and relative humidity.

xhich were measured at 1330 PST approximately 9 km off the coast of San Diego. Cali-

fornia. are shown in Fig. 21.

29

RELATIVE HUMIDITY (%)0 10 20 30 40 50 60 70 80 90 100

300 [-I I I I I I I I I I3000240

20 AIR TEMPAI

1800w1

P 1200

URELATIVE HUMIDITY

600 .

0 - _ - L10 15 20 25 30

TEMPERATURE (- C)

Figure 21. Profiles of air temperature and relative humidity measured withaltitude on 9 June 1988 off the coast of San Diego, California.

Near the time the meteorological parameters were obtained, measurements of8 12-,um horizon radiances were made with the thermal imaging system (AGA THER-MOVISION, model 780). The current and 24-hour averaged wind speeds (1, = 2.9 m/s andF = 2.8 m s) and the vertical profiles of meteorological parameters were used in LOW-FRAN 6 to calculate the radiance that matched the maximum radiance in the scene fornonunique combinations of air mass factors and visibilities. At the time of the measure-ments, the Los Coronodos coastal islands off San Diego were barely visible to the nakedeye at ranges between 25 and 35 km. Choosing a visibility of 37 km, an integer value of 3resulted as the appropriate air mass factor for the LOWTRAN 6 NAM. Figure 22 showsthe comparison of the measured and calculated IR radiances for zenith angles within aboutI degree above and I degree below the horizon, with an air mass factor of 3 and a visibilityof 37 km. Both the calculated sky (0 < 90.17 degrees) and sea (0 > 90.17 degrees) radiancesare in good agreement with the measured values for this low-wind-speed case.

The selected atmospheric model was used to calculate the contributions of thepropagation path and sea and reflected sky radiances to the total background radiance as afunction of altitude and zenith angle. In Fig. 23 the total apparent blackbody temperatureof the sea background from the three contributors is plotted versus zenith angle for sensoraltitudes of 500 meters and 2000 meters. At the 500-meter elevation, the dip in temperatureat about 97 degrees is a result of the rapid fall-off of propagation path emission withincreasing zenith angle (i.e.. shorter slant paths to earth than at the 2000-meter elevation).For zenith angles greater than about 100 degrees, there is little difference in the apparenttemperatures at each altitude, and both approach the measured sea surface temperature atthe nadir zenith angle. This is in contrast to Fig. 10, where propagation path emissionsbetween the sensor and sea surface made significant contributions to the total radiance atthe nadir zenith angle for both at an altitude of 1000 meters.

Figure 24 shows the course of the guided missile frigate USS Brooke (FFG I) offthe coast of San I)icgo, California. on 9 June 1989, this course was chosen to demonstrate

30

3.36/9/88

1508 PDT

S3.1E3.1 AA 8-12/um

E

Z 3.0 O -- MEASUREMENTS % %

LOWTRAN CALCULATIONS

2.9 - V;= 2.9 m/s, V = 2.8 rn/sAM 3, VIS = 37 km

2.8 I I89.0 89.5 90.0 90.5 91.0

ZENITH ANGLE (deg)

Figure 22. Comparison of measured and calculated IR radiances for zenithangles above (0 < 90.17 degrees) and below (0 > 90.17 degrees) the horizon.

18

SEA SURFACE TEMPERATURE

16

HORIZON

14 h=2000m

0S 12 -.

a.

W - APPARENT SEA SURFACE10 ........ TEMPERATURE (CALCULATED)

8

6; I I I I I I

80 90 100 110 120 130 140 150 160 170 180ZENITH ANGLE (deg)

Figure 23. Total apparent blackbody temperature of the sea backgroundversus zenith angle for sensor altitudes of 500 meters and 2000 meters.

31

32.76/9/88 SAN DIEGO HARBOR

BUOY NO. 6

0

32.65

2" 095219 0900 0800 PST

4k0---.----------------32.6 1200

• -" 1235

1031

32.55 -1300

32.5 11 1117.9 117.7 117.5 117.3 117.1

LONGITUDE (west)

Figure 24. Course of the USS Brooke (FFG1) on 9 June 1988.

the model. During the 5-hour time period, changes in the ship's heading allowed solar heat-ing of different sides of the ship. As the ship completed the course and returned to harbor,it passed close to the AGA thermal imaging system located near the entrance to the harbor.["he AGA system's data processing software allows subtraction of the sea backgroundradiance surrounding the ship and provides a histogram of the temperature distribution ofthe ship pixels within the chosen rectangular area. The mean temperature of the ship (uncor-rected for atmospheric effects) was 19.7°C. The measured radiance, N(meas), of the ship ata range r is related to its actual effective blackbody radiance, N(ship), and the atmosphericemission, N(path), along the path by

N(meas) = (ship)r(r) + N(path) (31)

\A, here T(r) is the atmospheric transmittance at a range r. The range to the ship, determinedby using the known vertical dimensions of the ship and their angular subtense within thelield-of-vie"' of the AGA, vkas approximately 1.7 km. LOWTRAN 6 calculations of trans-mittance and path emission were made to determine the temperature equivalent to N(ship).1hese calculations resulted in an adjusted AGA average temperature measurement of20.5°C, assuming that the surface emissivity of the ship was unity.

For the ship model calculations, the ship's initial position was taken to be near theentrance to San Diego harbor. The initial ship temperature, the ambient temperature, andthe relative humidity throughout the course were not recorded by the ship. These values\kcre taken to be constants as measured at the AGA site. The surface wind was south-\%estcrly (252 degrees true), and the depression angle of viewing was essentially broadsideat 0.6 degree. lhe average ship temperatures calculated for the port and starboard sides ofthe ship as a function of time are shown in Fig. 25. The most apparent features in thetemperature responses arc the rapid heating of the port side and the gradual cooling of thestarboard side as the ship steamed westward in the early morning, and their abrupt cooling

32

6/9/8829.00

27.00

LU

~ 2500STARBOARD

LU

1-- 23.00 -0 )

PORT

21.00AGA MSMT (ADJUSTED)

(E 1.0)

19.00 I I

0800 0900 1000 1100 1200 1300 1400

LOCAL TIME

Figure 25. Average temperature of port and starboard sides ofthe USS Brooke (FFG1) and the adjusted AGA measurements asthe ship entered San Diego harbor.

and heating after 1000 hours following the southeasterly course change at 0952 hours. Themagnitude of the port side average temperature is approximately PC greater (for an emis-si,,it\ of unity) than that measured by the AGA system as the ship returned to harbor near1345 hours. If indeed the emissivity. E. of the ship was 0.9. as is assumed in the model. themeasured average temperature would be in better agreement (22.7C). Allowing for theuncertainties in the meteorological parameters surrounding the ship throughout the course.the reasonable agreement between the adjusted AGA measurements and the model predic-tinon,, is gratifying.

In the performance calculations, the LOWRAN 6 code is used to directly calculatethe sum of the ship and path radiances received by the sensor at a range r as

.\(r), .V(r = 0),r(r) + .(r), (32)

k\ here .V(r = 0), is the ship radiance at /ero range and N(r),, is the path radiance. N(r),,,is the'n conserted to an equivalent blackbody temperature, Tr) , by an iterative solutionto Planck's hlackhodv formula. Similarly, an equivalent blackbody temperature, T(r),,. ofthe background radiance at the specified altitude and range is calculated, and the resultingapparent temperature difference. I T(r)a : T(r)\,1 , jJr), is determined. The range atk% hich the apparent temperature difference is equal to the system's minimum detectabletemperature difference (MDID) curve determines the maximum detectable range (M[)R)of the ship. I he I.IR system MDI)T) versus range (spatial frequency) curve was calculatedb\ using the formulation for a hypothetical ship operating against a rectangular target. InFig. 26 and 27. the calculated MDRs for the USS Brooke by an airborne FL1R operatingat altitudes of 0.5 km and 1.0 km, respectively, are shown. The MDRs could have been

33

30

MDR = 35 km29 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

28 - h 0.5 km

27

26 STARBOARD

25 25

:E24 -PORT

23

MDR 31 km22

21

200800 0900 1000 1100 1200 1300 1400

LOCAL TIME

Figure 26. Calculated MDR envelopes for the USS Brooke (FFG1I) by anairborne FLUR at an altitude of 0.5 km.

30

MDR = 56 km29 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

28 - h 1.0 km

27

26_ STARBOARDwD

26 -

24 PORTw

23

22 *.MDR 53 km

21

200800 0900 1000 1100 1200 1300 1400

LOCAL TIME

Figure 27. Calculated MDR envelopes for the USS Brooke (FFG1) by anairborne FLUR at an altitude of 1.0 km.

34

:LL1,ILd 11a I inili of imfc throughot the the ship's course:- hoxkever, for the sake of',,nIm1ph cit \ onk 1\i th \ iinero hi fit\ v dtection enveclopes, for thle entire duration of the ship's

.tII\C r h .InI 11he fItI-.ics. thle sipl's avejaac temperature for- both thle port and"I.1 hoard ides a ic sho\\n. InI Fig. 26, thle ship is seen to he \u I nerabie to I R detect ion01h0112h01.t It, course fro Ia altitUde of 0.5 kml at at range of 31 kml. Howecver, bevond

kilkm the: shp is not detetajble. SiialIn Fi27. thle ship IS vulnerable at a range of'SkinI 1,1n 11n altitudeI of 1 .0 kmll but IS Safe from detection beyond 56 km. In Fig. 28 at

0c[fiIIaIN01 )I, liox i of the \iI)Rs calcu-lated Wxth thle current algorithm and wilth tilenthdthat 1ass times a et n~Sta tter Clper-artire difference bet k cen the ship and its back-ritd0I (. When the CLurretit method is uised, considerable increases (-2() kmi at an

iltiude0 )I kin are obtinedCL comlpare-d to thle fixed-temperature method in predicted

SPRESENT METHODN CC14STANT-TEMPERATURE-DIFFERENCE METHOD

2.0 -(AT 5 C)

1,5

1 0

0,5

00 10 20 30 40 50 60 70 80 90

DETECTION RANGE (kin)

Figure 28. Comparison of the MIDRs (lower envelope) for the USS Brooke(FFG1) calculated with the current algorithm and with the method thatassumes a constant temperature difference of 500 between the ship andits background.

435

(ONCIUSIONS AND RECOMMENDATIONS

In the absence of transmission or radiometric measurements. the L.OW-I RAN codes1must currently be relied on, along with measured meteorological parameters and models of,jerosol sIe distributions, to predict the atmospheric effects on EO system performance.I hese codes haxe proven to be a versatile tool in predicting atmospheric radiance. NAMha, undcrgonc considerable caluation and should be considered a workable model inperformance predictions. While NOVAM is still in the development and validation stage. it,hovs promise of accounting for the dynamic changes within the mixed layer and shouldtransition into ILO I RAN in the near future. Still lacking is the ability to measure the airmass factors and slant-path visibility in operational conditions. Emphasis should be placedon alternati\c methods, such as the sky radiance technique, rather than on air mass charac-teri/ation for determining the required input parameters.

While the remote-sensing potential of lidars shows promise, the capability of deriv-ing extinction coefficients from backscatter measurements has not been demonstrated withMn assured degree of accuracy when the atmosphere is inhomogeneous. The two-angletechnique certainly would be useful in situations where the atmosphere can be shown to behori/ontally homogeneous, provided that the system calibration can be accurately deter-mined. -The double-ended technique resolves the requirement for knowing the relationshipbetween backscatter and extinction. However, the requirement for having an instrumentedrange at both ends is impractical in an operational situation. It appears that the double-ended technique is best suited for aerosol studies and model validation.

The reliability of the sea surface radiance model that uses the LOWTRAN 6 NAMto accurately represent measured values for low wind and moderate wind speed conditionshas been demonstrated. Whether or not it will be representative of other wind speed condi-tions needs to bc determined. Also, the preliminary evaluation of the average ship tempera-ture model showed promise, since this model responds to the differing solar conditions.[uture attempts at validation must ensure the accuracy of the ambient meteorologicalconditions. Onboard ground-truth radiometry measurements of the temperatures of differ-ent portions of the ship are also needed to aid in determining the accuracy of the adjustedaierage temperatures inferred from the AGA measurements. Finally, a controlled experi-ment %A ith an airborne operational system must be conducted to determine the validity ofthe predicted detction ranges under varying meteorological conditions.

36

REFERENCES

Bailex P.F. "Ship infrared Signatures (SIRS) Computer Model Technical Overviewand tsr Mlanual." Dl~ad 'laylor Naxal Ship Research and IDcxelopment CenterReport 51 F"'8-37 (1978).

H I',',on nett c.R ., 'Se n',It ixIt x A naksis of' I .dar In version AlIgorit hmis." Appi. Opt.. 25,

2122 1985).

Bums. R. 11. et al.. "lDcseription of the Ship Infrared Electro-Optical Scenario (SIR FOS)Comnputer MIodel. Dax id llo aal Ship Research and lDevelopment Center Report80) 09? ( 19,S).

(ov~ C'., and W. MIuink. NI easuremients of' Roughnless of the Sea Surf'ace from Photo-0raph", o1 thle Snn\" (ilitter7Jo.10.1 Opt. Soc. of' Am.. 44. 838 (1954).

dc Ic~m .~. -\ert ical IProtlles, of (i ant Particles Close A box c the Sea Surface.-Il 115It . -? XB. 5 I 19?86).

de: I ,c u x . (1-l1. K II n, anld C. -la inbert, s" H u nid itx F leet s onl the Backscatter[\tInction Ratio," Appl. Opt.. 25. 3971 (1986).

d& LCeeii Gi., K. I , D~a% idson. S.G. Gathmart, and R.V. Noonkester. "Ph\ysical Models forAerosol in thle Marine Nixed-Laver." Conference Proceedings of' the 44th AGARD.S\ Ilposlunl of the Electrontagnetic Wave Propagation Panel on "Operational DecisionA\id,, for Exploiting or NI iti gating Elect romagnet ic Propagation Effects." CP453. PaperNo0. 40) 1989).

I-ergeuson. A.. and l).H. Stephens. "Algorithm for In-verting Li1dar Returns." AppI. Opt.,-4, 3673 ( i983).

1 i~erad.J.W.." Approximate Formulas f'or the Equilibrium Size of an Aerosol Particlea, a F~unction of- Its D)r% Suec and Composition and the Ambient Relative Humidity." J.Arp. Meteorol.. 14, 1044 (1975).

1 it/gcrald, JiW.. "*Effect of RelaLiv Humidity on the Aerosol Backscattering Coefficient at.694-and-l0.6 pm Wax elcnigths.- AppI. Opt. 23. 411 (1984).

(iathman. S.G.. "Optical Properties of- the Marine Aerosol as Predicted by the Navy Aero-sol 'vlodel.- Opt. Fnig- 22. 57 (1983).

(iathman. S.G.. "A~ Preliminar% Description of NOVAM. the Navy Oceanic Vertical Aero-"o()I Mode)." Nax al Research I aboratorv Technical Report 9200 (1989).

(iathnian. S..and H. Vlfecrs, "On the Accuracy of' IR Predictions Made bv the Naxv\ero0s1l Model." American Meteorology Societv Ninth Conf'. on Aerospace and Aero-

nlautical Meteorology. pp. 194 198 (1983).

M.g .I.leasUrements of Atmospheric Aerosol Thickness using ERTS-1 Data." Jour.ijr Pol1lut. Control A\ssoc., 25. 622 (1975).

II ughces. HII .. "Ix alnation of the 1.0W) RAN 6 Nax v Maritime Aerosol Model Using 8 to12 pim Skx Radian rces." Opt. Eng., 26. 1 155 (1987 ).

lILnuhe\,. HI.(i. "Apparent Infrared Radiance of the Sea." Naval Ocean Systems Center.1 R-1271 (1989a).

11 ughes, . C.. -Sea and Skx Infrared Radiances Near the H-orlion." Nax a! Ocean Systems(enter. I R- 1294 (1989b).

11 ughes,. 11 Gi. ",Airborne El.!R IDetcction of Surf'ace Targets." Naval Ocean Systems('enter. I R-1 275) 1989c).

37

Hughes, H.G., and D.R. Jensen, "Aerosol Model Selection Using Surface Measurementsof IR Horizon Radiances and Satellite Detected Visible Radiances," Appl. Opt., 27,4367 119881

Hughes, H.G., and M.R. Paulson, "Double-Ended Lidar Technique for Aerosol Studies,"Appl. Opt.. 27. 2273 (1988).

Hughes. H.G.. and J.H. Richter, "Extinction Coefficients Calculated from Aerosol SizeDistributions Measuied in a Marine Environment," Proceedings of the Society ofPhoto-Optical Instrumentation Engineers (SPIE), San Diego, California, Volume 195,pp 39-45 (1979).

Klett, J.D., "Lidar Inversion with Variable Backscatter, Extinction Ratios," Appl. Opt., 11,1638(1985).

Kneizvs, F.X.. [P. Shettle, W.O. Gallery, J.H. Chetwynd. Jr., J.H. Abreu, J.E.A. Selby,S.A. Clough, and R.W. Fenn, "Atmospheric Transmittance/ Radiance: Computer CodeI.OWTRAN 6. Air Force Geophysical Laboratory Technical Report No. 83-0187(1983).

Kneizvs, F.X., E.P. Shettle, L.W. Abreu. J.H. Chetwynd, Jr., G.P. Anderson, W.O.Gallery, J.E.A. Selbv. and S.A. Clough. "Users Guide to LOWTRAN 7, AFGL-TR-88-0177 (1988).

Kunz, G.J., "Bipath Method as a Way to Measure the Spatial Backscatter and ExtinctionCoefficients with Lidar," Appl. Opt., 26, 794 (1987).

Kunz, G.J., "A Method for Measuring the Vertical Extinction and Backscatter Profile witha Scanning Lidar," TNO Physics and Electronics Laboratory, Report No. FEL 1988-65(1988).

Lindberg. J.D., W.J. Lentz, E.M. Measure, and R. Rubio, "Lidar Determination ofExtinction in Stratus Clouds," Appl. Opt., 23, 2172 (1984).

Miller. W.B., and i.C. Ricklin, "A Module for Imaging Through Optical Turbulence(IMTURB)," US Army Atmospheric Sciences Laboratory, Report No. ASL-TR-0221-27 (1987).

M ulders. J. M., "Algorithm for Inverting Lidar Returns: Comment," Appl. Opt.. 23, 2855(1984).

Ostrowski. P.P., and D.M. Wilson, "A Simplified Computer Code for Predicting Shiplnc-ared Signatures," Naval Surface Weapons Center TR 84-540 (1985).

Paulson, M.R., "Evaluation of a Dual-Lidar Method for Measuring Aerosol Extinction,"Naval Ocean Systems Center, TD-1075 (1987).

Paulson. M.R., Atmospheric Horizontal-Inhomogeneity Effects on the Optical DepthsDetermined by the Double-Elevation-Angle Lidar Technique," Naval Ocean SystemsCenter. TD-1600 (1989).

Russell, P.B., and J.M. Livingston, "Slant-Path Extinction Measurements and Their Rela-tion to Measured and Calculated Albedo Changes," J. Clim. and Appl. Meteor. 23,1204 (1984).

Salemink. H.W.M., P. Schotanus, and J.B. Bergwerff. "Quantitative Lidar at 532 nm forVertical Extinction Profiles and the Effect of Relative Humidity," Appl. Phys., B34,187 (1984).

Selby,,.A.F E.P. Shettle. and R.A. McClatchey, "Atmospheric Transmittance from 0.25to 28.5 pm: Supplement LOWTRAN 3B (1976)," AFGL-TR-76-0258 (1976).

38

Wollenweber, F.G., "Effects of Atmospheric Model Layering on LOWTRAN 6 Calcu-lations of 8 to 12 pm Near Horizon Sky Radiances," Naval Ocean Systems CenterID 1193 (1988a).

Wollenweber. F.G., "Infrared Sea Radiance Modeling Using LOWTRAN 6," Navn! OceanSystems Center ID 1355 (1988b).

39