imaging fourier-transform spectrometry for plume ... · imaging fourier-transform spectrometry for...

International Journal of Energetic Materials and Chemical Propulsion, 12 (1): 15–26 (2013)

IMAGING FOURIER-TRANSFORMSPECTROMETRY FOR PLUME DIAGNOSTICSAND CODE VALIDATION

Michael R. Rhoby,1 Jacob L. Harley,1 Kevin C. Gross,1,∗

Pierre Tremblay,2 & Martin Chamberland3

1Department of Engineering Physics, Air Force Institute of Technology,2950 Hobson Way, Wright-Patterson AFB, Ohio 45433, USA

2Centre d’optique, photonique et laser, Universite Laval, 2375 rue de laTerrasse, local 2104, Quebec, Qc, Canada G1V 0A6

3Telops, Inc., 100-2600 avenue St-Jean-Baptiste, Quebec, Qc, Canada G2E6J5

∗Address all correspondence to Kevin C. Gross E-mail: [email protected]

Laminar and turbulent flow fields found in smokestacks, flames, jet engine exhaust, and rocketplumes are of practical and academic interest and could greatly benefit from spatially resolved spectralmeasurements. Key physical flow field parameters such as temperature and species concentrations canbe extracted from spectral observations. Spectral images of flow fields produce rich information forplume diagnostics and could be used to validate next-generation plume codes. Laser-based diagnosticsare typically used to measure temperatures, concentrations, and flow velocities. Unfortunately, theselaser-based techniques are largely confined to a laboratory environment, and tracking multiple speciesconcentrations is complicated due to the limited bandwidth of tunable laser sources. The advantageof a passive sensor with high resolution across a broad bandwidth would make an imaging Fourier-transform spectrometer (IFTS) an attractive instrument for flow diagnostics, particularly when theflow field of interest cannot be studied in a laboratory. In this paper, we present an overview of IFTSand its uses for flow visualization and combustion diagnostics in various plumes. Examples fromrecent measurements of laminar flames and jet engine exhaust will be presented.

KEY WORDS: fourier-transform spectrometry, combustion diagnostics, spectralanalysis, radiative transfer, hyperspectral imagery

1. INTRODUCTION

Laminar and turbulent flow fields emanating from smokestacks, flames, jet engines,and rockets are of practical and academic interest and could benefit from spatially re-solved spectral measurements. Spectral emissions encode important flow field param-eters such as temperature, density, and species concentrations. Laser-based diagnostics

2150–766X/13/$35.00 c© 2013 by Begell House, Inc. 15

16 Rhoby et al.

are typically used to measure these parameters (Kohse-Hoinghaus and Jeffries, 2002).However, such techniques are a challenge to set up and are limited to a laboratory envi-ronment. The limited bandwidth of tunable laser sources makes tracking multiple speciesconcentrations difficult. The advantage of a passive sensor with high resolution across abroad bandwidth would make imaging Fourier-transform spectrometry (IFTS) an attrac-tive instrument for flow diagnostics, particularly when the flow field of interest cannotbe studied in a laboratory. In this paper, we present an overview of IFTS and its usesfor flow visualization and combustion diagnostics in various plumes. Examples from re-cent measurements of a laminar flame (Rhoby and Gross, 2012) and jet engine exhaust(Bowen, 2009; Bradley, 2009; Harley et al., 2012; Moore et al., 2009; Tremblay et al.,2009) will be presented.

2. INSTRUMENTATION

We have looked at various high-temperature laminar and turbulent flow fields using aTelops Hyper-Cam interferometer (Chamberland et al., 2005; Farley et al., 2006). ThisIFTS features a high-speed 320× 256 pixel InSb (1.5–5.5µm, 2 kHz full-frame) focal-plane array (FPA). Sequential scene imagery focused on the FPA is collected while look-ing through a scanning Michelson interferometer. The interferogram cube is thus a stackof broadband infrared images collected at fixed optical path differences (OPDs). Acqui-sition rate depends on spectral resolution and mirror speed, which in turn is affected byspatial resolution and camera integration time.

An ideal Michelson-based IFTS produces (at each pixel) an interferogramI(x) rep-resented by

I(x) =12

∫ ∞

0[1 + cos (2πxν)]G(ν) [Ls(ν) + Li(ν)] dν = IDC + IAC (x)

wherex is the optical path difference,Ls(ν) is the scene spectrum,Li(ν) are spectralemissions from within the instrument, andG(ν) is the spectral system response whichincludes the quantum efficiency of the detector. Here,IDC represents the integrated in-tensity andIAC(x) is the cosine transform of the (uncalibrated or raw) spectrum. Fouriertransformation ofI(x) − IDC yields the raw spectrum. This implicitly assumes thesource spectrum is static over the course of the measurement. For laminar flow, thisis typically true. However, the case of turbulent flow in whichLs(ν) may rapidly andstochastically change throughout an interferometric measurement is addressed in Sec-tion 3.2.

Two on-board blackbodies permit linear calibration to remove the effects of detectorresponseG(ν) and instrument self-emissionLi(ν). A schematic of an IFTS is presentedin Fig. 1. Also shown are an example interferogram for a single pixel and its correspond-ing spectrum upon Fourier transformation.

International Journal of Energetic Materials and Chemical Propulsion

Imaging Fourier-Transform Spectrometry for Plume Diagnostics and Code Validation 17

Movable Retrore�ector

Optical Path

Di�erence

Beam

1 Path

Beam

2 Path

Beam 1 & 2 Interfere

Scene

Fixed Retrore�ector

Beamsplitter

Detector

Ima

gin

g O

pti

cs

-0.06 -0.04 -0.02 0 0.02 0.04 0.06

5

10

15

20

2000 3000 4000 5000 6000

0

5

10

15

FIG. 1: Left panel: Schematic of an imaging Fourier-transform spectrometer. An in-terference pattern is measured at the focal-plane array detector by varying the phasebetween the two light beams via the movable retroreflector. Right panel: Illustration of asingle-pixel interferogram (top) and its corresponding spectrum (bottom) upon Fouriertransformation.

3. THEORY

3.1 Radiative Transfer for Ideal Turbulent Flow

The spectral radianceL(ν) from a non-scattering source in local thermodynamic equi-librium along a lengthl line-of-sight (LOS) can be expressed as (Thomas and Stamnes,2002):

L(ν) =∫ l

0e−τ(s)κ(ν, s)B[ν, T (s)] ds (1)

whereτ(s)=∫ lsκ(ν, s′) ds′ is the optical depth,κ(ν, s) is the absorption coefficient, and

B(ν, T ) is Planck’s blackbody distribution at temperatureT . The termκ(ν, s)B[ν, T (s)]accounts for photons “born” at the points along the LOS, ande−τ(s) accounts for thefraction of those photons absorbed as they travel through the remaining plume towardthe instrument. The dependence ofκ on bothT (s) and species concentrations~ξ(s) wassuppressed. For an ideal, high-temperature, two-dimensional flow field which is homo-geneous along the LOS, Eq. (1) can be approximated by

L(ν, T ) = τ(ν)ε(ν,~ξ, T )B(ν, T ) (2)

where the source emissivityε is defined byε(ν,~ξ, T ) = 1−e−κ(ν)l andτ(ν) representsthe transmittance of the material (atmosphere) between the source and instrument. Thismodel assumes the plume radiance dominates all other sources (e.g., photons emittedbehind or in front of the plume).

Volume 12, Number 1, 2013

18 Rhoby et al.

In this work, spectra are modeled using the line-by-line radiative transfer model(LBLRTM) (Clough et al., 2005) in conjunction with the high-temperature extension[HITEMP (Rothman et al., 2010)] to the HITRAN database (Rothman et al., 2009) ofspectroscopic line parameters.

Note that at all wavenumbersν, Planck’s distributionB(ν, T ) monotonically in-creases with temperature. Additionally, for many gas-phase systems in local thermo-dynamic equilibrium, this monotonicity is preserved, so we assumeT2 > T1 impliesL(ν, T2) > L(ν, T1) for all ν.

In a nonreactive turbulent flow field, the instantaneous temperatureT fluctuatesabout a mean temperature〈T 〉 according to a probability distributionP (T ) (Mathieuand Scott, 2000). Uncorrelated fluctuations in~ξ may also occur, but are ignored1. For anergodic flow field, the average of an ensemble of spectral measurements yields

〈L(ν, T )〉 =∫

L(ν, T ) P (T ) dT 6= L(ν, 〈T 〉) (3)

where the nonequality arises due to the nonlinear dependence ofL on T . To properlyinterpret〈L(ν, T )〉, a priori knowledge ofP (T ) would be required and simply fittinga single-T model to it necessarily results in biased temperatures and species concentra-tions. To address this problem we now consider flow measurement made by an interfer-ometer.

3.2 Quantile Interferogram Analysis for a Two-Dimensional Turbulent FlowField

Dynamic scenes are often considered problematic for IFTS as changes in scene radi-ance during the interferometric scan produce scene-change artifacts (SCAs) in the spec-trum. While time averaging can minimize the effects of this “source noise,” an alter-nate method is presented which, in addition to minimizing SCAs, can provide addi-tional information about the fluctuation statistics in the flow field. In the case of two-dimensional turbulent flow which is dominated by temperature fluctuations and is ho-mogeneous along the instrument’s LOS, temperature fluctuation statistics can be recov-ered.

To simplify the presentation, we assume an instrument response of unity and ig-nore instrument self-emission. Under these conditions, an ideal Michelson produces aninterferogramI(xi, Ti) at each OPDxi of the turbulent flow via

I(xi, Ti) =∫

[1 + cos(2πxiν)]L(ν, Ti) dν (4)

1If concentration fluctuations are significant, the one-to-one mapping of quantile spectra tounique temperatures to be described may not be valid. However, multiple quantile spectra docontain information complementary to and different from the mean spectrum.



whereTi represents a random sample fromP (T ) and is assumed constant over theshort FPA integration time. With a FPA, the DC component is preserved, and this iskey to the following development. Recall thatL(ν, T ) is a monotonic function of tem-perature at allν. Since1 + cos(2πxiν) ≥ 0 for any xi and all ν, it follows thatT2 > T1 → I(xi, T1) > I(xi, T2)∀xi. If an ensemble of interferometric measure-ments of the ergodic flow field are captured, then at eachxi a range of temperaturesweighted byP (T ) will have been observed. As the chain of probabilities demonstrates,the monotonicity ofL(ν, T ) permits sorting the ensemble of measuredI(xi)’s into var-ious quantiles

q = P {T ≤ Tq} = P {L(ν, T ) ≤ L(ν, Tq)}= P {I(xi, T ) ≤ I(xi, Tq) ≡ Iq(xi)} ∀xi (5)

whereTq is theqth quantile,P { } denotes probability of the argument, andIq(xi) de-fines the “quantile interferogram”. So long as a sufficient number of measurements aremade to enable robust quantile estimates,Iq(xi) is a valid interferogram correspondingto the spectrumLq(ν) ≡ L(ν, Tq).

The limitation to an unrealistic two-dimensional flow field may appear to limit theutility of this technique. However, the sorting of interferograms can still be performedto yield quantile spectra. These quantile spectra contain information which is comple-mentary to and distinct from the mean spectrum. An example from an axisymmetric jetis presented in Section 4.2 and demonstrates this point.

3.3 Extraction of Moderate-Speed Imagery from Interferometric Measure-ments

The Michelson interferometer encodes spectral information via intensity variations [asrepresented by the cosine term in Eq. (4)]. These variations occur at a frequency greaterthanf = vmνd, wherevm is the mirror scan velocity andνd is the lowest frequencyphoton (νd ∼ 1700 cm−1) that the camera detects. Thus, a temporal low-pass filter canbe applied to the interferogram cube yielding moderate-speed imagery. Also, if there arebroad regions in which no spectral emissions are observed, a temporal band-pass filtercan be applied to recover imagery (with no DC level) at higher frame rates. The mirrorscan velocity varies with spatial resolution and camera integration time.

A specific example illustrates the differences between camera and spectral imageacquisition rates. For a window size of 48× 156 pixels and an integration time of 5µs,the camera in the IFTS acquires images at nearly 10 kHz as the Michelson assemblycontinuously varies the optical path difference (OPD) between interfering beams. Eachimage corresponds to a change in OPD of 632.816 nm,2 and in this instrument config-uration, the mirror speed is 0.64 cm s−1. To achieve spectral images at a resolution of

2A HeNe reference laser is used to trigger the camera to capture images at regular OPD intervals.


20 Rhoby et al.

1.5 cm−1 between 1700 cm−1 < ν < 6667 cm−1 requires approximately 12,500 se-quential images collected between –0.4 cm< OPD < 0.4 cm. The spectral image isthus acquired at 0.8 Hz. While the camera frames at 10 kHz, intensity modulations atfrequencies greater thanf = 1700 cm−1 × 0.64 cms−1 = 1088 Hz could occur due tothe action of the Michelson; thus, the effective frame rate after low-pass imagery is ap-proximately 1 kHz. Broadband infrared imagery at these rates permits characterizationof many types of turbulent flow.

4. RESULTS & DISCUSSION

4.1 Laminar Flame

To demonstrate the utility of IFTS for combustion diagnostics, measurements of a Henc-ken burner were recently acquired (Rhoby and Gross, 2012) and the key results aresummarized here. A Hencken burner produces a nearly ideal adiabatic flame and is rou-tinely used as a calibration standard for testing new combustion diagnostics. In a seriesof experiments, an ethylene (C2H4) / air flame was produced at various equivalence ra-tios3 (Φ). Total volumetric flow rates were between 10.9 SLPM and 17.1 SLPM. Theinstrument collected 1000 spectral images at 1 cm−1 resolution on a 200× 64 pixelarray.

The observed spectra are dominated by broadband emission fromCO2 between 2150and 2400 cm−1. Emissions fromH2O are spectrally structured and are found between3000 and 4200 cm−1; weaker emissions can be found below 2000 cm−1. Spectra fromfuel-rich (Φ > 1) flames exhibitedCO emission lines on either side of the 2143 cm−1

band center. TheCO line intensities increased withΦ. An example spectrum is presentedin Fig. 2.

High-speed imagery was extracted from the interferometric cubes (see Section 3.3)and revealed that the flame was steady up to approximately 30 mm above the burner.Within this region, the flame is stable and nearly homogeneous with a very thin mix-ing layer. However, above 30 mm unsteady behavior was observed as revealed by theinset imagery in Fig. 2. The left panel provides the time-averaged flame intensity andcharacterizes the mean flow field. The middle panel shows the difference between aninstantaneous flame intensity and the mean flame intensity. Variations up to 50% of themean signal are evident. The standard deviation of each pixel’s intensity are provided inthe right panel.

Within this homogeneous portion of the flame, the radiative transfer model [Eq. (2)]can be used to simultaneously retrieve temperature and species concentrations from theobserved spectrum. To validate this approach, an ethylene flame measurement was taken

3The equivalency ratio is defined by the actual fuel:air ratio relative to the stoichiometric fuel:airratio.



0

500

1000

1500

2000 2500 3000 3500 4000 4500

0

20

40

60

80

100

120

140

160

180

200

2,0

00

0 –2

,00

0

–4

,00

0

2,5

00

2,0

00

1,5

00

1,0

00

50

0

4,0

00

2,0

00

6,0

00

10

,00

0

14

,00

0

Distance from center [mm]

He

igh

t a

bo

ve

bu

rne

r [m

m]

Counts [DN]

3100 3200 3300 3400

0

50

100

150

200

250

300

350

400DataModelFit Residual

-50

-20 0 20 -20 0 20 -20 0 20

FIG. 2: Mean single-pixel spectrum of an ethylene flame centered 20 mm above theburner. The large peak at 2250 cm−1 is due toCO2 and the structured emission between3000 and 4200 cm−1 is primarily due toH2O. The inset color panels present (1) thetime-averaged broadband infrared image (left), difference between an instantaneous andthe mean flame image (middle), and the standard deviation of the flame intensity (right).The inset spectrum compares an ethylene center flame spectrum at 10 mm with a modelfit. Fit quality can be judged by the residuals offset by 50µm/(cm2 sr cm−1).

corresponding toΦ = 0.91 via fuel and air flow rates of 0.78 and 12.2 standard litersper minute (SLPM), respectively. This was to permit comparison with measurementsof an identical flame studied using a tunable diode laser absorption4 technique (Meyeret al., 2005). Flame temperature and mole fractions ofH2O andCO2 were estimated bya nonlinear least-squares fit of Eq. (2) to the the IFTS spectrum at 10 mm above flamecenter. These fit parameters were adjusted using a Levenberg-Marquardt algorithm tominimize the sum of squared differences between the measured and model spectrum.The fit results were good as demonstrated in the inset spectrum of Fig. 2. The spectrallyestimated temperature ofT = 2172± 28 K was in excellent agreement with the OHlaser absorption temperature ofT = 2226± 112 K. Optimal concentrations forH2OandCO2 were 13.7± 0.6% and 15.5± 0.8%, respectively, exceeding expected resultsby 20% according to equilibrium calculations. Relative line heights determine the gastemperature, whereas absolute line heights determine species concentrations. The goodagreement in temperature suggests the relative instrument spectral calibration is good.However, the poor agreement in concentration could be caused by a systematic error inthe absolute calibration.

4The laser-based diagnostic measured the shape of a single hydroxy radical (OH) line to extracttemperature.


22 Rhoby et al.

4.2 Jet Engine

Having demonstrated the applicability of IFTS to a laminar flame, we now consider thehighly turbulent flow field produced by a jet engine. Rapid temperature fluctuations inthe flow field produce substantial changes in the instantaneous scene spectrum duringthe course of an interferometric measurement. The SCAs associated with the spectrumfrom a single interferometric cube appear as noise. Time-averaging reduces this “sourcenoise” and produces a recognizable spectrum. However, the quantile analysis discussedin Section 3.2 is evaluated for its utility in reducing SCAs as well as providing informa-tion on temperature fluctuation statistics.

The exhaust plume from a Turbine Technologies SR-30 turbojet was imaged by theIFTS. The SR-30 is a small turbojet designed for educational laboratory work. A single-stage centrifugal compressor operating between 39,000 and 87,000 rpm delivers air tothe 27 cm long× 17 cm diameter engine designed for combusting various fuels includingJet-A, JP-8, diesel, and kerosene. Maximum thrust of the SR-30 is approximately 178Nwith a nominal exhaust temperature of 720◦C; 800 spectra at 25 cm−1 were collected ona 48× 156 pixel window.

The collection of interferometric measurements was sorted into quantilesIq(xi) cor-responding toq ∈ {0.159, 0.5, 0.841}. These quantiles correspond to them − σ, me-dian, andm+σ of a normal distribution characterized by meanm and standard deviationσ. Quantile interferograms were converted to apparent radiance spectra. Plume spectraat all quantiles feature weak broadband emission between 2000 and 2800 cm−1 withlarge emission features arising from thermally excitedCO2. A map of brightness tem-perature5 TB[Lq(ν)] at ν = 2278 cm−1 from the median quantile is presented at thetop of Fig. 3. The plume appears fairly symmetric and spans nearly the full width ofthe FPA. The low-emissivity, polished metal engine appears substantially cooler. Themedian-quantile spectrumLq=0.5(ν) for a center pixel near the jet is also shown. Theimaginary part of the spectrum is also provided and appears as noise, indicating SCAshave been minimized. (In a properly calibrated FTS measurement of a static scene, thesignal is contained in the real part and noise is equitably distributed among the real andimaginary parts. SCAs can be detected by examination of the imaginary part.) Kinetictemperatures could be retrieved from the spectrum using an appropriate radiative transfermodel which properly accounts for the three-dimensional flow field.

At each pixel, the magnitude of temperature fluctuations can be characterized byestimating the standard deviation by differencing two brightness temperature quantiles,i.e.,σ+

B(ν) = TB[Lq=0.841(ν)]− TB[Lq=0.5(ν)]. A map ofσ+B(ν) at ν = 2278 cm−1 is

provided in the bottom panel of Fig. 3.While the map represents fluctuations in brightness temperature and not the gas ki-

netic temperature, the two are connected through the effective spectral emissivity of the

5Brightness temperature is defined byTB [L(ν)] = c2ν/ log[1 + c1ν

3/L(ν)]

wherec1 andc2

are the first and second radiation constants.



10

20

30

40

300

350

400

450

500

550

600

650

Ro

w N

um

be

r

10

20

30

40

0

5

10

15

20

25

30

35

Ro

w N

um

be

r

20 40 60 80 100 120 140

Column Number

Row 24, Column 132

Engine

94 cm

31

cm

Shear layer

Core

2000 2200 2400 2600 2800

0

50

100

150Real

CO2

Imag

FIG. 3: Top panel: Brightness temperatureTB at ν = 2278 cm−1 from the median quan-tile (q = 0.5) spectrum. The inset figure presents the spectrum for a center pixel atengine exit. Bottom panel: Brightness temperature standard deviationσ+

B estimated bydifferencing brightness temperatures from theq = 0.841 andq = 0.5 quantile spectra.Translucent lines are overlaid to distinguish the core and shear layers.

plume. Thus, this image indicates qualitatively the strength of temperature fluctuationsthroughout the plume and reveals asymmetry in the spatial distribution. The fluctuationsare strongest at the shear layer where the hot exhaust gases turbulently mix with the coldambient air. The wedge-shaped core is also evident, and while turbulent, appears lessso than at the shear layer as expected. While nonuniformities along the LOS compli-cate quantitative interpretation, we have demonstrated that IFTS can be used to studyturbulent flows and have presented a novel method to estimate temperature fluctuationstatistics.

Bulk flow field characterization is also possible, as demonstrated in a separate ex-periment. Recently, exhaust from an F109 turbofan engine was imaged with the IFTS(Harley et al., 2012). Examination of the time-averaged spectra from the exhaust plumeindicated that the spectral region above 4200 cm−1 was free of spectral emissions. Sincethe Michelson mirror was scanned at a speed of 0.18 cm s−1 in this experiment, intensityvariations at frequencies above 756 Hz could be attributable to fluctuations in the flowfield. A temporal high-pass filter (Butterworth, 756 Hz cutoff) was applied to the stackof images comprising a single interferometric cube. A sequence of images is providedin Fig. 4 and reveals the dynamic flow. Turbulent eddies are observed to move down


24 Rhoby et al.

256 pixels, 0.26 cm/pixelCamera frame rate: 2860 Hz

1

2

3

4

5

6

7

8

9

FIG. 4: Tracking turbulent eddies enables bulk flow velocity estimation as demonstratedin this sequential imagery of F109 engine exhaust. A Butterworth temporal high-passfilter with cutoff frequency of 756 Hz was applied to the imagery.

stream at a nearly constant velocity. Since the camera frame rate (2860 Hz) and pixeldimensions (0.26× 0.26 cm2) are known, frame-by-frame tracking of one eddy pro-vides a bulk flow velocity estimate of 181 m s−1. This compares well to the exit velocityof 176 m s−1 computed using measured fuel/air mass flow rates and a thermocoupletemperature measurement at the exhaust exit (Harley et al., 2012).

5. CONCLUSIONS

In this paper, we have summarized recent efforts at developing IFTS for combustion andflow field diagnostics. The IFTS enables highly resolved spectra across a wide band-width to be captured at each pixel in an image. We have demonstrated how this enablessimultaneous retrieval of temperature and multiple species concentrations. Moreover, theDC information captured by the focal-plane array in the IFTS yields high-speed, broad-band imagery “for free” enabling characterization of the bulk flow in a dynamic plume.This was used to successfully estimate bulk flow velocity from a jet engine. Additionally,the DC information permits the estimation of spectra at various total-intensity quantiles.These quantile spectra complement the information found in the mean spectrum and en-able qualitative estimates of temperature fluctuation statistics. The wealth of informationthat can be extracted from IFTS measurements of flow fields establishes it as a usefuldiagnostic tool. In particular, IFTS measurements could be used to validate predictionsfrom next-generation plume codes.



REFERENCES

Bowen, S. J., Hyperspectral imaging of a turbine engine exhaust plume to determine radiance,temperature, and concentration spatial distributions, Master’s thesis, Air Force Institute ofTechnology, 2009.

Bradley, K. C., Midwave infrared imaging Fourier transform spectrometry of combustion plumes,PhD thesis, Air Force Institute of Technology, AFIT/DS/ENP/09-S01, 2009.

Chamberland, M., Farley, V., Vallieres, A., Villemaire, A., Belhumeur, L., Giroux, J., andLegault, J.-F., High-performance field-portable imaging radiometric spectrometer technologyfor hyperspectral imaging applications,Proc. SPIE, vol. 5994, p. 59940N, 2005.

Clough, S., Shephard, M., Mlawer, E., Delamere, J., Iacono, M., Cady-Pereira, K., Boukabara, S.,and Brown, P., Atmospheric radiative transfer modeling: A summary of the AER codes,J.Quant. Spectros. Radiat. Transfer, vol. 91, no. 2, pp. 233–244, 2005.

Farley, V., Vallieres, A., Chamberland, M., Villemaire, A., and Legault, J.-F., Performance of theFIRST, a longwave infrared hyperspectral imaging sensor,Proc. SPIE, vol. 6398, p. 63980T,2006.

Harley, J. L., Rolling, A. J., Wisniewski, C. F., and Gross, K. C., Spatially-resolved infraredspectra of jet exhaust from an F109 turbofan engine,Proc. SPIE, vol. 8354, p. 83540H, 2012.

Kohse-Hoinghaus, K. and Jeffries, J. B. (Eds.),Applied Combustion Diagnostics, Taylor & Fran-cis, London, 2002.

Mathieu, J. and Scott, J.,An Introduction to Turbulent Flow, Cambridge University Press, 2000.

Meyer, T. R., Roy, S., Anderson, T. N., Miller, J. D., Katta, V. R., Lucht, R. P., and Gord, J. R.,Measurements of OH mole fraction and temperature up to 20 khz by using a diode-laser-basedUV absorption sensor,Appl. Opt., vol. 44, pp. 6729–6740, 2005.

Moore, E. A., Gross, K. C., Bowen, S. J., Perram, G. P., Chamberland, M., Farley, V., Gag-non, J.-P., and Villemaire, A., Characterizing and overcoming spectral artifacts in imagingFourier-transform spectroscopy of turbulent exhaust plumes,Proc. SPIE, vol.7304, p. 730416,2009.

Rhoby, M. R. and Gross, K. C., Mid-wave IFTS measurements of a laboratory-scale laminarflame,Proc. SPIE, vol. 8390, p. 83900E, 2012.

Rothman, L., Gordon, I., Barbe, A., Benner, D., Bernath, P., Birk, M., Boudon, V., Brown, L.,Campargue, A., Champion, J.-P., Chance, K., Coudert, L., Dana, V., Devi, V., Fally, S.,Flaud, J.-M., Gamache, R., Goldman, A., Jacquemart, D., Kleiner, I., Lacome, N., Lafferty, W.,Mandin, J.-Y., Massie, S., Mikhailenko, S., Miller, C., Moazzen-Ahmadi, N., Naumenko, O.,Nikitin, A., Orphal, J., Perevalov, V., Perrin, A., Predoi-Cross, A., Rinsland, C., Rotger, M.,Simeckova, M., Smith, M., Sung, K., Tashkun, S., Tennyson, J., Toth, R., Vandaele, A., andAuwera, J. V., The HITRAN 2008 molecular spectroscopic database,J. Quant. Spectros. Ra-diat. Transfer, vol. 110, no. 9-10, pp. 533–572, 2009.

Rothman, L. S., Gordon, I. E., Barber, R. J., Dothe, H., Gamache, R. R., Goldman, A.,Perevalov, V. I., Tashkun, S. A., and Tennyson, J., HITEMP, the high-temperature molec-ular spectroscopic database,J. Quant. Spectros. Radiat. Transfer, vol. 111, pp. 2139–2150,2010.


26 Rhoby et al.

Thomas, G. E. and Stamnes, K.,Radiative Transfer in the Atmosphere and Ocean, CambridgeUniversity Press, Cambridge, 2002.

Tremblay, P., Gross, K. C., Farley, V., Chamberland, M., and Villemaire, A., Understanding andovercoming scene-change artifacts in imaging Fourier-transform spectroscopy of turbulent jetengine exhaust,Proc. SPIE, vol. 7457, p. 74570F, 2009.


imaging fourier-transform spectrometry for plume ... · imaging fourier-transform spectrometry for...

Documents