Photoacoustic insight for aerosol light absorption aloft from meteorological aircraft and comparison with particle soot absorption photometer measurements: DOE Southern Great Plains climate research facility and the coastal stratocumulus imposed perturbation experiments

Photoacoustic insight for aerosol light absorption aloft from

meteorological aircraft and comparison with particle soot absorption

photometer measurements: DOE Southern Great Plains climate

research facility and the coastal stratocumulus imposed perturbation

experiments

W. Patrick Arnott,

John W. Walker,

Hans Moosmu

̈ller,

Robert A. Elleman,

Haflidi H. Jonsson,

Gintautas Buzorius,

William C. Conant,

Richard C. Flagan,

and John H. Seinfeld

Received 10 March 2005; revised 16 June 2005; accepted 10 August 2005; published 24 January 2006.

[

]

Aerosol light absorption can be intense close to local sources such as wildland and oil

fires, with smoke that disperses into the boundary layer and, with enough lift, into the

upper atmosphere where it may be transported around the globe. Filter-based methods

such as the Particle Soot Absorption Photometer (PSAP) are most commonly used to

quantify aerosol light absorption aloft. This paper reports first measurements of aerosol

light absorption aloft with photoacoustic instrumentation (PA). Three examples of aerosol

light absorption are presented. The first one illustrates a case of detached layers aloft

arising from intercontinental, interoceanic transport of smoke from wildland fires in

Siberia to the North American continent and the measurement campaign held at the

Department of Energy Atmospheric Radiation Measurement Program Climate Research

Facility in north central Oklahoma. Then, two examples of intense local fire smoke light

absorption from the Coastal Stratocumulus Imposed Perturbation Experiment near

Marina, California, USA, are presented. The first local fire was an oil fire burning in a

storage tank near Moss Landing, California, USA, and smoke from this fire was very dark,

indicating a low single scattering albedo. By contrast, the second local fire was

predominantly burning wood, vegetation, and structures near Fort Ord in Marina,

California, USA, and the smoke was very bright, indicating a high single scattering

albedo. In all examples, PA measurements at 676 nm were compared with those from a

PSAP modified to measure at three wavelengths, including 660 nm.

Citation:

Arnott, W. P., J. W. Walker, H. Moosmu

̈ ller, R. A. Elleman, H. H. Jonsson, G. Buzorius, W. C. Conant, R. C. Flagan, and

J. H. Seinfeld (2006), Photoacoustic insight for aerosol light absorption aloft from meteorological aircraft and comparison with particle

soot absorption photometer measurements: DOE Southern Great Plains climate research facility and the coastal stratocumulus imposed

perturbation experiments,

J. Geophys. Res.

111

, D05S02, doi:10.1029/2005JD005964.

1. Introduction

[

] Aerosol light absorption and scattering are the main

pathways for particulate interaction with atmospheric radi-

ation. Regions downwind of major urban areas are impacted

by light absorbing aerosol, chiefly black carbon from

combustion sources, and health, radiation, and cloud life-

time issues are associated with these particles [

Andreae

2001]. These particles may mix into the atmospheric

boundary layer, or get lofted into the upper atmosphere by

convective processes. Wildland fires also produce black

carbon that may very well be transported into the upper

atmosphere during large fires. In these cases where aerosol

are in the upper atmosphere their characterization cannot be

inferred by in situ measurements at ground level.

[

] Black carbon absorbs light across the entire solar

spectrum and typically has an inverse wavelength depen-

dence where light absorption at blue wavelengths is greater

than it is at red wavelengths. This is not always the case, as

was noted recently [

Kirchstetter et al.

, 2004] where evi-

dence was presented that light absorption by wildland fire

smoke may have a very different wavelength dependence.

Aerosol light absorption, both on the ground and aloft, is

most often measured using filter-based methods such as the

Particle Soot Absorption Photometer (PSAP), and a new

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111, D05S02, doi:10.1029/2005JD005964, 2006

Division of Atmospheric Sciences, Desert Research Institute, Reno,

Nevada, USA.

Department of Atmospheric Sciences, University of Washington,

Seattle, Washington, USA.

Naval Postgraduate School, Monterey, California, USA.

California Institute of Technology, Pasadena, California, USA.

D05S02

1of16

three-wavelength instrument has been developed for both

applications [

Virkkula et al.

, 2005].

[

] Measurements aloft are usually undertaken on rare

occasions when groups assembled to analyze particular

hypotheses initiate an intensive operational period. An

exception to this is the program at the Department of Energy

Atmospheric Radiation Program Climate Research Facility

Southern Great Plains site, denoted simply as the SGP in the

remainder of this paper. Aerosol light scattering and PSAP

absorption measurements are measured routinely in the

boundary layer over the SGP using an unpressurized sin-

gle-engine light aircraft. This program may grow in coming

years, both in terms of the number of measurements and in

the aircraft capability. One finding of this routine sampling

program aloft is an indication that aerosol single scattering

albedo systematically diminishes with height. Single scat-

tering albedo is defined as the ratio of light scattering to the

sum of light scattering and absorption, also known as

extinction. The significance of this finding is that such an

effect would stimulate relative heating of the atmosphere at

levels above the ground where clouds try to form, and

would tend to shadow the ground from direct radiative

forcing by sunlight with potential implications to reductions

of photosynthetic plant growth and the initiation of convec-

tion. Aerosol amount at the SGP is usually modest, though

can be larger during episodes where farmers burn local

fields [

Sheridan et al.

, 2001;

Arnott et al.

, 2003b]. Aerosol

amount, and the potential for strong aerosol impacts, is

much larger in places like Brazil where the rain forest is

being cleared by the slash and burn method [

Reid et al.

1998].

[

] The most common method of aerosol light absorption

measurement involves use of a filter to sample ambient air

and an optical source and detector to determine the change

in filter transmittance or reflectance due to particulate matter

deposited on the filter since the previous optical measure-

ment. Filter optical transmittance measurement is the most

common of analytical method and single- and multiple-

wavelength versions of these instruments are commercially

available, called Aethalometers by Magee Scientific, and

PSAP’s by Radiance Research. Aethalometers are calibrated

by the manufacturer to provide black carbon mass concen-

tration, though recent efforts have sought to interpret data

from these instruments also in terms of aerosol light

absorption [

Weingartner et al.

, 2003;

Arnottetal.

2005b]. Single-wavelength PSAP’s have been calibrated

to provide aerosol light absorption measurements [

Bond et

al.

, 1999], and a new three-wavelength prototype has been

developed and calibrated [

Virkkula et al.

, 2005] during the

Reno Aerosol Optics Study [

Sheridan et al.

, 2005]. It is this

new prototype that will be compared in this paper with

photoacoustic measurements of aerosol light absorption

aloft. It has been recognized that the optical characterization

of particles on quartz fiber filters common to these

instruments formally involves multiple scattering theory

[

Horvath

, 1997;

Gorbunov et al.

, 2002], and that non-

absorbing aerosol gives rise to an apparent absorption

equal to roughly 2% of the scattering coefficient [

Bond et

al.

, 1999;

Arnott et al.

, 2005b;

Virkkula et al.

, 2005]. A

strong motivation for use of the photoacoustic instrument

on aircraft is to provide a separate method for aerosol

light absorption measurement that does not employ filters,

and a method that can have its calibration evaluated using

light absorbing gases such as NO

[

Arnott et al.

, 2000;

Sheridan et al.

, 2005] to compare with the more com-

monly used PSAP, and to characterize the atmosphere.

[

] It should be noted that filter-based measurements of

aerosol light absorption are quite sensitive, as the multiple

scattering enhancement of light absorption by the filter

substrate amplifies the absorption signal, yet the challenge

is to provide suitable calibration. The largest remaining

unknown factor is the effect of particle distribution in the

filter since a monolayer of particles sitting on the filter

surface will have an optical effect quite different from the

same particles embedded uniformly throughout the filter. A

new filter-based instrument has been developed as the

Multiple Angle Absorption Photometer (MAAP) by

Thermo Electron that measures both filter transmission

and reflection at numerous angles to better constrain the

energy budget for the optical interaction of the filter and

particles [

Petzold and Scho

̈nlinner

, 2004;

Petzold et al.

2005] by obtaining absorptivity from (1

filter transmis-

sivity

filter reflectivity). The MAAP does not require

ancillary use of a nephelometer for determination of the

scattering coefficient needed to correct filter transmission

measurements as does the PSAP and the Aethalometer.

[

] Surface measurements of aerosol light absorption in

the ambient, and from source samples, by the photo-

acoustic method, have been reported by several groups

[

Bruce and Pinnick

, 1977;

Terhune and Anderson

, 1977;

Japar and Killinger

, 1979;

Roessler and Faxvog

, 1980;

Japar and Szkarlat

, 1981;

Roessler

, 1984;

Adams et al.

1990a, 1990b;

Petzold and Niessner

, 1992, 1994, 1995;

Moosmu

̈ller et al.

, 1998;

Arnott et al.

, 1999, 2003b,

2005b]. While this list covers the flavor of most previous

efforts, it is not meant to be complete. However, no references

have been found for previous measurements of aerosol light

absorption aloft from a meteorological aircraft by the photo-

acoustic method, but are reported in this paper. It should be

noted that a remarkable report has been made of photo-

acoustic measurements of water vapor and NO concentrations

at an altitude of 28 km using a large tethered balloon system

that was operated by the National Center for Atmospheric

Research and flew out of Palestine Texas [

Patel et al.

, 1974].

[

] An extensive literature exists on photoacoustic trace

gas detection and it is relevant to discuss the general merits

of this spectroscopic method. Use of high-quality micro-

phones (e.g., 1

diameter capacitive versions that polarize

the plates with an externally provided 200 V) as detectors

gives a dynamic range for acoustic pressure measurements

from around

40 dB (Re 20

Pa) to around 135 dB (Re

Pa). The acoustic signal at the low end of this range is

Pa, and as light absorption levels aloft generally

are low, it is a challenge to accurately quantify such low

sound pressure values. These detectors are completely color-

blind in that they can be used with radiant energy at any

wavelength, are stable over many years, and a well-aligned

unit has negligible background. Relatively simple acoustical

shielding and use of analog acoustic filters to block certain

frequency ranges can allow these instruments to be used for

source sample measurements for even very loud tethered jet

aircraft [

Arnott et al.

, 2005a]. At the low signal level phase

sensitive lock-in amplifier detection and large laser power

are additionally needed to optimize sensitivity.

D05S02

ARNOTT ET AL.: BLACK CARBON LIGHT ABSORPTION ALOFT

2of16

D05S02

[

] This paper is organized as follows. Section 2 covers

the acoustical issues associated with use of resonant photo-

acoustic instruments aloft in an environment where ambient

pressure can change rather quickly. This discussion is rather

technical and is specialized toward instrument issues so the

reader eager for intercomparisons of filter-based measure-

ments with photoacoustic values can skip this section on a

first read and go directly to the next section. Examples are

given in this next section illustrating measurements aloft

above the atmospheric boundary layer of distinct layers of

aerosol light absorption from wildland fires at source

locations an ocean and 1/2 of a continent away from the

observation area. Further examples illustrate use of the

instrument to characterize locally intense plumes of aerosol

light absorption from very dark smoke emitted during a

fuel oil fire and from very bright smoke from a wildland

fire.

2. Resonant Photoacoustics Aloft: Acoustical

Considerations

[

] Photoacoustic instruments have been used both in

source and ambient sampling of light absorbing aerosol.

Sample air is pulled continuously through an acoustical

resonator and is illuminated by laser light that is power

modulated at the acoustical resonance frequency. Light

absorption results in particle heating and this heat transfers

rapidly to the surrounding air, inducing periodic pressure

fluctuations that are picked up with a microphone on the

resonator. This is a rather generic description of the photo-

acoustic method, though key points about the ability to

sample continuously, to accommodate microphone calibra-

tion by balancing the static pressure on both sides of the

membrane, and to acoustically shield the instrument from

background noise and coherent window light absorption

signals can be found elsewhere [

Rosengren

, 1975;

Arnott et

al.

, 1999, 2003a, 2005a].

[

] The photoacoustic instrument operates at a conve-

nient wavelength of 676 nm where gaseous interference is

negligible and where a diode laser source is available that

allows for direct electronic modulation of the optical power

at the resonator frequency. The laser power is around 500 mW

at the operating acoustical resonance frequency of nominally

1500 Hz, and the laser is fiber-coupled to the resonator. The

multimode fiber output is collimated with a lens attached to

the fiber end. The lens is attached directly to the resonator to

minimize effects of aircraft vibration on optical align-

ment, though as the beam did not proceed completely

cleanly through the resonator a background signal of

around 10 Mm

was observed from absorption of light

by the resonator walls. A misaligned system produces a

very large amount of stray light in comparison with the

aligned system where light scattering particles are present.

The lens system used at the time of these experiments

had no alignment capacity, though in later work employ-

ing a more sophisticated lens system, stray radiation has

been eliminated.

[

] It has been demonstrated that ammonium sulfate

aerosol of a pure scattering nature at 532 nm, with negli-

gible light absorption, produce negligible photoacoustic

signals [

Arnott et al.

, 2005b;

Sheridan et al.

, 2005]. The

absorption measured by the photoacoustic instrument for

ammonium sulfate aerosol was perhaps 0.77 Mm

and the

scattering coefficient was 519 Mm

for a single scattering

albedo of 0.9985. By comparison, the Aethalometer had a

filter attenuation coefficient of around 27 Mm

at 521 nm

while the scattering-corrected PSAP had an absorption

coefficient of 2.2 Mm

at 550 nm for this case.

[

] In any case, the slightly misaligned photoacoustic

instrument with a relative broad laser beam diameter neces-

sitated background subtraction through the use of a motor-

ized valve that switched between sample air and air filtered

of particles, and the valve position was automatically

changed under computer control as part of the operation

of the instrument control software. Data were acquired at a

rate of 1 Hz and instrument zeros were obtained every 500

measurements. The background measurement was time

averaged over 10 measurements to improve the signal-to-

noise ratio for background determination. It should be noted

that in other instances a properly aligned system with a

clean laser beam from a solid state laser at 532 nm exhibited

negligible background so no background subtraction was

necessary [

Arnott et al.

, 2003b].

[

] Photoacoustic signal background could generally

arise from two sources. Electronic background could arise

from bleed of the modulation square wave signal to the

microphone data acquisition channel. This background

source would be independent of pressure and temperature.

On the other hand, stray laser light being absorbed some-

where, for example on a window or on the resonator wall,

could give rise to a background photoacoustic signal that is

dependent on the thermodynamic conditions of the gas in

the resonator. Extensive tests have demonstrated that the

electronic background is negligible. The goal of the remain-

ing section is to develop an understanding of the operation

of the photoacoustic instrument aloft where pressure, rela-

tive humidity, and ambient temperature may change dra-

matically on timescales of minutes. Photoacoustic light

absorption is obtained from

abs

mic

Laser

res

ðÞ

;

;

ðÞ

;

;

ðÞ

mic

Laser

res

;

where

mic

is the microphone pressure,

laser

is the laser

power,

res

is the cross-sectional area of the resonator,

the ratio of isobaric to isochoric specific heats,

is the

resonance frequency, and

is the quality factor of the

resonator [

Arnott et al.

, 1995, 1999, 2000, 2003b;

Raspet et

al.

, 2003]. The first form of equation (1) indicates the

explicit dependencies on pressure, temperature, and RH.

Relative humidity, temperature, and pressure are measured

downstream of the photoacoustic resonator. The resonance

frequency and quality factor are also measured every 100 to

200 s using a piezoelectric transducer attached to the

resonator. The resonance scan is accomplished in about 3 s

as 5 points are used at different frequencies along with a

routine to determine the resonance frequency and quality

factor from a fit of the resulting curve to a theoretical

expression with 3 parameters. The third parameter is the

peak acoustic pressure at resonance. Though it is not needed

in equation (1) to determine light absorption, it is very useful

to have available as a means of quantifying microphone

calibration and performance over time, and its dependence on

D05S02

ARNOTT ET AL.: BLACK CARBON LIGHT ABSORPTION ALOFT

3of16

D05S02

ambient conditions is a good check on the viability of making

sound pressure measurements aloft.

[

] Equation (1) indicates that several parameters are a

function of relative humidity (RH), temperature and pres-

sure. It is instructive to consider the dependence of these

thermophysical parameters on environmental conditions of

temperature, pressure and

. First, define the fraction of

water vapor molecules as

, and note that it is given by

:

ðÞ

;

where the saturation vapor pressure of water vapor at

temperature

ðÞ¼

:

ðÞ

exp

ðÞ

;

with

= 17.269,

= 35.860, and

= 273.15 for

273.15 K. The value of

for moist air is

:

Consider a relatively extreme example of air being fully

saturated at temperature of 30 C and a total air pressure

of 500 mb. The saturation vapor pressure is

42 mb,

and

= 0.084. The value of (

1) for dry air is 0.4,

while for moist air it is 0.3934, for a percentage

difference of less than 2%. In other words, the variation

with air pressure and relative humidity in the

photoacoustic equation (1) is generally negligible, though

of course one could easily take it into account since the

requisite measurements to obtain it are available.

[

] The speed of sound in moist air is

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1000

;

where

is the universal gas constant and MKSA units are

used [

Bohn

, 1988]. The denominator of equation (5) is the

average molecular weight of moist air, and equation (4)

arises from the water vapor molecules having one more

degree of rotational freedom than the main diatomic

constituents of air. The density of moist air,

,is

Pmb

ðÞ

:

:

ðÞ

:

The resonator quality factor can be expressed as

Acoustic Power Lost per Cycle

Energy Stored

transport

loss

ðÞ

loss

;

where acoustic boundary layer losses occur in the thermal

and boundary layer thickness

ffiffiffiffiffiffiffiffiffi

;

ffiffiffiffiffiffiffiffiffiffiffiffiffi

ffiffiffiffiffiffiffi

;

respectively, and

= 0.838 cm is the resonator radius, and

= 22.5 cm is the resonator length. The term involving

resonator length is due to dissipation of thermal energy in

the boundary layer at the ends of the resonator. No viscous

energy is dissipated here. A similar expression applies to

radial wave acoustical resonators [

Kamm

, 1976]. In

equation (8),

is the viscosity of the moist air,

is the

thermal conductivity,

is the isobaric heat capacity per

unit mass, and

is the Prandtl number. The

Prandtl number of dry air at standard conditions is

approximately 0.7. Other fractional losses in equation (7)

might include microphone flexing due to its compliance,

bulk acoustic losses in the gas mixture, and possibly other

fluid dynamical motions of the gas such as vortex

generation at the relatively sharp corners in the resonator

section that occur where the resonator takes a perpendicular

corner [

Arnott et al.

, 2003a, 2005a]. In general, the other

fractional losses in the bulk of the gas [

Johnson et al.

1981] are due to free-space viscous and thermal damping

and are small compared to the losses of fluid kinetic energy

associated with the viscous boundary layer and fluid

potential energy associated with the thermal boundary layer

at the resonator wall, as the Q is a relatively modest value

of around 73.

[

] The resonator quality factor associated with transport

losses of thermal conduction and viscosity can be expressed

transport

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Pmb

ðÞ

:

:

ðÞ

ffiffiffiffiffiffiffi

;

neglecting the relatively small term in equation (7)

associate with the resonator length. The resonator is

operated in an acoustical mode such that one full acoustic

wavelength is spanned by the resonator length

22.4 cm. For example, the measured value in equation (7)

73, with calculated

transport

= 98 and inferred

loss

= 300 for one atmosphere pressure and 20 C. The

condition for resonance can be approximated roughly by

considering wave propagation in a lossy resonator [

Arnott

et al.

, 1996], and taking the real part of the complex

propagation constant times the resonator length equal to

. This approach assumes that the boundary conditions

at the resonator ends are that the acoustic velocity goes to

zero. However, because of dissipation of acoustic potential

energy at the resonator terminations in the thermal boundary

layers at the microphone and piezoelectric transducer, this

boundary condition is only approximate. This effect is small

for the resonator discussed here. The resonance frequency can

be expressed in terms of the sound speed, quality factor, and

resonator length as

transport

;

assuming that the frequency shift is due entirely to wall

absorption [

Johnson et al.

, 1981]. A lossless acoustical

resonator would have resonance frequency

[

] Equations (9) and (10) show that the primary

variation of the resonator quality factor and resonance

D05S02

ARNOTT ET AL.: BLACK CARBON LIGHT ABSORPTION ALOFT

4of16

D05S02