Heard D.E. (editor) Analytical Techniques for Atmospheric Measurement

Подождите немного. Документ загружается.

478 Analytical Techniques for Atmospheric Measurement

298 nm 305 nm 310 nm 325 nm

380

420

25 25

Actinic flux (photons cm

–2

–1

)

25 2510

10 June

Measured

Simulated

Altitude (km)

Figure 9.41 Measured and simulated vertical proﬁles of the actinic ﬂux over the Agean Sea in Greece

for selected wavelengths on a cloudless day (10 June, 1996). The solar zenith angle was between 17



and



during the ﬂight measurements (adapted from Hofzumahaus et al., 2002, printed with permission

from American Geophysical Union).

in Figure 9.42. A strong reduction of measured jNO

 values caused by an optically

thick cloud layer was observed in the lowest troposphere. The gradual jNO

 decrease in

the cloud layer is well described by model simulations which used micrometeorological

measurements of cloud droplets and particles as model input. When compared to clear-

sky conditions, an increase of the actinic flux above clouds and a strong decrease below

clouds can be expected (Madronich, 1987b; Van Weele & Duykerke, 1993). Measurements

have shown, for example, enhancements above clouds as high as +200% and attenuations

below clouds up to −90% (Lefer et al., 2003). While the influence of closed cloud layers

on the actinic flux can be described consistently by one-dimensional RT models, the

effect of broken clouds cannot be well represented by the same models (Kylling et al.,

2005). Here, more research is needed to better understand the three-dimensional effects

of clouds on the radiation field.

9.8.3 Instrumental intercomparisons

The performance of actinic-flux-sensitive measurement instruments was investigated in

various field comparisons. Radiometry-based techniques were tested against chemical

actinometers for jO

D (Müller et al., 1995; Hofzumahaus et al., 2004) and jNO



(Kraus et al., 1998, 2000; Shetter et al., 2003). Other investigations involved radiometric

Measurement of Photolysis Frequencies in the Atmosphere 479

j (NO

) (10

–3

–1

)

Height (km)

2.0

1.5

1.0

0.5

0.0

0246810

Figure 9.42 Vertical proﬁle of jNO

 (symbols) measured by a broadband ﬁlter radiometer near

Dresden, Germany, on a cloudy day (9 December, 1996) at a solar zenith angle  =74



. The solid and

dotted lines represent simulations with two different radiative transfer models. The cross-hatched area

indicates an optically thick cloud layer (adapted from Früh et al., 2000, printed with permission from

American Geophysical Union).

techniques only, comparing spectrally resolved actinic fluxes (e.g. Webb et al., 2002b;

Bais et al., 2003; Edwards & Monks, 2003; Eckstein et al., 2003) or results for jO

D and

jNO

 (e.g. Kraus & Hofzumahaus, 1998; Lefer et al., 2001; Eckstein et al., 2003; Kanaya

et al., 2003).

The most comprehensive field comparison so far, the International Photolysis

Frequency Measurement and Modeling Comparison (IPMMI), was performed in Boulder,

Colorado, on four days with cloudy and clear-sky conditions in summer 1998 (see

overview by Cantrell et al., 2003). The assessment was carried out as a blind inter-

comparison that included chemical actinometers, spectroradiometers, and broadband

filter radiometers from eight international research groups. Some examples of diurnal

profiles of jO

D and jNO

 measured by actinic-flux spectroradiometers and chemical

actinometers on a clear day are shown in Figure 9.43.

The main results of the experimental data assessment can be summarised as follows:

•

Actinic-flux measurements were compared for three spectroradiometers. Two double-

monochromator instruments agreed to within ±6% independent of solar zenith angle

and wavelength (300–420 nm) when the measurements were transformed to a common

spectral resolution. A third spectroradiometer which was based on a single monochro-

mator showed significant disagreements (up to 30%) with the two other spectrora-

diometers as a result of instrumental stray light. For further details, see Bais et al.

(2003); Edwards and Monks (2003).

480 Analytical Techniques for Atmospheric Measurement

48 812 16 20

CA-NCAR

SR-NCAR

SR-FZJ

Daytime (MDT)

4121620

CA-NCAR

CA-UMD

SR-NCAR

SR-FZJ

Daytime (MDT)

j (NO

) (10

–3

–1

)

D) (10

–5

–1

)

Figure 9.43 Comparison of diurnal proﬁles of photolysis frequencies measured by different chemical

actinometers (CA) and actinic-ﬂux spectroradiometers (SR) during the IPMMI ﬁeld campaign in Boulder,

Colorado, on 19 June 1998 (Shetter et al., 2003; Hofzumahaus et al., 2004). The cross sections and

quantum yields used for the evaluation of the spectroradiometer data are in agreement with the IUPAC

recommendations from 2004 (Atkinson et al., 2004).

•

General good agreement for jO

D within 10% was found among filter radiometers and

spectroradiometers at solar zenith angles less than 60



, provided that the instruments

used the same cross sections and quantum yields. The deviations were generally larger

(up to a factor of 2) when the zenith angles were larger than 60



. Two spectroradiometers

and one filter radiometer showed excellent agreement within 5% with the chemical

actinometer for zenith angles less than 80



, when recently published O

D quantum

yield data were used and when the temperature dependence of jO

D was explicitly

taken into account. For further details, see Hofzumahaus et al. (2004).

•

Measurements of jNO

 from filter radiometers, spectroradiometers and two chemical

actinometers showed consistent agreement within the stated instrumental uncer-

tainties (from 8 to 20%) at zenith angles below 70



. Spectroradiometers and chemical

actinometers agreed within less than 10% when more recent NO

absorption cross

sections from the literature were used. Filter radiometers appeared to be generally less

accurate, with larger discrepancies (20–50%) at larger zenith angles above 70



. For

further details, see Shetter et al. (2003).

9.8.4 Molecular data assessments

The determination of photolysis frequencies from radiation spectra requires absorption

cross sections  T  and quantum yields  T  of the photodissociation reactions

of interest. In general, there is a choice between published molecular data from different

research studies that sometimes show discrepancies larger than their specified uncer-

tainties. This is of considerable relevance in actinic-flux spectroradiometry, because the

absorption cross sections and quantum yields make the largest contribution to the error

of derived photolysis frequencies (Kraus & Hofzumahaus, 1998; Cantrell et al., 2003) (see

also Table 9.6). For a decision on which data are most suitable and accurate, it is generally

a good idea to consult the recommendations which are periodically published and

Measurement of Photolysis Frequencies in the Atmosphere 481

updated by institutions like the Jet Propulsion Laboratory (JPL) under contract of the

American National Aeronautics and Space Administration (NASA) or the Subcommittee

on Gas Kinetic Data Evaluation for Atmospheric Chemistry of the International Union

on Pure and Applied Chemistry (IUPAC). Both institutions make their own evalua-

tions available via the World Wide Web (http://jpldataeval.jpl.nasa.gov and www.iupac-

kinetic.ch.cam.ac.uk).

For the two most important tropospheric photolysis frequencies, jO

D and jNO

,

the accuracy of the molecular data was explicitly tested in field experiments by comparing

chemical actinometer measurements against photolysis frequency data that were derived

from measured actinic-flux spectra and different sets of absorption spectra and quantum

yields (Müller et al., 1995; Shetter et al., 1996; Kraus et al., 2000; Cantrell et al., 2003;

Shetter et al., 2003; Hofzumahaus et al., 2004).

9.8.4.1 Ozone photodissociation

The most comprehensive study for jO

D was performed in the IPMMI field campaign

(see Section 9.8.3) over a temperature range of 5–45



C and zenith angles from 165



to 90



(Cantrell et al., 2003; Hofzumahaus et al., 2004). The derived jO

D data were

found to have little sensitivity <3% to the choice of literature ozone absorption cross

sections, which were chosen for example from Bass and Paur (1985), Molina and Molina

(1986), and Malicet et al. (1995). A large sensitivity, however, was found to be the choice

of O

D quantum yields which were taken from different releases of the NASA/JPL

recommendations from 1994 to 2003. Compared to chemical actinometry, the differences

in the derived jO

D values were generally found to increase with the solar zenith angle.

The largest differences were observed in case of the recommendation from 1994 (DeMore

et al., 1994), yielding discrepancies over a factor of 2 at low sun  > 70



. The application

of the most recent recommendation for the O

D quantum yield (Matsumi et al., 2002;

Sander et al., 2003; Atkinson et al., 2004) gave the best absolute agreement between

chemical actinometry and spectroradiometry, with very small systematic deviations (<5%)

at zenith angles <80



(Figure 9.43). These deviations were smaller than the specified

instrumental uncertainty (10%) of the chemical actinometer used for the study.

9.8.4.2 Nitrogen dioxide photodissociation

The cross sections and quantum yields of the NO

photolysis were tested in two field

studies, JCOM97 (Kraus et al., 2000) and IPMMI (Cantrell et al., 2003; Shetter et al.,

2003). The use of absorption cross sections (Davidson et al., 1988) and quantum yields

(Gardner et al., 1987; Roehl et al., 1994) recommended by NASA/JPL (DeMore et al.,

1997; Sander et al., 2003) led in both studies to spectroradiometer results that were

consistently higher than the chemical actinometer measurements by 10–20%. Improved

agreement with deviations generally less than 10% was obtained in both studies when

IUPAC recommended data (Atkinson et al., 2004) were applied, favouring more recent

measurements of the absorption cross sections in accord with the Merienne et al. (1995)

data and revised quantum yields from Troe (2000).

482 Analytical Techniques for Atmospheric Measurement

9.9 Conclusions and outlook

Various experimental techniques are now available for routine measurements of

atmospheric photolysis frequencies. The most direct method is chemical actinometry,

which has been applied in field campaigns almost exclusively for the photolysis of O

and NO

. It is calibrated by a gas-phase standard and has the particular advantage that it

does not require the knowledge of cross sections and quantum yields. The requirement to

handle pressurised gases and the instrumental difficulty of detection of reactive photolysis

products is probably the reason why it is not more widely used for other photolysis

reactions.

The most versatile method is actinic-flux spectroradiometry. It measures the spectrally

resolved actinic flux and can be used to determine photolysis frequencies of a variety

of chemical compounds from the same measured radiation spectrum. As a further

advantage, the absolute calibration is traceable to very accurate radiation standards

from national laboratories. For the determination of photolysis frequencies, the method

requires knowledge of absorption cross sections and quantum yields. Although the

relevant photochemical data have been measured thoroughly, they remain the largest

source of uncertainty in the calculation of photolysis frequencies.

The quality of photochemical data for O

and NO

has been improved greatly owing

to new laboratory studies. The improvement is consistent with the outcome of field

comparisons between chemical actinometry and actinic-flux spectroradiometry, which

agree well when the applied cross sections and quantum yields comply with the most

recent IUPAC recommendations (Atkinson et al., 2004). The comparisons also demon-

strate that jO

D and jNO

 can be measured with spectroradiometers as accurate as by

chemical actinometers.

In the future it would be desirable to test critically the photochemical database for other

important molecules, like formaldehyde, and perform corresponding field comparisons

with chemical actinometers.

Broadband filter radiometers are useful, because they are small and can be run

automatically with high time resolution. However, their calibration is laborious and their

data evaluation requires careful corrections for their imperfect spectral matching of the

respective photolysis processes.

Spectroradiometers developed at the time of writing are based on multichannel-detector

spectrograph systems. They combine the advantages of spectroradiometry with the small

size and high time resolution known from filter radiometers. Spectrograph instruments

offer the largest potential for field studies; however, unsolved stray light problems preclude

at present their usefulness in the UV-B region. Their further development will significantly

facilitate the fast measurement of actinic radiation and photolysis frequencies, in particular

on aircraft and under cloudy conditions.

Acknowledgements

The author gratefully acknowledges fruitful cooperation with present and previous co-

workers and colleagues in the field of photolysis frequency measurements, in particular

Birger Bohn, Theo Brauers, Alexander Kraus, Martin Müller, and Franz Rohrer;

Measurement of Photolysis Frequencies in the Atmosphere 483

furthermore Arve Kylling, Christos Zerefos, Rick Shetter and the colleagues who

contributed to the IPMMI project, and Rainer Schmitt at Meteorologie Consult GmbH.

He also likes to thank Gus Hancock for the stimulating and fruitful collaboration on the

photolysis of ozone. Many thanks to Birger Bohn for critical reading of the manuscript

and Dwayne Heard for his encouragement in writing this chapter.

Appendix

A.1 Radiation quantities and their relationships

Basic properties of radiance, irradiance, and actinic flux and their relations relevant for

the determination of atmospheric photolysis frequencies have been discussed for example

by Madronich (1987b) and Van Weele et al. (1995). This section summarises some of

the important relationships.

The radiance in the atmosphere can be split into three components:



= L

+L ↓+L ↑ (A.1)

where L

is the spectral radiance of the direct sunlight, L ↓ represents the downwelling

diffuse sky radiation, and L ↑ describes the upwelling diffuse radiation. When

equation A.1 is inserted into equation 9.10, the actinic flux can be partitioned accordingly

into a direct flux F

, a diffuse downwelling flux F ↓, and a diffuse upwelling flux F ↑:



= F

+F ↓+F ↑ (A.2)

In a similar way, the downward irradiance (cf. equation 9.14) can be written as the sum

of the direct downward irradiance E

and the diffuse downward irradiance E ↓:



= E

+E ↓ (A.3)

The upward directed irradiance E ↑ is related to the downward irradiance by a surface

albedo A

, which is defined as:

E ↑

+E ↓

(A.4)

By definition the direct actinic flux F

is related to the direct irradiance E

by the cosine

of the solar zenith angle :

cos

(A.5)

Ratios can also be defined for the upwelling and downwelling diffuse components:

F ↓

E ↓



L ↓   d



L ↓   cos d

(A.6)

F ↑

E ↑



L ↑   d



L ↑   cos d

(A.7)

484 Analytical Techniques for Atmospheric Measurement

Here, the ratios r

and r

can be calculated only if the angular distribution for the

corresponding diffuse radiance is known.

Inserting equations A.4–A.7 into equation A.2 allows to express F



in terms of the

downwelling direct and diffuse irrandiance components:



cos 

E ↓+r

E

+E ↓ (A.8)

Next the parameter 

is defined as the direct fraction of the total downward irradiance:



+E ↓



(A.9)

Combining Eqs. A.8 and A.9, a general relationship between F



and E



can be derived:





cos

1 −

 +r

= 



cos 

−r







(A.10)

It is often assumed that the reflecting surface giving rise to the upward directed radiance

behaves like a Lambertian reflector, that is produces upward radiance without angular

dependence. For this special case (isotropy assumption) r

is calculated to be

F ↑

E ↑

= 2 (A.11)

Then equation A.10 assumes the following form:



= 



cos

−r





+2A



(A.12)

This is the basic equation for estimating actinic fluxes from downward irradiances above

Lambertian surfaces with albedo A

Another useful equation can be derived from equations A.4 and A.7 for the upward

actinic flux F ↑ above Lambertian surfaces r

= 2:

F ↑= 2A

E

+E ↓ (A.13)

Inserting equations A.5 and A.6, one finds:

F ↑= 2A

cosF

F ↓ (A.14)

Assuming a nearly isotropic distribution for the downward radiance L ↓ yields

F ↓

E ↓

= 2 (A.15)

Measurement of Photolysis Frequencies in the Atmosphere 485

and

F ↑= A

2cosF

+F ↓ (A.16)

When the parameter 

is defined as the direct fraction of the total downward actinic

flux



+F ↓

(A.17)

and is inserted into equation A.16, the following relation between the upward and

downward actinic flux is found:



F ↑

+F ↓

= A





2cos −1 +1



(A.18)

Here, the ratio 

, applicable to actinic fluxes, is the formal counterpart to the albedo

which is defined for irradiances and can be used to estimate F ↑ from the total

downwelling actinic flux. Besides an albedo A

the estimate requires values for the solar

zenith angle  and the fraction 

of the direct actinic flux. A detailed discussion of

reflected actinic fluxes and their relevance above ground and cloud surfaces has been

given by Madronich (1987b).

A.2 Effective transmission of solar radiation into a chemical

actinometer

We consider the effective transmission of ultraviolet and visible solar radiation into the

photolysis cell of a chemical actinometer as discussed in Section 9.3.3. It is assumed

that the photolysis cell is either a spherical bulb or an infinitely long cylindrical tube

with non-absorbing walls made of quartz. The wall thickness is assumed to be small

compared with the radius of the cell. Under the latter assumption, the reactor wall will

behave approximately like a plane-parallel plate at the point at which an incident light

ray penetrates into the interior of the reactor.

An appreciable amount of incident light is lost by reflections at the two interfaces of the

wall, when the radiation enters the photolysis reactor. The reflectance  at each interface

between air and quartz depends on the refractive indices (n

and n

) of the two media,

the angle of incidence 

 and the polarisation of light. For the polarisation component

parallel  and perpendicular ⊥ to the plane of incidence (the plane containing a ray

incident on the surface and the normal of the surface at the point of incidence),  can

be calculated according to Fresnel’s formulas (McCluney, 1994):







cos 

−n

cos 



and 

⊥



cos 

−n

cos 



(A.19)

Here 

is the angle of the refracted light ray in the quartz and is related to 

by Snell’s

law

sin 

= n

sin 

(A.20)

486 Analytical Techniques for Atmospheric Measurement

The reflectances of both polarisation components are equal in the case of normal

incidence 



. With n

=1 (air) and n

=146 −15 (quartz at  = 650−250 nm),

the reflectance is about 3.8% at each interface. For grazing incidence 

→ 90



, the

reflectances both approach 100%.

The effective reflectance R and transmittance T of a plane-parallel plate of quartz

surrounded by air is the result of multiple reflections at the two interfaces of the plate

(McCluney, 1994). Consider an incident ray of light, with intensity I

denoting the flux

of photons carried by the ray (Figure 9.44). At the first air–quartz interface, part of the

light will be reflected with intensity I

=  ×I

, while the remaining fraction 1 − ×I

will be refracted into the quartz. Part of the refracted light will be transmitted through

the second interface, leaving the quartz as a ray of intensity I

= 1 −

×I

, with a

direction that is collinear to the original light. The radiation that is not transmitted at the

second interface will undergo successive internal reflections in the quartz plate, each with

reflectance . On each reflection, a fraction 1 − of light will escape from the plate into

the air. As a result, an infinite number of rays with intensities I

I

I

i →

will be transmitted through the quartz plate, adding up to the total intensity I

. The

intensities can be written as

= I

1 −

= I

1 −



etc

= I

1 −



(A.21)





i=0

= I

1 −





i=0



(A.22)

For x ≤1, the following relation is valid





i=0

1 −x

(A.23)

so that

T =

1 −

1 −

1 +

(A.24)

In a similar way, a succession of rays with the intensities I

I

I

i → will be

reflected with total intensity I

. For reasons of energy conservation, I

must be equal to

−I

. It follows that

R =

2

1 +

(A.25)

Equations A.24 and A.25 are applicable to light with parallel or vertical polarisation by

setting  =



or  =

⊥

, respectively. Unpolarised light can be treated as the superimpo-

sition of two components with parallel and perpendicular polarisation. The transmitted

Measurement of Photolysis Frequencies in the Atmosphere 487

Figure 9.44 Illustration of multiple reﬂections at the two interfaces of a plane-parallel plate. n

denotes

the refractive index of the transparent plate material (e.g. quartz) and n

= 1 is the refractive index of

the surrounding air. I

is the intensity of the incident light ray, I

and I

are the total intensities of the

reﬂected and transmitted rays, respectively. 

is the angle of incidence and 

the angle of refraction

of the original ray.

(reflected) intensities are determined separately for each of the two polarisation compo-

nents, which finally add up to the total transmitted (reflected) intensity. Note that

Equations A.24 and A.25 are valid only for non-coherent light, like solar radiation. The

treatment of coherent, monochromatic radiation must take into account interference

effects which play no role for broadband radiation.

Next we consider the sun light that has penetrated into the interior of the photolysis

reactor. The transmitted ray with intensity I

=1−R×I

is collinear to the original ray

, which has an angle of incidence 

(Figure 9.9). The ray I

strikes the interior wall

surface at an angle which is essentially equal to 

, if the wall thickness is small, compared

with the radius of the reactor. The interior reflection of I

produces a succession of an

infinite number of reflected rays with intensities I

I

, etc. Each of these rays has the

same angle of incidence at the interior wall and the same path length like ray I

. The

collective contribution of all interior rays to the photolysis in the actinometer is therefore

given by

internal





i=1

(A.26)

with

= I

1 −R

= I

1 −RR

etc

= I

1 −RR

(A.27)