Professional Documents
Culture Documents
Proceedings of Spie
Proceedings of Spie
SPIEDigitalLibrary.org/conference-proceedings-of-spie
ABSTRACT
Multi AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) instruments, as solar straylight satellites,
require an accurate characterization and elimination of Fraunhofer lines from solar straylight spectra to measure the
atmospheric column abundance of reactive gases that destroy toxic and heat trapping ozone and form climate cooling
aerosols, like glyoxal (CHOCHO), iodine oxide (IO), or bromine oxide (BrO). The currently achievable noise levels
with state-of-the-art DOAS instruments are limited to δ’DL ≈ 10-4 (noise equivalent differential optical density, δ’);
further noise reductions are typically not straightforward, and the reason for this barrier is not well understood. Here we
demonstrate that the nonlinearity of state-of-the-art CCD detectors poses a limitation to accurately characterize
Fraunhofer lines; the incomplete elimination of Fraunhofer lines is found to cause residual structures of δ’ ≈ 10-4, and
only partially accounted by fitting of an “offset” spectrum. We have developed a novel software tool, the CU Data
Acquisition Code that overcomes this barrier by actively controlling the CCD saturation level, and demonstrates that
δ’DL on the order of 10-5 are possible without apparent limitations from the presence of Fraunhofer lines. The software
also implements active control of the elevation angle (angle with respect to the horizon) by means of a Motion
Compensation System for use with mobile MAX-DOAS deployments from ships and aircraft. Finally, a novel approach
to convert slant column densities into line-of-sight averaged concentrations is discussed.
Keywords: DAC software, CCD, nonlinearity, solar straylight, radiative transfer, O4, IO, glyoxal, MAX-DOAS, motion
compensation, mobile platforms
1. INTRODUCTION
Fraunhofer lines are caused by light absorption of chemical elements (primarily H, Fe, Ni, Ca, Mg, Na) present in the
upper layers of the solar atmosphere. There are thousands of Fraunhofer lines tabulated in the ultraviolet, visible and near
infrared spectral ranges; these lines are superposed as “dark” areas to the thermal back body radiation of the sun. The
elimination of Fraunhofer lines poses experimental challenges if solar photons are used to measure Earth atmospheric
composition of trace gases.1-2 First, Raman scattering leads to the partial “filling-in” of these lines (the “Ring effect”).3
Second, the spectral features of an atmospheric absorber gas measured using “structured” sunlight is different from that
of the same gas measured in the laboratory using a broadband light source (“solar I0“ effect).2,4 Third, for measurements
at moderately low resolution (0.5nm FWHM) the apparent optical depth of Fraunhofer lines is 2-3 orders of magnitude
larger than the optical density of reactive atmospheric trace gases that are of interest to atmospheric chemistry research.
The accurate characterization of Fraunhofer lines is a prerequisite to measure atmospheric trace gases by means of
Differential Optical Absorption Spectroscopy (DOAS).5 DOAS measures atmospheric trace gases by use of their specific
narrow band (<5 nm) ultraviolet–visible light absorption structure in the open atmosphere, separating trace gas
absorption from broadband molecule and aerosol extinction. Nowadays, DOAS is being used from ground, mobile
laboratories, ships, aircraft, and satellite platforms, and can be considered as a reference technique for measurements of
trace gases like bromine oxide (BrO), iodine oxide (IO), chlorine dioxide (OClO), iodine dioxide (OIO), nitrogen dioxide
(NO2), formaldehyde (HCHO), and glyoxal (CHOCHO).1,6-9 Multi AXis DOAS (MAX-DOAS) observes scattered
sunlight under different viewing directions, enabling relatively portable instruments for the spectroscopic remote sensing
of atmospheric trace-gases even if trace layers travel decoupled from ground.10 Furthermore, parameters like the aerosol
extinction coefficient can be derived by (1) measuring trace gases with a known spatial distribution, like the oxygen
dimer O4,11-13 (2) quantitative retrievals of the Ring effect,14 and (3) by intensity ratios at different wavelengths.11
Ultraviolet and Visible Ground- and Space-based Measurements, Trace Gases, Aerosols and Effects VI,
edited by Jay R. Herman, Wei Gao, Proc. of SPIE Vol. 7462, 746203 · © 2009 SPIE
CCC code: 0277-786X/09/$18 · doi: 10.1117/12.826792
2. CU SHIP MA
AX-DOAS INSTRUMENT (CU SMAX-DOAS)
The components of CU’s Ship MAX-DO OAS are shown in Figure 2. It consists of a rotating prism m telescope (f/4), which
is mounted on the outside railing (Fig.2, bottom right), and coupled with a clinometer (bottom leeft) mounted inside the
ship, which provides pitch and roll anglees of the ship to the CompactRIO processor/computer. Th he two clinometers are
actively temperature controlled at 23 ± 0.25° C, resulting in negligible contributions of zero and sppan drifts to the overall
accuracy (zero drift < 0.002°; scale factorr drift <0.03%). The absolute sensor accuracy is 0.1° C.
Fig. 2. Spectral ranges covered by 3 sppectrometers (upper panel), and the inside and outside componen
nts of the CU Ship
MAX-DOAS (lower panel).
The functionality of the motion compensation system was tested by means of a separate (third) clinometer that was
placed on top of the motion controlled telescope. The tests were performed at an elevation angle of zero degrees (line of
sight to the horizon) under conditions when waves acting on the ship caused the ship roll angle to vary by 6° (peak to
peak), and the pitch angle varied by 1.3° (peak to peak). The telescope was pointing near perpendicular to the ship
motion, and the roll angle variations very closely resemble the angle compensated by the motion compensation system.
The variability of the elevation angle could be determined with 0.1° precision, and was found to be 0.56° (1-sigma).
4. RES
SULTS: SIGNAL TO NOISE TESTS
Solar straylight spectra from field and laaboratory experiments were analyzed using the WinDOA AS software.19 The
software performs a non-linear least squares fit by simultaneously adjusting the relevant atmospheric absorbers in
the respective wavelength range to the m measured spectra. To account for broad band effects (in paarticular caused by
Rayleigh- and Mie-scattering) a polynom mial of degree 3 was included. The fitting procedure is performed with the
logarithm of the spectra (i.e. in optical ddensity space). In some cases a pre-logarithmic linear inntensity offset was
included (see introduction). The fitted crooss sections include: two O3 cross section references (T = 223K and 243K),
two NO2 cross section references (T = 2220K and 294K), a water vapor and a glyoxal cross section, as well as a Ring
reference spectrum;20 a linear shift on thee measured spectrum was allowed during the fitting procedure. The 1-sigma
RMS noise of the residuals from these fitts is shown in Figure 4, left panel. The theoretical RMS noise
n is also shown
as a solid line. Field data were collectedd at a high mountain observatory in Colorado, and were evaluated using a
standard MAX-DOAS evaluation proceddure. The “laboratory data” represent solar straylight speectra recorded at a
constant elevation angle, and using diffferent saturation levels (constant over any single acqu uisition). Figure 4
demonstrates that for spectra collected at a constant elevation angle the RMS decreases as expected based on photon
counting statistics. At RMS values on thhe order of 1x10-4 near 350nm (5-8x10-5 near 450nm) th he RMS no longer
decreases as expected with increasing nnumber of photons, but instead it is found that it depend ds (approximately
linearly) on the difference in saturatioon levels between the two spectra. The saturation lev vel represents the
percentage number of the dynamic rangee of the 16-bit ADC at which the maximum of the spectrum was digitized
(in principle the wavelength range where the maximum is determined can be chosen in our Labview w code).
Fig. 4. Dependence of the RMS residuall noise of DOAS retrievals on the number of solar straylight ph hotons (left), and on
the difference (sample – reference) oof CCD saturation levels (see text) used to record the spectra (right).
The factor f can be rationalized as a correction factor to derive accurate concentrations as ccorr = cretrieved / f; it is a
convoluted quantity that accounts for differences in the vertical profile of O4 and the trace gas among other factors.
Notably, at 360nm f360nm = 0.57 is suited to derive the trace gas concentration within 25%; at 477nm the dependence on
boundary layer height is even smaller, and f477nm = 0.64 gives results within better 20% over the range of boundary layer
heights studied. We conclude that a value of f = 0.6 is representative for overhead sun conditions, and gives results
within 30% that are independent of wavelength and boundary layer height. Sensitivity tests that varied SZA showed that
f converges between different elevation angles also at higher SZA, with values of f approaching unity at higher aerosol
optical depth.
6. CONCLUSIONS
Two approaches are investigated to maximize the signal-to-noise of using MAX-DOAS to measure trace gas layers of
weak absorbers: (1) Airborne MAX-DOAS allows to maximize the spatial overlap between the photon paths through
absorber layers. (2) A novel software tool was developed to actively lower the attainable noise level from DOAS
retrievals using solar straylight photons; the software overcomes existing limitations towards lowering the noise level by
actively controlling the saturation level of measured spectra within narrow bounds. It is demonstrated that this is needed
to characterize Fraunhofer lines in a consistent way, and RMS values as low as 2x10-5 are demonstrated for
measurements at constant elevation angle. Active controls of the saturation level are a prerequisite to use the information
contained in the Ring effect to infer aerosol optical properties. The University of Colorado Ship MAX-DOAS and
Airborne MAX-DOAS instruments provide two innovative, and flexible mobile MAX-DOAS instruments that are well
suited to measure the column abundance of reactive gases like CHOCHO, IO, BrO and other gases that are of relevance
to atmospheric composition in urban, remote and marine air, as well as their vertical distribution throughout the free
troposphere.