Monitoring contaminants in the atmosphere by means of a Fourier spectroradiometer
S. K. Dvoruk,V. N. Kornienko, I. V. Kochikov, M. V. Lel’kov, A. N. Morozov,
S. I. Svetlichny , and S. E. Tabalin
Physics Department and Center for Applied Physics, N. E
´
. Bauman Moscow State Technical University,
Moscow
Submitted October 14, 2003
Opticheski Zhurnal 71, 7–13 May 2004
An algorithm based on a model of constancy of temperatures on the path and in a cloud of
contaminants is proposed for solving the problem of identifying and determining contaminant
concentrations by means of a Fourier spectroradiometer. The main problems of the
procedure for processing the initial interferograms and the experimental spectra are considered,
including those with small radiance contrasts of the paths. The experimental results of
field tests of a spectroradiometer based on the proposed algorithm are presented. © 2004 Optical
Society of America
INTRODUCTION
The problem of identifying and determining the con-
taminant concentration has been thoroughly and methodi-
cally treated for laboratory Fourier spectrometers. Software
exists for solving such problems, including a set of math-
ematical procedures that make it possible to pick out the
desired spectral components from an experimental spectrum
and to determine the concentrations. Various spectral data-
bases are available that make it easy as a rule to solve the
problem of identifying a substance. Of course, such proce-
dures require a fairly highly qualified investigator and take
quite a bit of time. This task is much harder for mobile
Fourier spectroradiometers FSRs intended to operate under
field conditions and in real time. This is caused by the very
essence of high-speed analysis and by the rigorous time con-
straints for processing and outputting the final results, as well
as by the nonconstancy of the target conditions, such as tem-
perature variations of the underlying surface, variation of the
concentrations of contaminant clouds in time and space, etc.
The goal of this article is to demonstrate one possible
algorithm for operating an FSR in real time, based on accu-
mulated experience in operating a mobile FSR with spectral
resolution of about 4 cm
-1
, a working range of
700– 1300 cm
-1
, a scan time less than 0.4 sec, and an in-
stantaneous field of view of 1.5° 1.5°. The photodetector is
a Cd–Hg–Te photoresistor cooled to 80 K. Both a self-
contained version of the photodetector and one with a split-
Sterling microrefrigerator were used. The detectivity of the
photoreceptor was about (1–2) 10
10
cm Hz
1/2
/W. The
FSR was equipped with an electromechanical scanning sys-
tem and a television viewer
1,2
and was controlled by means
of a notebook-type PC.
FUNDAMENTAL CONSIDERATIONS
General radiation-transfer equations can be used to pro-
cess the self-radiation spectra of objects in the atmosphere.
However, to solve them even approximately, one must have
significant computing resources, must know the local
weather conditions, and must use spectral databases concern-
ing the radiative properties of the underlying surfaces and the
paths. This problem can be simplified by assuming that the
temperatures of the contaminant cloud and the path are
equal, and this is justified for near-earth paths, where the
temperature variations along the path do not exceed a few
degrees. With these assumptions, the observed spectrum at a
given wave number is
B = P -
t
s
B
0
* - P , 1
where B ( ) is the experimentally determined spectrum,
P ( , T ) is Planck’s function, B
0
* ( ) is the spectrum of the
underlying surface behind the contaminant cloud,
t
( )
and
s
( ) are the spectral transmittances of the path and the
contaminant cloud, and T is the temperature of the path.
Further processing can be carried out directly from Eq. 1
by emulating the experimental data with the corresponding
spectra of the underlying surface and the contaminants. This
is not the most rational processing method for near-ground
observation paths, since it requires an extensive database
concerning the spectral properties of various underlying sur-
faces and observation paths. However, this is the optimum
method for a number of specific problems and sometimes is
the only one possible.
It is possible to simplify Eq. 1, if the spectrum of the
path ( B
0
) is first recorded in the absence of a contaminant
cloud (
s
=1)
B
0
= P -
t
B
0
* - P . 2
Combining Eqs. 1 and 2, we finally obtain
B = P +
s
B
0
- P or
s
=
B - P
B
0
- P
. 3
It is possible to obtain a relationship analogous to Eq. 3
but for conditions in which the temperature T
1
of the con-
taminant cloud and the temperature T
2
averaged over the
path do not coincide:
s
=
B - P
2
+
t
P
2
- P
1
B
0
- P
2
+
t
P
2
- P
1
. 4
271 271 J. Opt. Technol. 71 (5), May 2004 1070-9762/2004/050271-06$15.00 © 2004 The Optical Society of America