Abstract

The United Kingdom Meteorological Office (UKMO) has developed an airborne interferometer to act as a simulator for future satellite-based infrared meteorological sounders. The Airborne Research Interferometer Evaluation System (ARIES) consists of a modified commercial interferometer mounted on the UKMO C-130 aircraft. The instrument is sensitive to the wavelength range 3.3–18 μm and has a maximum optical path difference of ±1.037 cm. This paper describes the design and performance of ARIES, discusses instrument calibration, and presents some preliminary results. An important problem associated with the use of the new generation of high-spectral resolution infrared meteorological sounders is that improvements need to be made to knowledge of atmospheric spectroscopy and radiative transfer. These improvements are necessary to extract the promised vertical and absolute resolution in temperature and humidity retrievals from these new high-spectral resolution sounders. By virtue of the extensive instrumentation that is available on the C-130 aircraft for observing and measuring the basic meteorological and atmospheric parameters (e.g., in situ temperature, humidity, and ozone), it is hoped that ARIES will be an important tool for use in studying this issue.

1. Introduction

The Remote Sensing branch of the United Kingdom Meteorological Office (UKMO) has developed a number of airborne radiometer simulators for forthcoming meteorological satellite instruments. These satellite instruments will provide improved temperature and humidity profiles for operational meteorology, ultimately leading to improvements in forecasting (Eyre 1990). Airborne simulators are used to test and to improve knowledge of the relevant atmospheric spectroscopy, surface characteristics, and radiative transfer models. This testing is a necessary precursor to the routine extraction of the required atmospheric data from the satellite radiance measurements.

Until now, much of the work at UKMO has focused on the development of simulators for the Advanced Microwave Sounding Unit (AMSU). This work has led to a pair of microwave radiometers (English 1995; Hewison and English 1999) mounted on the UKMO C-130 aircraft that are capable of making measurements at a number of AMSU-B frequencies (23.8, 50.1, 89, and 157 GHz). Recently, however, an infrared interferometer called the Airborne Research Interferometer Evaluation System (ARIES) has been built as a simulator for a number of new infrared sounding satellite instruments, in particular the Infrared Atmospheric Sounding Interferometer (IASI; Amato 1995; Diebel et al. 1996). IASI is being developed by the Centre National d’Etudes Spatiales, Toulouse, France, for the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT). It will fly on the EUMETSAT Polar Systems polar-orbiting series, the first of which is due to be launched in about 2003. These new sounders, which also include the Atmospheric Infrared Sounder (AIRS;Aumann and Pagano 1994) under development by the Jet Propulsion Laboratory, and projected to fly on the National Aeronautics and Space Administration (NASA) Earth Observing System EOS-PM1 satellite, offer considerably higher spectral resolution and spectral coverage than do current meteorological sounders, for example, the High-Resolution Infrared Sounder (HIRS; Smith et al. 1979). Increased spectral resolution will provide data that are representative of shallower sections down through the atmosphere. It also will allow measurement between individual gas emission lines (Fig. 1). The coverage gives the opportunity to vary the spectral position and extent of the data being processed. The final goal is to yield temperature and humidity retrievals with higher vertical resolution and accuracy. A much vaunted target (Neuendorffer 1989; Diebel et al. 1996; Prunet et al. 1998) is to retrieve temperature and humidity with a typical vertical resolution of 1 km in the troposphere and mean errors of 1 K (temperature) and 10% (humidity). The feasibility of retrieving the improved data from such higher-resolution sounders has been demonstrated experimentally, using the High-Resolution Interferometer Sounder instrument mounted on the high-flying NASA ER-2 aircraft (Huang et al. 1992).

Fig. 1.

A comparison of the spectral resolution of typical current and forthcoming IR satellite sounders. The spectrum (solid line) is a simulation calculated using the GENLN2 line-by-line radiative transfer code based on a midlatitude summer atmosphere and an instrument with the spectral resolution of IASI (2-cm maximum optical path difference). The shading represents the half-power bandwidth of HIRS/2 channels. Note that HIRS channels 2–7 overlap, the half-power bandwidth being typically 16 cm−1.

Fig. 1.

A comparison of the spectral resolution of typical current and forthcoming IR satellite sounders. The spectrum (solid line) is a simulation calculated using the GENLN2 line-by-line radiative transfer code based on a midlatitude summer atmosphere and an instrument with the spectral resolution of IASI (2-cm maximum optical path difference). The shading represents the half-power bandwidth of HIRS/2 channels. Note that HIRS channels 2–7 overlap, the half-power bandwidth being typically 16 cm−1.

As with the other UKMO simulators, ARIES is mounted on the C-130 aircraft, which is an extremely flexible and well-instrumented platform. The aircraft has a range of 5000 km, can fly between 16-m and 10-km altitude, and offers a wide range of instrumentation to measure meteorological, radiative, cloud microphysical, and related chemical parameters. Some of these measurements, particularly temperature, humidity, and ozone concentration, provide collocated “truth” for comparison with the radiometer simulators. A summary of these basic measurements is presented in Tables 1 and 2.

Table 1.

Specifications of aircraft in situ temperature and humidity measurements.

Specifications of aircraft in situ temperature and humidity measurements.
Specifications of aircraft in situ temperature and humidity measurements.
Table 2.

Specificiations of aircraft dropsonde system (Vaisala RD82-41G GPS dropsonde).

Specificiations of aircraft dropsonde system (Vaisala RD82-41G GPS dropsonde).
Specificiations of aircraft dropsonde system (Vaisala RD82-41G GPS dropsonde).

ARIES is mounted in a converted fuel tank on the starboard wing. The instrument consists of a rugged commercial infrared interferometer, modified by UKMO for aircraft usage. The instrument is sensitive to infrared radiation in the wavelength range 3.3–18 μm (550–3000 cm−1) and is capable of viewing both the upwelling and downwelling radiation (as well as various angular downward views). Fuller details of the instrument design are provided in the remainder of this paper.

The primary aim for ARIES data will be to provide experimental airborne infrared spectra for comparison with simulations from line-by-line radiative transfer models, for example, GENLN2 (Edwards 1992). The intention in this exercise is to identify and to study areas of uncertainty in the modeling and in the relevant atmospheric spectroscopy, prior to the launch of IASI. This work is expected to include infrared spectral features associated with carbon dioxide, water, and other minor constituents, for example, ozone, nitrous oxide, methane, and chlorofluorocarbons. ARIES also will be used for studies on cloud detection and clearing in infrared satellite views (Smith and Eyre 1996), cloud microphysics and aerosol retrieval (Smith et al. 1995; Moncet and Clough 1997), and sea and land infrared emissivity measurements (Fiedler and Bakan 1997).

2. Design

Interferometers have been exploited successfully to record infrared emission spectra in a number of air and ship (Smith et al. 1996; Korb et al. 1996), satellite (Hanel et al. 1971, 1972; Ogawa et al. 1993), and spaceborne (Gunson et al. 1996; Christensen and Pearl 1997) missions. Such instruments have fundamental advantages when compared to more traditional filter or grating radiometers—they are multiplex (Felgett advantage), sampling all frequencies simultaneously, and they can accept light over a greater solid angle, giving greater throughput (Jaquinot advantage). These advantages lead to interferometers that have a higher signal-to-noise ratio and greater spectral resolution (Griffiths and de Haseth 1986). These considerations are important in the drive to improve vertical resolution and absolute accuracy of temperature and humidity retrievals for assimilation into numerical weather prediction (NWP) models.

At its heart, ARIES has a rugged commercial interferometer (MR 200), developed by BOMEM, Inc., of Canada. The interferometer views the external scene using a pointing mirror assembly designed by UKMO. The pointing mirror also can be directed to view a pair of onboard calibration targets. ARIES data are recorded on a personal computer (PC) located in the main body of the aircraft and connected to the instrument by a fiber-optic cable, which runs down the starboard wing. Figure 2 shows a schematic diagram of the package located in the aircraft wing pod. A summary of the ARIES package is presented in Table 3. The design of the interferometer, scan mirror assembly, and acquisition system is now considered.

Fig. 2.

A schematic diagram of the ARIES fit within the C-130 wing pod.

Fig. 2.

A schematic diagram of the ARIES fit within the C-130 wing pod.

Table 3.

ARIES specifications.

ARIES specifications.
ARIES specifications.

a. ARIES interferometer

As with other “Michelson”-type interferometers, radiation incident on the instrument is partially reflected and partially transmitted by a beam splitter. A variable optical path difference is introduced between the reflected and transmitted paths, prior to recombination of the beams. The resulting interference pattern is monitored as a function of the path difference; this so-called interferogram contains all the spectral information about the source, which can be recovered using the technique of Fourier transform (Griffiths and de Haseth 1986).

Figure 3 shows a schematic of the ARIES interferometer optics; the interferometer itself is based on a double-pendulum arrangement. In brief, light enters the instrument through a zinc selenide (ZnSe) window and is partially reflected and partially transmitted by a coated potassium bromide (KBr) beam splitter. Note that virgin KBr is insufficiently reflective (∼1%) in the infrared to act as an efficient beam splitter. The KBr therefore is coated to increase the reflectivity; the coating is applied to selective regions of the beam splitter to minimize unwanted reflections from stray beams. The beam splitter directs the resulting beams to a pair of corner-cube reflectors, which are mounted on a pivoting arm. The motion of this arm rotates the reflectors in opposite directions with respect to the beam splitter, providing a mechanism for the introduction of a variable path difference. The arm moves through a maximum of ∼0.259 cm from its central position. Taking into account the path to and away from the reflector, this movement therefore introduces a maximum path variation of ∼0.518 cm and a maximum optical path difference between the transmitted and reflected paths of ∼1.03 cm. At the corner cubes, the beams are shifted to a higher vertical plane. Following reflection at the corner cubes, the beams interfere on return to the beam splitter and the interference pattern is monitored by a detection system. ARIES has two infrared detectors arranged in a sandwich. A mercury–cadmium telluride (HgCdTe; MCT) detector is sensitive to the wavelength range 5.5–18 μm (550–1800 cm−1) and an indium antimonide (InSb) detector covers the range 3.3–5.9 μm (1700–3000 cm−1). The detectors are cooled to approximately 80 K by a Stirling microcooler (EG&G Judson).

Fig. 3.

A schematic diagram of the ARIES interferometer optics.

Fig. 3.

A schematic diagram of the ARIES interferometer optics.

ARIES is a dual-port instrument, meaning that the signal at the detector is proportional to the difference between the radiance entering the instrument through the ZnSe window and that entering the optical system from a second port located inside the instrument. An internal blackbody is mounted in the second port of the instrument to allow easy characterization of the radiance from this port. With reference to Fig. 3, the blackbody is mounted below mirror A at the level of the input radiation, and its temperature is monitored using a platinum resistance thermometer (PRT) and a thermistor (for redundancy), so that its emission into the interferometer can be calculated [see Eq. (5)]. Referring to Fig. 4, it is clear that the beam splitter–reflected path from the second port involves one internal and one external reflection at the beam splitter (Fig. 4d), whereas the beam splitter–transmitted path involves no beam splitter reflections (Fig. 4c). Since an internal reflection introduces a phase reversal compared with an external refection, a 180° phase shift is introduced between the beams from the second port. Thus for the radiance from the second port, there is cancellation when the optical path difference is zero. By contrast, for the main signal, both beams have the same number of reflections and hence there is constructive interference at zero optical path difference (Figs. 4a,b). Clearly, the strength of the resultant (main and second port) modulation increases as the difference between the external scene temperature and second port temperature increases. By a more detailed consideration of the reflections at the beam splitter, it is possible to show that the resultant modulation is zero when the external scene and second port temperatures are the same. The second port blackbody also acts as a dump for any reflected light from the detector system that happens to enter this part of the interferometer.

Fig. 4.

Reflection and transmission of radiance from the main and second ports. The beams from the main port both undergo (a) one internal and (b) one external reflection. Whereas (c) one beam from the second port undergoes no internal reflections, (d) the other undergoes one internal and one external reflection.

Fig. 4.

Reflection and transmission of radiance from the main and second ports. The beams from the main port both undergo (a) one internal and (b) one external reflection. Whereas (c) one beam from the second port undergoes no internal reflections, (d) the other undergoes one internal and one external reflection.

The optical path difference sampling interval is set by the fringe spacing in an interference pattern, generated by a helium–neon (He–Ne) laser. The maximum optical path difference (OPDmax) defines the instrument resolution. However, definitions of resolution are complicated by any modulation of the interferogram amplitude caused by the instrument’s optical design (self-apodization) or applied by the user to suppress sidelobes (apodization). To avoid the description of a particular apodization function, the resolution can be described in terms of the spectral sampling interval δυ̃ [=1/(2OPDmax)]. This is the spectral-domain “Nyquist” sampling interval, in that the continuous spectrum limited by the maximum OPD of the interferometer can be reconstructed completely from any set of discrete samples with regular spacing δυ̃ (Griffiths and de Haseth 1986). For ARIES, OPDmax = 1.037 cm and therefore the spectral sampling interval is 0.482 cm−1. Typically, at this “resolution,” two full scans per second in each direction are obtained, but this frequency can be increased by decreasing the resolution, that is, decreasing the OPDmax (see Table 3). The associated software provides options for a number of such degraded settings. The instrument line shape of the interferometer is related to the gain dependence of the instrument as a function of OPD variation, which has been measured and simulated; these data have been reported elsewhere (Atkinson and Wilson 1997).

The interferometer and detector system (see Fig. 2,“optical head”) is mounted in a machined aluminum box, which is pressurized with dry nitrogen (∼3 × 104 Pa overpressure). A series of thermostatically controlled heaters (Minco “foil”-type) maintain the unit within its operational temperature range, typically 20°–40°C. The entire unit is mounted to the pod via antivibration mounts designed to dampen the severe aircraft vibrations.

b. Scan mirror and calibration targets

A pointing mirror assembly is coupled to the interferometer input window to allow ARIES to view the external scene. The mirror is gold coated with a silicon oxide protective layer. It can be directed to view upward through an aperture in the pod (3° off zenith) and downward through a slot, with views at nadir and at a series of angles up to 55° toward the main fuselage (across the aircraft track). The ARIES field of view is 44 mrad. The mirror also can be directed to view either of two blackbody heated calibration targets, which employ foil-type heaters, under proportional temperature control. The calibration targets have an aluminum substrate and are coated with an infrared black paint (Nextel Velvet Coating 811–21). The surface of the calibration targets comprises an array of pyramids (square base side is 7.2 mm, height is 12.5 mm) machined onto a circular base with a diameter of 67.5 mm. Each target has a black-alochromed cylindrical shroud with a height of 66 mm to improve the emissivity of the overall target system. The temperature of these calibration targets is monitored via a series of PRTs (Minco S201PD) and can be set from software (up to 80°C). Each target has 5 PRTs mounted in the base and an additional PRT in the shroud. Typically, the temperatures are set to straddle those of the scene to facilitate calibration (see section 3).

c. Software and data acquisition

ARIES interferogram data are sent to a PC located within the main fuselage of the aircraft via two fiber-optic cables located in the wing. The data are handled by a dedicated digital signal processing card. Other data, including the PRT temperatures and communication to the scan mirror mechanism, are exchanged via RS-422 and RS-485 twisted-pair connections. Using software developed by BOMEM, the operator can specify the acquisition timing, angular view, and blackbody temperatures, and can observe temperatures from various other PRTs located around the instrument. In addition to recording the interferometer data, the software produces a “housekeeping file,” which records all of these parameters for later use.

3. Calibration

a. Basic calibration

As described above, ARIES can view two temperature-stabilized blackbody calibration targets, which are mounted on the pointing mirror assembly. Viewing these targets provides the information necessary to calibrate the raw spectral information. The calibration procedure is similar to that described in Revercomb et al. (1988). The Fourier transform convention used in the analysis of ARIES data is described by

 
formula

where j and k are integers, Vj is the real interferogram at point j, Sk is the complex raw spectrum at point k, and i = √−1⁠, and N is the total number of points. In brief, following Fourier transformation of the observed interferogram (Fig. 5), the relationship between the measured complex raw spectrum S (Fig. 6) and the true scene radiance spectrum R is given by

 
S = G(RRport2C),
(2)

where G is a complex gain and C is a radiance offset, both dependent on frequency. Note that G and C refer to the entire system, rather than purely the instrument detectors. Here Rport2 is the radiance from the internal blackbody, which can be calculated from its measured temperature [see Eq. (5)].

Fig. 5.

A typical ARIES interferogram (MCT detector), expanded in the amplitude axis to demonstrate the information content across the entire range of the OPD variation. The inset shows the full amplitude of the center burst.

Fig. 5.

A typical ARIES interferogram (MCT detector), expanded in the amplitude axis to demonstrate the information content across the entire range of the OPD variation. The inset shows the full amplitude of the center burst.

Fig. 6.

Real (solid line—bottom) and imaginary (dashed line—top) components of a raw ARIES spectrum, following Fourier transformation of the interferogram shown in Fig. 5 to the frequency domain.

Fig. 6.

Real (solid line—bottom) and imaginary (dashed line—top) components of a raw ARIES spectrum, following Fourier transformation of the interferogram shown in Fig. 5 to the frequency domain.

Complex gain G takes into account the following:

  1. the detector efficiency as a function of frequency,

  2. amplifier gain,

  3. phase offsets, and

  4. the optical efficiency of the instrument, for example, signal attenuation by the mirror or entrance window.

Note that, given the normalization convention presented in Eq. (1), the magnitude of the gain is independent of instrument resolution.

Radiance offset C is expected to be a slowly varying function of instrument temperature; it takes into account the following:

  1. any errors in the calibration of the internal blackbody PRT,

  2. nonideal behavior of the internal blackbody, and

  3. nonideal transmission of the entrance window and reflectance of the pointing mirror. In these events, a small fraction of the signal will originate from the mirror/window and will depend on the mirror/window temperature.

Making measurements of the two blackbody calibration targets set at different temperatures yields two expressions in the form of Eq. (2):

 
S1 = G(R1Rport2,1C),
(3)

and

 
S2 = G(R2Rport2,2C),
(4)

where S1 and S2 are the measured complex raw spectra and R1 and R2 are the true scene radiance spectra corresponding to the two blackbody views. Spectra R1 and R2 can be calculated using the Planck function:

 
formula

where υ̃ is wavenumber (cm3), c1 = 1.1911 × 10−8 W m−2 sr−1 (cm−1)−4, c2 = 1.439 K (cm−1)−1, and R is in units of W m−2 sr−1 (cm−1)−1. Given that R1, R2, and Rport2 are known and that S1 and S2 are measured, Eqs. (3) and (4) can be solved to yield G and C (Fig. 7):

 
formula
Fig. 7.

(a) ARIES gain G and (b) radiance offset C functions calculated from blackbody calibration views recorded immediately prior to the data shown in Fig. 5. Real (solid line—top) and imaginary (dashed line—bottom) components are shown.

Fig. 7.

(a) ARIES gain G and (b) radiance offset C functions calculated from blackbody calibration views recorded immediately prior to the data shown in Fig. 5. Real (solid line—top) and imaginary (dashed line—bottom) components are shown.

The contamination of the gain and offset functions by spectral features shown in Fig. 7 results from the small atmospheric path between the calibration targets and instrument, and from any residual atmosphere within the instrument itself because of an incomplete purge. Such features may have the effect of increasing the instrument noise because of the lowering of the system gain performance at certain wavelengths. Apart from the increased noise, all other effects will be “calibrated out” except where the signal is so strong that the gain is reduced to zero. In this case, information is lost (i.e., at a few discrete frequencies in the water vapor band: ∼1400–1800 cm−1). Note that in the intermediate case, where the gain is degraded but not to zero, the data are still calibrated but are more noisy than they would be otherwise. The calibration is still valid because the atmosphere between the blackbodies and the ARIES scan mirror is expected to be at the same temperature as the equivalent path between the scan mirror and atmospheric scene. The lack of any significant local heating of the atmosphere by the blackbodies arises because of the good ventilation around the instrument (particularly in flight). In the cases where the system gain is zero such data are of little interest to our group, since at these wavelengths that correspond to very strong transitions the instrument effectively is sounding the “local” rather than the “remote” atmosphere in the lower-tropospheric altitude range of the UKMO aircraft. To some extent, a similar comment can be made about those wavelengths that show any spectral “leakage” into the gain and offset values.

Assuming that the phase of the raw spectrum (at a given frequency) is the same for both blackbody measurements, then S2/S1 will be real and hence C will be real [this fact can be seen by dividing the top and bottom of Eq. (7) by S1]. This case occurs, provided that both calibration sources are uniform spatially and fill the beam. After these calibration constants are calculated, Eq. (2), which corresponds to the atmospheric scene, can be inverted to give the measured radiance (Fig. 8).

Fig. 8.

Real (solid line) and imaginary (dashed line) components of a calibrated ARIES radiance spectrum, following calibration of the raw data shown in Fig. 6, as described in the text. Note that virtually all the spectral information is in the real component, as might be expected.

Fig. 8.

Real (solid line) and imaginary (dashed line) components of a calibrated ARIES radiance spectrum, following calibration of the raw data shown in Fig. 6, as described in the text. Note that virtually all the spectral information is in the real component, as might be expected.

Note that forward and reverse scans of the interferometer (OPD = −OPDmax → +OPDmax and OPD = +OPDmax → −OPDmax) must be treated separately during this calibration, since in general they will have different phase offsets. However, the final radiances may be averaged, giving an improvement in signal-to-noise ratio by a factor of 21/2. Further, although the calculated radiance is a complex array, it is expected that the imaginary component is pure noise, and that brightness temperatures therefore should be calculated from the real component alone.

b. Cold calibration target

To achieve the best calibration, it is important that the target temperatures straddle those of the atmospheric scene. Currently, the targets can be stabilized at user-defined temperatures between ambient and 80°C. It clearly is not possible to straddle the scene temperature when making a zenith view through cloudless skies, since the scene out to space is extremely cold. To alleviate this problem, a ground-use-only cold calibration target has been developed. The target design is similar to the onboard calibration targets (see above), including a grooved target together with a shroud. Both target and shroud are coated internally with infrared black paint and cooled using circulated liquid nitrogen. A thermocouple embedded into the target is used to monitor the target cooling and to indicate when that target temperature has stabilized. The actual temperature of the target is not measured; clearly, the minimum possible temperature using liquid nitrogen as the coolant would be 77 K. It is expected that the target temperature is close enough to 77 K that the difference in radiance terms from that at 77 K is below the level of detection for all but the lowest frequencies of the ARIES spectral range. The target also incorporates a gaseous nitrogen purge, which is used to avoid the formation of “clouds” of condensed water over the target. Thus far, the cold target has been used only in the laboratory, where both ARIES and the target were held in a purged gaseous nitrogen atmosphere. Figure 9 shows a spectrum of the cold target, calibrated using the two onboard calibration targets (set to 333 and 300 K, respectively). Generally the measured radiance values give confidence to the use of the onboard targets to calibrate “cold” atmospheric scenes. Further refinement and characterization of the targets and the technique, in particular the removal of the slight negative radiance offset apparent in Fig. 9, may lead to the quantitative use of such cold target views.

Fig. 9.

ARIES observations of a liquid nitrogen-cooled blackbody target. The upper spectrum corresponds to the data from the MCT detector; the lower spectrum is from the InSb detector. Superimposed on the spectra are representative blackbody temperature profiles (dashed lines).

Fig. 9.

ARIES observations of a liquid nitrogen-cooled blackbody target. The upper spectrum corresponds to the data from the MCT detector; the lower spectrum is from the InSb detector. Superimposed on the spectra are representative blackbody temperature profiles (dashed lines).

c. MCT detector nonlinearity

It was found that the MCT detector has a slightly nonlinear output as a function of radiance. Specifically, the MCT detector sensitivity decreases with increasing incident photon flux. Similar behavior with MCT and other photoconductive detectors has been observed elsewhere (Abrams et al. 1994; Rahmelow 1997). No such behavior is exhibited by the photovoltaic InSb detector.

Characterization of this behavior as a function of photon flux is required to correct the measured interferograms. For an ideal (linear) photoconductive detector operated with constant bias voltage, the detector current Idet is given by

 
Idet = + I0,
(8)

where ϕ is the photon flux and A and I0 are constants. The detector current is converted to a voltage in the preamplifier. However, because capacitive coupling is used to reduce dynamic range and unwanted low-frequency noise, only the alternating current component of the signal is available at the preamplifier output. Therefore an indirect method is required to estimate the total photon flux within the detector pass band, which depends on the intensity of radiation from both the main instrument port (scene) and the second input port (viewing a blackbody at ambient temperature).

Figure 10 shows schematically the contributions to the ARIES signal at the detector, assuming an ideal beam splitter with a reflectivity of 50%. The intensity at maximum OPD is equal to the mean of the intensities from the two input ports (ϕ1, ϕ2), whereas at zero OPD the intensity is equal to that from the main port alone (ϕ1), since the transmitted and reflected paths waves from the second port interfere destructively and exactly cancel (see section 2a).

Fig. 10.

Contributions to ARIES signal at the detector; see text for details.

Fig. 10.

Contributions to ARIES signal at the detector; see text for details.

This finding suggests the following route to correcting for nonlinearity.

  1. Take two calibration measurements and calculate the gain, assuming a linear detector.

  2. For each calibration measurement, calculate the second port radiance, given by the Planck function at the measured port temperature, R(υ̃). Using the gain from step 1, calculate a weighted mean of the second port radiance Rport2: 
    formula
  3. For each calibration measurement, measure the interferogram voltage at zero OPD V(0). This value is by definition equal to the mean of the raw spectrum (the Fourier transform of the interferogram), and the raw spectrum in turn is approximately equal to (RmainRport2)G(υ̃) [see Eq. (3)]. Hence one can calculate the voltage offset, ΔV: 
    formula
  4. For each calibration measurement, add the offset to the measured interferogram voltage (V) to synthesize a voltage signal V′, which is proportional to photon flux. Linearize this voltage assuming a quadratic relationship between V′ and its corrected value Vlin: 
    V′ = ΔV + V,Vlin = V′ + βV2.
    (11)
  5. Recalculate the gain and radiance offset using the newly calculated interferogram voltage Vlin.

  6. Repeat steps 2–4 for any “measurement” (as opposed to calibration) interferogram and use the gain and offset from step 5 to calculate radiance.

This method has introduced a constant β; the question of how to select the value of β remains. One way to fine-tune the nonlinearity factor β is by comparison with the InSb detector in the region of spectral overlap (Fig. 11). In this manner, a β value of ∼0.02 was found to be appropriate. Note that this β value was derived by comparison of a number of different spectra, including the simpler blackbody spectra as well as atmospheric observations over a range of conditions. The value is not “tuned” on an individual spectrum basis. The correction is remarkably stable and effective and, therefore, correction to higher powers is not necessary.

Fig. 11.

This figure shows data from the overlap region between the ARIES MCT and InSb detector regions from a typical airborne observation. Here, the data are used to demonstrate the effect of the MCT detector nonlinearity factor β. (a) MCT detector data (solid line) assuming β = 0.02, together with InSb detector data (shaded line). (b) For comparison, MCT detector data (solid line) assuming β = 0.0 and β = 0.04, together with InSb detector data (shaded line).

Fig. 11.

This figure shows data from the overlap region between the ARIES MCT and InSb detector regions from a typical airborne observation. Here, the data are used to demonstrate the effect of the MCT detector nonlinearity factor β. (a) MCT detector data (solid line) assuming β = 0.02, together with InSb detector data (shaded line). (b) For comparison, MCT detector data (solid line) assuming β = 0.0 and β = 0.04, together with InSb detector data (shaded line).

4. ARIES performance

a. Noise performance

ARIES noise-equivalent temperature difference NEΔT was determined experimentally to check the interferometer manufacturer’s values, quoted at two spot frequencies (Table 3) and to provide details of the variation over the entire frequency band of the detectors. The method chosen to determine NEΔT at any particular temperature relies on the fact that it can be calculated from NEΔR, the standard deviation of the radiance using the Planck function [Eq. (5)] evaluated at that temperature. In this method, multiple interferograms are recorded while observing a target at constant temperature and each single interferogram is stored separately, that is, no coaddition occurs. Each interferogram then is transformed and calibrated as described in section 3a, making use of calibration views of the onboard blackbody targets. Taking the real part of these radiance spectra, which should contain all the spectral information, NEΔR can be calculated as the standard deviation of the radiance measurement. This value can be converted to NEΔT at any desired temperature using the relationship

 
formula

where dR/dT is the differential of the Planck function [Eq. (5)] evaluated at the temperature of interest.

For this exercise, 30 single interferograms were recorded of the constant temperature target, followed by standard observations of the onboard calibration targets. ARIES was operated in its lowest spectral resolution mode, nominally 16 cm−1, which in more accurate terms corresponds to a spectral sampling bin of 7.7 cm−1 (OPDmax = 0.065 cm); see Table 3 for a fuller description of ARIES resolution modes. This mode is appropriate given that the NEΔT variation is expected to show relatively low resolution and that the results easily can be extrapolated to the higher-resolution modes, as described below. The results at 280 K for the MCT and InSb detectors are shown in Figs. 12a and 12b, respectively. The degradation in noise performance in the region of 2350 cm−1 is associated with the lower system gain caused by a strong spectral feature (in this case, part of the 4.3-μm carbon dioxide band), as described in section 3a.

Fig. 12.

Measured single-scan ARIES NEΔT values at 16 cm−1 nominal resolution for the ARIES (a) MCT and (b) InSb detectors. Deduced NEΔT values at 1 cm−1 nominal resolution for (c) MCT (1-min coaddition) and (d) InSb (2-min coaddition) detectors also are shown. See text for details.

Fig. 12.

Measured single-scan ARIES NEΔT values at 16 cm−1 nominal resolution for the ARIES (a) MCT and (b) InSb detectors. Deduced NEΔT values at 1 cm−1 nominal resolution for (c) MCT (1-min coaddition) and (d) InSb (2-min coaddition) detectors also are shown. See text for details.

To calculate the NEΔT profile at other instrument resolution settings, random noise is assumed to dominate other sources of noise, for example, 1/f noise, where f is the frequency; this assumption is reasonable over the period of a scan. Random noise per sample is constant in the interferogram domain, so the total noise power is proportional to the scan period. By reference to Eq. (1), it can be seen that the spectrum at any given frequency is equal to the weighted sum of all the interferogram components, where the weighting is simply a constant-amplitude phase function. Thus the spectral noise power at a given frequency also is proportional to the scan period. It follows that the noise amplitude at any frequency, and hence NEΔT, is proportional to the square root of the scan period. NEΔT for ARIES operating in its highest spectral resolution mode, nominally 1 cm−1, thus can be calculated from the 16 cm−1 data by multiplying by a factor of 161/2 = 4, since the scan period is directly proportional to the nominal resolution setting.

The BOMEM specification is for NEΔT < 0.2 K viewing a 280 K target at 1000 cm−1 for 1 min and at 2700 cm−1 for 2 min at maximum (1 cm−1) resolution. For comparison with the BOMEM data, it clearly is necessary to account for coaddition of interferograms, with 240 scans recorded per min, assuming that both directions of the interferometer sweep are utilized. Co-addition of interferograms has the effect of reducing the random noise by a factor inversely proportional to the square root of the number of scans averaged. Figures 12c and 12d present NEΔT profiles calculated for 1- and 2-min observations for the MCT and InSb detectors, respectively. The data have been scaled to represent operation at maximum (1 cm−1) instrument operation, as described above. Also shown on the plots is a dotted line indicating NEΔT = 0.2 K. Clearly the measurement at 1000 cm−1 of ∼0.02 K (Fig. 12c) easily surpasses the specification by an order of magnitude; in fact, the measured NEΔT over the approximate range 600–1700 cm−1 meets this target. At 2700 cm−1 (Fig. 12d), the measured NEΔT of ∼0.4 K is just outside the specification, though it is within the 0.2 K target at 2550 cm−1.

b. ARIES data

ARIES made its first flights on the UKMO C-130 aircraft during 1996 and since then has participated in a number of campaigns. However, a hardware fault associated with insufficient shielding of the detector preamplifiers had necessitated a complicated, imperfect recalibration procedure to extract scientifically usable information from the data. Once the fault had been identified, a repair was effected by the interferometer manufacturers during the first quarter of 1997. Since then, ARIES has been performing according to expectation.

Figures 13a and 13b show an example of an ARIES nadir brightness temperature spectrum across the spectral range of the MCT and InSb detectors, respectively. The spectrum was recorded at the instrument’s maximum resolution, corresponding to a sampling bin of ∼0.5 cm−1. The data were obtained during sortie A545 of the UKMO C-130 on 10 June 1997. The aircraft was flying over the North Sea (∼54°0′N, 0°30′E), flying in clear air, straight and level at approximately 3.7-km altitude (∼645 mb). The relevant interferogram consisted of coaveraging 100 observations, over a period of approximately 50 s. At a ground speed of ∼100 m s−1, this observation time corresponds to a track of 5 km over the sea. Indicated on the spectra are the molecular species responsible for some of the major spectral features in this range.

Fig. 13.

ARIES brightness temperature spectra (unapodized) plotted across the range of the (a) MCT and (b) InSb detectors, respectively. Indicated on the spectra are the key molecular species responsible for the major emission features observed. See text for details of the flight.

Fig. 13.

ARIES brightness temperature spectra (unapodized) plotted across the range of the (a) MCT and (b) InSb detectors, respectively. Indicated on the spectra are the key molecular species responsible for the major emission features observed. See text for details of the flight.

Of primary interest is the comparison of these and other ARIES data with the output of line-by-line radiative transfer models. Initially, this comparison can be used at a general level to test the performance of the instrument; subsequently such comparisons can be used to identify failings in the models or spectral line databases. During this flight, in situ temperature and humidity measurements were made using the aircraft’s general instrumentation (see Table 1). By flying the aircraft from the minimum safe altitude (16 m) to an altitude above 3.6 km, the entire atmospheric temperature and humidity profile relevant to the ARIES observation at 3.6 km was recorded, to use as an input to line-by-line radiative transfer models for comparison with the ARIES observation. Figure 14 shows the temperature and humidity profiles recorded in this manner. These data have been used as inputs to the GENLN2 (version 3) line-by-line radiative transfer model to generate a simulation of the ARIES observation. At UKMO, this model currently is being used with the HITRAN 92 (high-resolution transmittance) molecular spectral line database (Rothman et al. 1992); however, both model and database are due to be updated in the near future (GENLN2 version 4, HITRAN 96).

Fig. 14.

(a) Atmospheric temperature and (b) humidity profiles recorded during UKMO C-130 flight A545.

Fig. 14.

(a) Atmospheric temperature and (b) humidity profiles recorded during UKMO C-130 flight A545.

A comparison of ARIES data and GENLN2 simulation, together with a difference plot, shows generally good agreement (Fig. 15), with discrepancies typically less than 1 K. Improved agreement already has been observed in the preliminary analysis of data recorded on flights with better characterized atmospheric temperature and humidity profiles; a further improvement also is expected with the updated spectral line database and model described above. A notable exception to the general level of agreement between ARIES data and simulation shown in Fig. 15 is at the center of the major carbon dioxide (CO2) emission feature at ∼ 670 cm−1, where the experimental observation is noticeably warmer than that predicted by the simulation. This wavenumber region corresponds to the center of the molecular transition, where the transition strength is strongest;consequently the CO2 responsible for the observed emission is very close to the instrument. The ambient temperature in the aircraft pod is slightly warmer than the surrounding atmosphere, and this difference gives rise to the observations of slightly warmer CO2 at these wavenumbers. This situation can be modeled by including a very shallow atmospheric layer at the aircraft altitude, which is slightly warmer than the temperature indicated by the measured profile. As a guide, the temperature measurement from a PRT mounted on the ARIES rotating scan drum can be used. An example of the results of such an approach is shown in Fig. 16, where an additional 20-cm atmospheric path, 1 K warmer than the surrounding atmosphere, has been included in the line-by-line simulation. Clearly this inclusion improves the agreement between simulation and experiment in this region. Further such improvements could be made by refining the description of this pod layer; however, such an improvement generally is not of importance, given that UKMO’s interest lies in sounding the remote atmosphere rather than the conditions in the aircraft pod.

Fig. 15.

Comparison of ARIES brightness temperature (BT) data (a) over the wavenumber range 600–800 cm−1 corresponding to part of the spectrum in Fig. 13 with (b) a line-by-line radiative transfer simulation. (c) A difference plot (ARIES − simulation).

Fig. 15.

Comparison of ARIES brightness temperature (BT) data (a) over the wavenumber range 600–800 cm−1 corresponding to part of the spectrum in Fig. 13 with (b) a line-by-line radiative transfer simulation. (c) A difference plot (ARIES − simulation).

Fig. 16.

Comparison of ARIES BT data (a) in the region of the 670 cm−1 CO2 feature with (b) a line-by-line radiative transfer simulation, which includes a warm pod layer. See text for details.

Fig. 16.

Comparison of ARIES BT data (a) in the region of the 670 cm−1 CO2 feature with (b) a line-by-line radiative transfer simulation, which includes a warm pod layer. See text for details.

c. Future applications

Of fundamental interest to UKMO are improvements in the ability to assimilate data from satellite sounders into NWP models either by direct assimilation of the raw radiance data or via an intermediate fast modeling stage. Either way, to extract the promised vertical and absolute resolution in temperature and humidity retrievals from new high-spectral resolution infrared sounders such as IASI and AIRS, it is likely that improvements will need to be made to knowledge of atmospheric spectroscopy and radiative transfer (Han et al. 1997; Husson et al. 1992; Jaquinet-Husson et al. 1999; Rothman et al. 1992, 1998). For example, the 670 cm−1 (15 μm) CO2 band will be used for temperature sounding; however, the presence of several hot bands and Q branches renders simulation of atmospheric brightness temperature spectra difficult to achieve at the noise levels of ARIES/IASI-type instruments. Similarly, aspects of the spectroscopy of water vapor, such as air-broadened line widths in the fundamental υ2 (bending) region and line and so-called continuum absorption in the window region are very likely to require improvement. Also of concern are minor atmospheric component species such as ozone, nitrous oxide, methane, and chlorofluorocarbons, as well as major species such as oxygen and nitrogen that have weak continuum absorption. All these gases will require some improvement in their spectral data, though possibly with a lower priority than water and CO2. Considerable effort on some of these issues already has been applied within the U.S. Atmospheric Radiation Measurement program and in preparation for the AIRS satellite instrument (Strow et al. 1998a, b; Tobin et al. 1996).

With this in mind, a flight campaign was planned that involved the detailed comparison of ARIES clear air observations with numerous atmospheric truth measurements, to test thoroughly and to identify failings in current line-by-line radiative transfer models. The truth measurements included aircraft in situ temperature, humidity, and ozone measurements, aircraft dropsondes, and the aircraft microwave radiometers. The sorties were flown over a very well instrumented coastal ground site, equipped with facilities to launch radiosondes and to make lidar measurements of water vapor and ozone concentration (Bösenberg 1997). This Validation of IASI Radiative Transfer: Experiments and Modeling campaign took place in September 1998.

Other important issues to be studied using ARIES include cloud clearing and microphysics retrievals, surface emissivity measurements, and aerosol. Clouds present several problems in the assimilation of infrared satellite data into NWP models. When clouds are opaque and completely fill the field of view, no information is retrievable from below the cloud. Even when the cloud is semitransparent or only partially filling the field of view, the uncertainties in the cloud microphysical structure and in modeling the radiative transfer through and around the cloud lead to increased uncertainties in other retrieved variables. When cloud-top temperatures are close to the expected surface temperature, when cloud covers only a small fraction of the area, or when the cloud is nearly (but not quite) transparent, the field of view may mistakenly be taken as free of cloud. At UKMO (Gadd et al. 1995), data from any particular sounder field of view that appear to be affected by cloud are assimilated with a reduced number of channels—using only those microwave and stratospheric channels that are unaffected by cloud. This approach reduces the quantity of data assimilated but it does avoid degrading the quality of the assimilation. Accurate cloud detection clearly is vital in this process (Smith and Eyre 1996). ARIES will be used to investigate the signature of cloud, concentrating on the most difficult cases such as low stratocumulus and semitransparent cirrus.

Characterization of surface emissivity is of importance given the influence such characterization has on the observed infrared spectra (Smith et al. 1996). The spectral resolution of ARIES is sufficiently high that the system can be used in a novel manner, described by Fiedler and Bakan (1997), to measure sea surface emissivity. This method is based on the fact that emission from the sea surface in the atmospheric window region (∼750–1150 cm−1) should result in a smooth spectrum;however, the reflected components from the downwelling atmospheric radiation will contribute distinct spectral features that add to the otherwise featureless spectrum. The magnitude of the reflected features clearly is sensitive to the reflectivity and, hence, emissivity of the surface. This multispectral method therefore allows the sea surface temperature and emissivity effects on the observed spectrum to be separated. It is planned to use this method to study sea surface emissivity with ARIES in its airborne role to observe the effect of a wide range of conditions (e.g., scan angle, sea state), but also to extend the method to study different land types. This aim is desirable given that, as with other NWP centers, infrared satellite sounder data for the troposphere recorded over land rather than over sea are not assimilated into the UKMO weather prediction models (Gadd et al. 1995). This omission is because of inadequate characterization of the complex influence of the land surface emissivity and the lack of a sufficiently detailed surface emissivity global map.

Acknowledgments

The authors thank their colleagues at the Remote Sensing branch of UKMO and at BOMEM, Inc. for helpful discussions. The authors also thank the flight crew of the UKMO C-130 for their support during the measurements described in this paper.

REFERENCES

REFERENCES
Abrams, M. C., C. G. Toon, and R. A. Schindler, 1994: Practical example of the correction of FT spectra for detector nonlinearity. Appl. Opt., 33, 6307–6314.
Amato, U., V. Cuomo, and C. Serio, 1995: Assessing the impact of radiometric noise on IASI performances. Int. J. Remote Sens., 16, 2927–2938.
Atkinson, N. C., and S. H. S. Wilson, 1997: Measurement of interferometer instrument lineshape. Proc. IGARSS ’97, 1997 Int. Geoscience and Remote Sensing Symp. on Remote Sensing—A Scientific Vision for Sustainable Development, Singapore, IEEE, 2011–2014.
Aumann, H. H., and R. J. Pagano, 1994: Atmospheric Infrared Sounder on the Earth Observing System. Opt. Eng., 33, 776–784.
Bösenberg, J., 1997: Remote sensing of the troposphere using differential absorption LIDAR. Max-Planck-Institut für Meteorologie, Report 236, 134 pp. [Available from Max-Planck-Institut für Meteorologie, Bundesstr. 55, D-20146 Hamburg, Germany.]
.
Christensen, P. R., and J. C. Pearl, 1997: Initial data from the Mars Global Surveyor thermal emission spectrometer experiment: Observations of the Earth. J. Geophys. Res., 102, 10 875–10 880.
Diebel, D., F. Cayla, and T. Phulpin, 1996: IASI mission rationale and requirements. Tech. Rep. IA-SM-0000-10-CNE/EUM Issue 4a, EUMETSAT/Météo-France/CNES, 35 pp. [Available from EUMETSAT, Am Kavalleriesand 31, D-64295, Darmstadt, Germany.]
.
Edwards, D. P., 1992: GENLN2: A general line by line atmospheric transmittance and radiance model. Tech. Note 367, NCAR, 158 pp. [Available from National Center for Atmospheric Research, P.O. Box 3000, Boulder, CO 80307-3000.]
.
English, S. J., 1995: Airborne radiometric observations of cloud liquid-water emission at 89 and 157 GHz: Application to retrieval of liquid-water path. Quart. J. Roy. Meteor. Soc., 121, 1501–1524.
Eyre, J. R., 1990: The information content of data from satellite sounding systems: A simulation study. Quart. J. Roy. Meteor. Soc., 116, 401–434.
Fiedler, L., and S. Bakan, 1997: Interferometeric measurements of sea surface temperature and emissivity. Dtsch. Hydrog. Z., 49, 1–9.
Gadd, A. J., B. R. Barwell, S. J. Cox, and R. J. Renshaw, 1995: Global processing of satellite sounding radiances in a numerical weather prediction system. Quart. J. Roy. Meteor. Soc., 121, 615–630.
Griffiths, P. R., and J. A. de Haseth, 1986: Fourier transform infrared spectrometry. Vol. 83, Chem. Analysis, John Wiley & Sons, 671 pp.
Gunson, M. R., and Coauthors, 1996: The Atmospheric Trace Molecule Spectroscopy (ATMOS) experiment—Deployment on the ATLAS Space Shuttle missions. Geophys. Res. Lett., 23, 2333–2336.
Han, Y., J. A. Shaw, J. H. Churnside, P. D. Brown, and S. A. Clough, 1997: Infrared spectral radiance measurements in the tropical Pacific atmosphere. J. Geophys. Res, 102 (D4), 4353–4356.
Hanel, R. A., B. Schlachman, D. Rogers, and D. Vanous, 1971: The Nimbus 4 Michelson interferometer. Appl. Opt., 10, 1376–1382.
——, B. J. Conrath, V. G. Kunde, C. Prabhakara, I. Revah, V. V. Salomonson, and G. Wolford, 1972: The Nimbus 4 infrared spectroscopy experiment 1: Calibrated thermal emission spectra. J. Geophys. Res., 77, 2629–2641.
Hewison, T. J., and S. J. English, 1999: Results of a microwave airborne campaign over snow and ice. IEEE Trans. Geosci. Remote Sens., 37, 1871–1879.
Huang, H.-L., W. L. Smith, and H. M. Woolf, 1992: Vertical resolution and accuracy of atmospheric infrared sounding spectrometers. J. Appl. Meteor., 31, 265–274.
Husson, N., B. Bonnet, N. A. Scott, and A. Chedin, 1992: Management and study of spectroscopic information: The GEISA program. J. Quant. Spectros. Radiat. Transfer, 48, 509–518.
Jaquinet-Husson, N., and Coauthors, 1999: The 1997 spectroscopic GEISA databank. J. Quant. Spectros. Radiat. Transfer, 62, 205–254.
Korb, A. R., P. Dybwad, W. Wadsworth, and J. W. Salisbury, 1996:Portable Fourier transform infrared spectroradiometer for field measurements of radiance and emissivity. Appl. Opt., 35, 1679–1692.
Moncet, J. L., and S. A. Clough, 1997: Accelerated monochromatic radiative transfer for scattering atmospheres: Application of a new model to spectral radiance observations. J. Geophys. Res., 102, 21 853–21 866.
Neuendorffer, A. C., 1989: The feasibility of 1-deg/1-km tropospheric retrievals. RSRM ’87: Advances in Remote Sensing Retrieval Methods, A. Deepak, H. E. Fleming, and J. S. Theon, Eds., A. Deepak, 323–335.
Ogawa, T., H. Shimoda, M. Hayashi, R. Imasu, A. Ono, S. Nishinomya, and H. Kobayashi, 1993: IMG, Interferometric measurement of greenhouse gases from space. Adv. Space Res., 14, 25–28.
Prunet, P., J.-N. Thépaut, and V. Cassé, 1998: The information content of clear sky IASI radiances and their potential for numerical weather prediction. Quart. J. Roy. Meteor. Soc., 124, 211–241.
Rahmelow, 1997: Electronic influences on an infrared detector signal:Nonlinearity and amplification. Appl. Opt., 36, 2123–2132.
Revercomb, H. E., H. Buijs, H. B. Howell, D. D. LaPorte, W. L. Smith, and L. A. Sromovsky, 1988: Radiometric calibration of IR Fourier transform spectrometers: Solution to a problem with the high-resolution sounder. Appl. Opt., 27, 3210–3218.
Rothman, L. S., and Coauthors, 1992: The HITRAN molecular database: Editions of 1991 and 1992. J. Quant. Spectros. Radiat. Transfer, 48, 469–507.
——, and Coauthors, 1998: The HITRAN molecular spectroscopic database and HAWKS (HITRAN Atmospheric Workstation). J. Quant. Spectros. Radiat. Transfer, 60, 665–710.
Smith, J. A., and J. R. Eyre, 1996: Initial cloud detection by comparing brightness temperatures derived from a numerical weather prediction model with those observed. Proc. Satellite Remote Sensing and Modeling of Clouds and the Atmosphere, SPIE Proceedings, Vol. 2961, Taormina, Italy, Int. Soc. Opt. Eng., 21–32.
Smith, W. L., H. M. Woolf, C. M. Hayden, D. Q. Wark, and L. M. McMillin, 1979: The TIROS-N operational vertical sounder. Bull. Amer. Meteor. Soc., 60, 1177–1187.
——, H. E. Revercomb, R. O. Knuteson, F. A. Best, R. Dedecker, H. B. Howell, and H. M. Woolf, 1995: Cirrus cloud properties derived from high spectral resolution infrared spectrometry during FIRE II. Part I: The High Resolution Interferometer Sounder (HIS) systems. J. Atmos. Sci., 52, 4238–4245.
——, and Coauthors, 1996: Observations of the infrared radiative properties of the ocean—Implications for the measurement of sea surface temperature via satellite remote sensing. Bull. Amer. Meteor. Soc., 77, 41–51.
Strow, L. L., D. C. Tobin, W. W. McMillan, S. E. Hannon, W. L. Smith, H. E. Revercomb, and R. O. Knuteson, 1998a: Impact of a new water vapor continuum and line shape model on observed high resolution infrared radiances. J. Quant. Spectros. Radiat. Transfer, 59, 303–317.
——, H. E. Motteler, R. G. Benson, S. E. Hannon, and S. DeSouza-Machado, 1998b: Fast computation of monochromatic infrared atmospheric transmittances using compressed look-up tables. J. Quant. Spectros. Radiat. Transfer, 59, 481–493.
Tobin, D. C., L. L. Strow, W. J. Lafferty, and W. B. Olson, 1996: Experimental investigation of the self- and N–2-broadened continuum within the υ2 band of water vapor. Appl. Opt., 35, 4724–4734.

Footnotes

Corresponding author address: Dr. J. A. Smith, Remote Sensing Branch, Meteorological Office, Building Y70, DERA, Farnborough, Hampshire GU14 0LX United Kingdom.