The ability to accurately validate high–spectral resolution infrared radiance measurements from space using comparisons with a high-altitude aircraft spectrometer has been successfully demonstrated. The demonstration is based on a 21 November 2002 underflight of the AIRS on the NASA Aqua spacecraft by the Scanning-HIS on the NASA ER-2 high-altitude aircraft. A comparison technique which accounts for the different viewing geometries and spectral characteristics of the two sensors is introduced, and accurate comparisons are made for AIRS channels throughout the infrared spectrum. Resulting brightness temperature differences are found to be 0.2 K or less for most channels. Both the AIRS and the Scanning-HIS calibrations are expected to be very accurate (formal 3-sigma estimates are better than 1 K absolute brightness temperature for a wide range of scene temperatures), because high spectral resolution offers inherent advantages for absolute calibration and because they make use of high-emissivity cavity blackbodies as onboard radiometric references. AIRS also has the added advantage of a cold space view, and the Scanning-HIS calibration has recently benefited from the availability of a zenith view from high-altitude flights. Aircraft comparisons of this type provide a mechanism for periodically testing the absolute calibration of spacecraft instruments with instrumentation for which the calibration can be carefully maintained on the ground. This capability is especially valuable for assuring the long-term consistency and accuracy of climate observations, including those from the NASA EOS spacecraft (Terra, Aqua and Aura) and the new complement of NPOESS operational instruments. The validation role for accurately calibrated aircraft spectrometers also includes application to broadband instruments and linking the calibrations of similar instruments on different spacecraft. It is expected that aircraft flights of the Scanning-HIS and its close cousin the NPOESS Airborne Sounder Test Bed (NAST) will be used to check the long-term stability of AIRS and the NPOESS operational follow-on sounder, the Cross-track Infrared Sounder (CrIS), over the life of the missions.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The need for higher accuracy and more refined error characterization of radiance measurements from space (and corresponding geophysical products) to improve both weather forecasting and climate change monitoring has led to a new emphasis on conducting direct tests of in-orbit performance, referred to as “validation.” Validation as described here involves collecting high-quality reference data from specially maintained airborne and ground-based facilities that can support refined analyses using a controlled set of well-understood measurements, instead of, for example, statistical analyses of data having inconsistent pedigree and unknown error characteristics. This is a positive trend that will help take full advantage of our satellite systems. The validation activities initiated for the NASA Earth Observing System (EOS) platforms, which are well underway for the Terra and Aqua platforms and now initiated for Aura, are setting the stage for enhanced validation of new observational satellite systems. Plans for validation of the National Polar-orbiting Operational Environmental Satellite System (NPOESS) and future geosynchronous systems are underway. The Scanning High-Resolution Interferometer Sounder (Scanning-HIS) aircraft instrument discussed in this paper is an important validation tool that is currently being used for both EOS and NPOESS. For NPOESS, Scanning-HIS use is coordinated with the NPOESS Airborne Sounder Test Bed (NAST) to optimize payload compatibility with joint field campaigns and for critical intercomparison tests of accuracy.
 The Atmospheric Infrared Sounder (AIRS) [Aumann et al., 2003] is the first of a new generation of high–spectral resolution infrared sounders. It operates on the EOS Aqua platform, making twice daily global observations from a 705 km sun synchronous polar orbit. It is a hyperspectral grating spectrometer that measures the thermal infrared spectrum from 3.75–4.59 μm (2181–2665 cm−1), 6.20–8.22 μm (1217–1614 cm−1), and 8.8–15.4 μm (650–1136 cm−1) with 2378 spectral channels with resolving power (λ/Δλ) ranging from 1080 to 1590 (full width at half maximum, FWHM, ranging from 0.41 to 2.18 cm−1). It has infrared footprints approximately 13.5 km in diameter at nadir and utilizes cross track scanning to collect 90 cross track footprints every 2.667 s with a swath width of ∼1650 km.
 In addition to the primary objective of improving weather forecasting, AIRS and its follow-on operational counterparts also have the potential to provide a long-term record of accurately calibrated spectral radiances for climate monitoring and other climate related studies. This is due in part to high spectral resolution, which offers inherent advantages for both radiometric [Goody and Haskins, 1998] and spectral calibration. Details of the AIRS radiometric calibration is provided by Pagano et al. . Both high sensitivity and calibration accuracy are required for atmospheric sounding. When minimizing differences between observed and calculated radiances, a detailed understanding of the various sources of bias (observation, surface and atmospheric state, clouds, and forward model) as a function of wave number is required in order to understand the error characteristics of the retrieved products. Assessment of the observation accuracy is the first step toward achieving this understanding. Therefore, since the launch of AIRS in May 2002, the Scanning-HIS has participated in several aircraft based field experiments that have provided data sets coincident with Aqua overpasses, with a primary objective of providing an accurate assessment of the AIRS radiances. This is both an accurate and a relatively straightforward approach to validating AIRS, providing comparisons of satellite based radiance measurements with similar radiance measurements made from a high-altitude aircraft. For nearly all spectral channels (with the exception of channels with significant sensitivity above the aircraft altitude, ∼20 km) it provides accurate comparisons without a first-order dependence on the accuracy of coincident atmospheric profile or surface measurements or models or on the absolute accuracy of radiative transfer calculations.
Section 2 provides an introduction to the Scanning-HIS design and calibration. In section 3, the Scanning-HIS/AIRS radiance comparison technique is introduced and an example validation case is presented. One of the first AIRS underflights conducted, this is a clear sky case over ocean with spatially homogeneous conditions, making it an ideal case for creating an accurate comparison and demonstrating the comparison approach. A summary of the results in given in section 4.
 This section provides an introduction to the Scanning-HIS and an overview of its design and calibration approach. The Scanning-HIS is a follow-on of the original University of Wisconsin HIS [Revercomb et al., 1988a, 1988b, 1996] that was flown successfully on the NASA ER2 aircraft from 1986 to 1998. Its design and calibration techniques have matured from experience with the HIS and from the ground based Atmospheric Emitted Radiance Interferometer (AERI) instruments developed for the DOE Atmospheric Radiation Measurement (ARM) program. The nadir-only spatial sampling of the original HIS has been replaced by programmable cross-track coverage with similar ∼2 km footprints, while at the same time the Scanning-HIS is smaller, more robust, and easier to operate. In addition to the NASA ER2, the Scanning-HIS has been successfully flown on the NASA DC8, the Scaled Composites/Northrop Grumman Proteus, and most recently on the NASA WB-57. Applications of the data include satellite validation, development and validation of clear and cloudy sky absorption models, retrieval of atmospheric state, surface, and cloud properties, and design trades and risk reduction for future satellite sensors.
 The basic components of the Scanning-HIS include a scene mirror module, telescope, Michelson plane mirror interferometer, aft optics, detector module with mechanical cooler, laser metrology, calibration blackbodies, and onboard signal processing and solid state storage. Figure 1 is a diagram of the Scanning-HIS opto-mechanical layout. Following the basic optical flow, scene radiance incident on a 45° gold-coated scene mirror passes through an afocal telescope, through the interferometer, and is focused at the entrance to the detector module where a field stop is located. The field stop limits angles through the interferometer to 40 mrad (full angle). The telescope has an afocal ratio of 2.5 resulting in a 100 mrad spatial field of view, producing 2 km diameter nadir footprints from an altitude of 20 km. A 4.5 cm aperture stop is located on the fixed mirror of the interferometer, and its image is focused onto the detectors via refractive optics within the detector assembly. The detector package is a “sandwich” configuration with three physically overlapping detectors with a shared focal plane, eliminating the need for multiple coolers and dichroic beamsplitters and providing a compact and simplified optical design. All three detectors share the same physical field and aperture stops. Two high-emissivity calibration blackbodies, one at flight ambient temperature (ABB) and one maintained at an elevated temperature (HBB), are viewed by rotation of the scan mirror. The scan mirror sequence is programmable; the sequence used for the validation case presented in section 3 consists of 13 cross track Earth scene views followed by 5 HBB views and 5 ABB views. On the Proteus and WB-57, zenith view data may also be collected. Wavelength separation is provided by a dynamically aligned plane mirror interferometer with a voice coil driven linear slide mechanism. A stable HeNe metrology laser source and detectors are introduced along the center of the interferometer optical axis for fringe counting and for dynamic alignment of the fixed mirror. The scan speed is 4 cm/s and the maximum optical path difference (MOPD) is ±1.037 cm, resulting in the collection of an interferogram with spectral resolution (1/2/MOPD) of about 0.48 cm−1 (FWHM ≈ 0.58 cm−1) every 0.5 s. The raw interferograms from each view are compressed in real time using a numerical filter and decimation process performed on a digital signal processor (DSP), while a second DSP is used for controlling the instrument.
 As with the original HIS and ground-based AERI instruments, accurate calibration has been a primary objective of the Scanning-HIS. For a comprehensive description of the AERI calibration approach and algorithms, which are very closely related to those used for the Scanning-HIS, the reader is referred to Knuteson et al. [2004a, 2004b]. Accurate radiometric calibration of the Scanning-HIS is achieved primarily by frequent viewing of the onboard blackbody calibration references. The ABB is unheated but is thermally coupled to the aircraft pod temperature while the HBB is maintained at approximately 310 K when in flight. The onboard blackbodies were built at the University of Wisconsin Space Science and Engineering Center with NIST traceable calibration and have formal 3 sigma (i.e., not to exceed) absolute uncertainties for temperature and cavity emissivity of 0.10 K and 0.001, respectively [Best et al., 1997, 2003]. Linear calibration using these references and proper handling of phase is achieved following the complex calibration approach introduced by Revercomb et al. [1988a]. Nonlinearity corrections are applied to the photoconductive HgCdTe longwave and midwave band complex spectra following the same general approach used for the AERI instruments.
 In addition to the onboard calibrations, characterization tests are performed at the University of Wisconsin before and after field campaigns to verify radiometric nonlinearity, spectral calibration, and other issues relevant to the calibration. The predeployment and postdeployment efforts typically include blackbody thermistor and electronics calibration, blackbody alignment verification, views of external calibration targets (LN2 bath and high-emissivity blackbodies at various temperatures), analysis of out-of-band harmonics, and comparisons of ground-based zenith view data with coincident clear sky data measured by an AERI. Along with other analyses, the character and magnitude of the radiometric nonlinearity of the longwave and midwave bands (and the linearity of the shortwave InSb band) are characterized using these data. A check of the Scanning-HIS calibration by direct comparison to a NIST-maintained TXR radiometer is planned for later this calendar year. Previous comparisons of the University of Wisconsin designed blackbodies against NIST maintained standards have shown agreement to within 0.03 K over a range of blackbody temperatures [Minnett et al., 2001].
 Accuracy of the in-flight data is also assessed directly by inspection of the imaginary part of the calibrated spectra (which can be used to diagnose phase errors), inspection of spectral regions of detector band overlap (1030–1200 cm−1 and 1760–1810 cm−1), and inspection of numerous housekeeping variables. Recently added is the ability to collect and record 128 unfiltered interferogram data points (bypassing the numerical filter and decimation processes) centered around zero optical path difference, allowing the out-of-band harmonics to be monitored. When flying on the Proteus and WB-57 aircraft, zenith view sky spectra collected from high altitude can serve as a pseudo space view and are used to assess the calibration in spectral regions with low opacity.
 Spectral integrity of the Scanning-HIS spectra is achieved through use of a stable HeNe metrology laser, which provides fixed interval optical path difference sampling. Self-apodization effects are removed on the basis of the optics geometry (yielding constant spectral resolution and line shape as a function of wave number) and the spectra are resampled to a standard wave number grid following the same approaches as used for AERI. The absolute spectral calibration is determined by adjusting the effective metrology laser frequency to create optimal agreement with the positions of well-known spectral features present in clear sky calculated spectra. Analysis of ensembles of such cases [Tobin et al., 2003] has determined the Scanning-HIS spectral calibration with an uncertainty of ±0.5 ppm (3 sigma uncertainty in the mean) with no detectable changes with time.
3. Validation of AIRS With the Scanning-HIS, 21 November 2002 Gulf of Mexico Case
 The best conditions for validation of AIRS with Scanning-HIS to date were encountered over the Gulf of Mexico on 21 November 2002 as part of the Texas Aqua Experiment (TX2002). This case has also been used by Moeller et al.  to provide an assessment of the Aqua Moderate resolution Imaging Spectroradiometer (MODIS) calibration. The logistics of this case are illustrated in Figure 2. During this mission, the ER-2 flew a straight and level flight leg at ∼20.0 km altitude (50 mbar) along the suborbital track of EOS Aqua. Coincidence of the ER-2 and EOS Aqua nadir tracks occurred at approximately 1940 UTC (early afternoon). Various data sets from the ER-2 payload including MODIS Airborne Simulator (MAS) 50 m visible and infrared data, and the Aqua AIRS and MODIS data have been analyzed to assess the cloudiness of this scene, leading to the conclusion that this is a purely clear sky scene. Selection of which AIRS footprints to include in the comparisons takes into consideration the spatial and temporal collocation with the Scanning-HIS observations, spatial uniformity of the scene, and measurement noise reduction. The use of spatially uniform scenes reduces differences due to collocation errors and also removes the need for exact representation of the Scanning-HIS and AIRS footprints. Using coincident MODIS data to assess spatial uniformity, eight near nadir AIRS footprints collected between 1940:31 and 1940:39 UTC were selected. Even though this is a relatively spatially uniform scene, the range in MODIS window region brightness temperatures within these eight AIRS footprints is ∼0.3 K. Therefore it is necessary that the Scanning-HIS footprints provide full coverage of the AIRS footprints, and multiple Scanning-HIS footprints collected over a range of scan angles are required. A total of 416 Scanning-HIS footprints providing nearly contiguous coverage of the eight AIRS footprints are selected. These data were collected from 1934:22 to 1939:47 UTC and include 13 distinct view angles ranging from approximately −36 to +36° off nadir. In addition to coadding multiple footprints, random measurement noise in the AIRS and Scanning-HIS observations is further reduced using the Principle Component Noise Filter algorithm described by Antonelli et al. . This algorithm removes a large percentage of the spectrally random noise while retaining the spectral and radiometric fidelity of the spectra.
 Computing the mean of the eight selected AIRS footprints and of the coincident Scanning-HIS spectra, a comparison of the AIRS and Scanning-HIS spectra is shown in Figure 3. The comparison shown in Figure 3 presents the AIRS and Scanning-HIS data at their native spectral resolutions and sampling and without any attempt to account for the different viewing geometries. In spectral regions where AIRS has significant contributions from above the aircraft altitude (e.g., 4 and 15 μm CO2, 9 μm O3), large differences are observed, as expected. In spectrally flat window regions and regions where the spectral resolutions are similar (e.g., 1400–1600 cm−1) good agreement is apparent. The remainder of this section presents the Scanning-HIS radiometric uncertainty estimates for this case and then presents a more detailed comparison, taking into consideration the differences between AIRS and Scanning-HIS viewing geometries and spectral characteristics between the two sensors.
 Uncertainty in the radiometric accuracy of the Scanning-HIS data are estimated by a perturbation analysis of the calibration equation [Revercomb et al., 1988a] as applied to this Earth scene data and confirmed using the in-flight spectral band overlap agreement. Time series of various Scanning-HIS instrument temperatures are shown in Figure 4. Note that the temperatures of the ambient blackbody, optics bench, and blackbody support structure (near the scene mirror) vary significantly throughout the duration of the flight. As discussed in section 2, frequent viewing of the two calibration blackbodies and use of a stable laser allows for robust calibrations throughout the flight. At the time of the Aqua overpass, temperatures required for the radiometric calibration are approximately 310 K (HBB temperature), 258 K (ABB temperature), and 256 K (“reflected” temperature, i.e., temperature outside blackbody aperture). Using formal 3 sigma absolute uncertainties of the blackbody temperatures (0.1 K), reflected temperature (5.0 K), and blackbody cavity emissivities (0.001), uncertainties of the mean Scanning-HIS Earth scene spectrum are computed by perturbation of the linear calibration equation. Uncertainty due to the detector nonlinearity correction is computed as 10% of the full correction for this case (i.e., 10% of difference between calibrations performed with and without the nonlinearity corrections). The total uncertainty estimate is then computed as RSS (Root Sum of Squares) of the individual components. These uncertainties are converted to brightness temperature and are shown in Figure 5 as a spectrum and versus scene brightness temperature. Errors are minimized between the two onboard blackbody temperatures but grow rapidly when extrapolating to colder scenes. In this case the ambient blackbody was at 258K, so we find that for scene temperatures greater than 250 K (235 K), the 3 sigma total uncertainty is less than 0.15 K (0.3 K). Different nonlinearities for the longwave and midwave bands result in the discontinuity observed at 1070 cm−1. For lower scene temperatures, the uncertainties are grouped according to the three distinct regions of the spectrum with low scene temperatures (15 μm CO2, 5–7 μm H2O, 4.5 μm CO2) and the wavelength dependence of the Planck function. Improved coupling of the ambient blackbody to the ambient air temperature has led to significant improvement in the Scanning-HIS radiometric performance on subsequent flights. Note that for carefully calibrated infrared spacecraft instruments, the accuracy should be relatively better for cold scene temperatures, because space provides a well-known cold reference.
 Comparison of the longwave and midwave bands and of the midwave and shortwave bands of Scanning-HIS in the spectral overlap regions are shown in Figure 6. The shortwave InSb band is linear and shows good agreement with the nonlinear midwave band. The midwave and longwave bands have nonlinearity of different magnitudes, and also show good agreement. Differences in both overlap regions are random (spectrally) in nature with mean values <0.03 K. This band overlap agreement is a consistency check on the nonlinearity correction used in Scanning-HIS longwave and midwave bands.
 The Scanning-HIS and AIRS have different observation altitudes, footprint sizes, and spectral characteristics. These are illustrated in Figure 7. As discussed previously, to avoid collocation errors when creating comparisons of AIRS and Scanning-HIS observations, temporally and spatially coincident data collected under clear sky and spatially uniform conditions are typically used. However, despite a careful selection of such conditions, surface emission variations within the AIRS footprints can be significant. To ensure that both sensors are looking at the same scene it is necessary to use multiple ∼2 km Scanning-HIS footprints collected over a range of scan (view) angles to provide contiguous coverage of the ∼13.5 km AIRS footprints. In addition to the different observation altitudes and spectral characteristics, these differences also need to be accounted for when comparing AIRS and Scanning-HIS spectra. The technique selected for doing this is to make use of calculations that include the actual spectral and spatial characteristics of each instrument. The calculated spectra allow the observed minus calculated residual for each instrument to be compared, avoiding the first-order effects of the altitude and view angle differences. To further improve the comparison, the observed and calculated spectra of each instrument are convolved with the Spectral Response Functions (SRFs) of the other. This is equivalent to eliminating AIRS grating contributions from optical path differences larger than those measured by Scanning-HIS and apodizing the Scanning-HIS spectra to match the effect of the AIRS SRFs. The expression for the final radiance differences, RDIFF, is given by
where RAIRS is the mean AIRS radiance spectrum for a selection of spatially uniform AIRS footprints, RSHIS is the mean Scanning-HIS radiance spectrum of footprints collocated with the AIRS footprints, RAIRS′ are simulated AIRS radiances using monochromatic calculations performed for the AIRS viewing geometry and convolved with the AIRS SRFs, RSHIS′ are simulated Scanning-HIS radiances using monochromatic calculations performed for the Scanning-HIS viewing geometries and reduced to Scanning-HIS spectral resolution, and ⊗ denotes a spectral convolution. In practice, RAIRS ⊗ SRFSHIS is implemented using SRFSHIS (sinc functions) centered on the AIRS channel centroids, and spectrally oversampled Scanning-HIS spectra are used when computing RSHIS ⊗ SRFAIRS. RDIFF is therefore computed on the AIRS spectral grid. Since RSHIS is the mean of many individual spectra collected over a range of view angles, RSHIS′ is computed as the mean of individual calculations performed for the same set of view angles. AIRS SRFs are provided by the AIRS project and are computed with a sensor model for a grating temperature of 155.1325 K, a grating/detector focal plane offset of −13.5 microns, and filter window temperature of 156.157 K [Strow et al., 2003]. AIRS channels not recommended for use in geophysical products retrievals by the AIRS Project (as noted by the “L2_Ignore” flag in the AIRS channel properties files) are excluded. Also, some numerical errors are encountered when convolving the spectrally discreet AIRS spectra with SRFSHIS; a small number of channels affected by this, typically on the edges of the AIRS detector arrays, are excluded.
 RDIFF is essentially the difference between AIRS and Scanning-HIS observed minus calculated residuals, reduced to lowest common spectral resolution. When performing the radiance calculations, the same surface conditions, atmospheric state, and forward model physics are used for both AIRS and Scanning-HIS. This results in systematic errors that are common to both sets of calculations, and to first order removes the fundamental effects of altitude and view angle differences from the comparison. For this case, the monochromatic calculations are performed using the LBLRTM line-by-line radiative transfer model [Clough et al., 2005] with coincident analysis fields (pressure, temperature, water vapor, ozone, surface pressure and temperature) from the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Masuda et al.  ocean emissivity. If there are significant errors in the assumed atmospheric state and/or forward model physics used in the calculations, however, two types of biases can be introduced. The first type of uncertainty arises when there are errors in the atmospheric state and/or forward model which manifest in errors in the computed emission above the aircraft, differentially affecting RAIRS′ versus RSHIS′. Examples are errors in the assumed stratospheric ozone concentrations and non-Local Thermodynamic Equilibrium (non-LTE) effects; issues specific to this case are discussed below. To eliminate this type of error, channels with sensitivity to altitudes above the aircraft altitude are objectively excluded from the final comparisons. A second-order effect can also arise because the Scanning-HIS view angles are larger than those of AIRS. Errors in computed radiances are amplified for large scan angles, therefore introducing a larger bias into RSHIS′ as compared to RAIRS′. To provide an upper bound on this effect, a 5% error in optical depths for all spectral regions is assumed. To simulate the biases that would arise in equation (1), calculations are performed for nadir and for the mean off-nadir Scanning-HIS angle, with and without 5% errors in the optical depths (e.g., Rnadir5% error − Roff-nadir5% error minus Rnadirno error − Roff-nadirno error). The magnitude of the resulting bias is 0.11 K or less for all spectral regions.
Figures 8–10 show the comparisons for the longwave, midwave, and shortwave spectral regions, respectively, where the differences in spectral resolution and viewing geometry between the two sensors have been taken into account following equation (1). Brightness temperatures and brightness temperature differences are shown in Figures 8–10, where the Planck function is used to convert from radiance to equivalent brightness temperature. The middle plots of each figure show the comparison of AIRS and Scanning-HIS observed minus calculated residuals. For most spectral regions, similar residuals are observed, because of common errors in the two sets of calculations. The bottom plots show the differences between AIRS and Scanning-HIS, without excluding channels with significant sensitivity above the aircraft. In the longwave spectral region, the 650–680 cm−1, 720 cm−1, and 1000–1070 cm−1 regions have significant contributions from above the aircraft and large differences are observed. This is attributed to a cold bias in the ECMWF stratospheric temperature fields [Dethof et al., 2004]. For other regions, generally good agreement is found. For AIRS detector module M-05 (1055–1140 cm−1), however, the differences show a dispersive behavior on spectral lines, which is indicative of a spectral calibration error. This and comparisons of clear sky observed and calculated spectra have demonstrated that the AIRS array module M-05 SRF centroids are off by approximately 3 percent of δν, which has since been corrected in the AIRS SRF tables. In the midwave spectral region, channels with significant emission from above the aircraft are mainly isolated to the 1305 cm−1 methane region. Good agreement is observed throughout the midwave region, including the mid and upper tropospheric water vapor regions. For the shortwave region, good agreement is not necessarily expected beyond ∼2500 cm−1 because this is a daytime case with solar contamination, and the AIRS and Scanning-HIS view angles are different. Channels from ∼2240 cm−1 to 2390 cm−1 have significant emission from above the aircraft, and additionally, the AIRS spectra in this region are affected by non-LTE which is not accounted for in the calculations while the Scanning-HIS spectra are not. Good agreement is found as expected for wave numbers less than ∼2240 cm−1.
 The final step in the comparison process is to objectively exclude AIRS channels with significant atmospheric emission contributions from above the ER-2 altitude. This is accomplished by computing AIRS radiances for a profile with perturbed values above 20 km (2 K increase in temperature, 30% increase in water vapor and ozone amounts) and with an unperturbed profile. Channels with differences in the perturbed calculations of 0.1 K or less are retained in the final comparisons, which are presented in Figures 11–13. Channels with solar contamination are excluded. Figure 11 displays the brightness temperature differences as a spectrum. (For these comparisons, the AIRS M-05 SRF centroids have now been shifted by 3% of δν). Channels are retained starting from 706 cm−1 in the longwave temperature sounding region, throughout the longwave window and midwave water vapor sounding channels, and on the longwave end of array M-02b. For the midwave upper water vapor band, all channels except those on the centers of the strongest water vapor lines are retained. Distributions of differences for each AIRS detector array are shown in Figure 12. Mean differences are ∼0.1 K or less for all arrays with the exception of array M-04b (1460–1525 cm−1) that has a mean difference of ∼0.2 K. Standard deviations are 0.25 K or less for all arrays. Figure 13 presents the AIRS minus Scanning-HIS differences as a function of scene brightness temperature. Only a small systematic dependence in the midwave region is observed. However, considering the formal 3- sigma Scanning-HIS uncertainty estimates for this case and the possible biases in the comparison process due to differing scan angles discussed previously, most of the observed differences between AIRS and Scanning-HIS are not significant. Largest differences for individual channels are found in regions with spectral structure; these differences could possibly be due to AIRS, Scanning-HIS, the comparison technique, or most likely some combination of all three, and require further investigation to resolve.
 In this paper, an approach to validate satellite based AIRS high spectral resolution infrared observations with the aircraft based Scanning-HIS has been introduced and an example case from 21 November 2002 over the Gulf of Mexico has been presented. This is a daytime case with clear skies and high spatial uniformity and good temporal and spatial collocation of the two sets of observations. The comparisons are not limited to window regions channels; accurate comparisons are made for spectral channels throughout the infrared spectrum including nearly all channels used for sounding of the surface to upper troposphere. Resulting brightness temperature differences are found to be 0.2 K or less for most spectral channels, and nearly all differences fall within the Scanning-HIS 3 sigma absolute radiometric uncertainty estimate and within the uncertainty of the comparison process.
 Aircraft comparisons of this type provide a mechanism for periodically testing the absolute calibration of spacecraft instruments with instrumentation for which the calibration can be carefully maintained on the ground. This capability is especially valuable for assuring the long-term consistency and accuracy of climate observations, including those from the NASA EOS spacecrafts (Terra, Aqua and Aura) and the new complement of NPOESS operational instruments. The validation role for accurately calibrated aircraft spectrometers also includes application to broadband instruments and linking the calibrations of similar instruments on different spacecraft. The very good agreement observed here between AIRS and Scanning-HIS throughout the longwave CO2 and midwave H2O spectral regions is particularly relevant given observed radiometric differences between AIRS and Aqua MODIS [Tobin et al., 2006]. In addition to the case presented here, approximately 15 other AIRS underflights suitable for radiance validation have been collected to date. These include a range of conditions including nighttime cases (allowing the shortwave channels to be validated) and cases of cold, uniform stratus clouds in the Arctic, as shown in Figure 14. A detailed analysis of these cases will be presented in a subsequent paper. Quick-look analyses suggest agreement between AIRS and Scanning-HIS on the order of 0.2 K for these cases, similar to the case presented here. Good radiometric stability of AIRS window region channels over the AIRS lifetime has been shown by H. H. Aumann et al. (unpublished manuscript, 2006) using comparisons to sea surface temperatures. It is expected that aircraft flights of the Scanning-HIS and NAST will be used to check the long-term stability of AIRS and the NPOESS operational follow-on sounder, the Cross-track Infrared Sounder (CrIS), over the life of the missions.
 We gratefully acknowledge the Department of Energy Atmospheric Radiation Measurement (ARM) program, NASA, and the Integrated Program office for support of the development and field deployment of the Scanning-HIS. This research was supported by the EOS Science Project Office under NASA contracts NAS5-31375 and NNG04GG31G and by the Integrated Program Office under contract 50-SPNA-1-00039. The authors would like to extend our thanks to George Aumann for discussions regarding various aspects of this work. Chris Moeller provided science and ER-2 flight coordination during TX-2002. Thanks to the Atmospheric Spectroscopy Laboratory, University of Maryland Baltimore County, for supplying various AIRS software modules and to the European Centre for Medium-Range Weather Forecasts for use of analysis fields.