1. Introduction
One of the key instruments supported by the U.S. Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) instrument development program was the Atmospheric Emitted Radiance Interferometer (AERI) (Stokes and Schwartz 1994). The AERI instrument measures the downwelling atmospheric emission spectrum at the surface with high spectral resolution and high absolute accuracy. The design of the AERI instrument meeting the requirements of the ARM Program is described in a companion paper (Knuteson et al. 2004, hereafter referred to as Part I). All the AERI systems built by the University of Wisconsin Space Science and Engineering Center (UW-SSEC) for the ARM Program include custom data processing software used to produce calibrated radiance spectra in real time. A description of the real-time data processing flow used in the AERI system is provided in Minnett et al. (2001). In this paper, the authors describe the algorithms used in the real-time data processing to achieve the desired radiometric performance. The actual performance achieved by the various AERI instruments built at UW-SSEC is presented, along with appropriate calibration verification data.
2. AERI performance
The AERI is a ground-based Fourier transform spectrometer (FTS) for the measurement of atmospheric downwelling infrared thermal emission at the earth's surface. As described in Part I, the AERI instrument was built as an operational facility instrument for the ARM Program for the routine measurement of downwelling infrared radiance to better than 1% absolute accuracy. This section summarizes the detailed performance characteristics of each of the eight AERI instruments built for the ARM Program based upon laboratory tests and clear-sky intercomparisons performed at UW-SSEC prior to the delivery of each instrument. The details presented here represent the AERI instrument performance at the time of initial deployment into the ARM field network. A characterization of the data record of field observations of the AERI instruments within the ARM network is deferred to a future paper.
a. Radiometric performance
This section describes how the performance of the AERI systems meets the requirements listed in Part I for the production of calibrated infrared radiance spectra. The AERI real-time data processing applications convert the raw interferometer data to calibrated radiances by implementing a sequence of operations including 1) correction for detector nonlinearity in the longwave band, 2) radiometric calibration using the onboard reference blackbodies, 3) correction for spectral line shape effects, and 4) resampling of the radiance spectra to a common wavenumber grid. Details of the AERI data processing algorithms are presented in the following sections on system linearity, radiometric calibration, spectral coverage and instrument line shape, wavenumber calibration, noise, and reproducibility.
1) Nonlinearity correction
The AERI system uses a mercury cadmium telluride (HgCdTe) and indium antimonide (InSb) detector package. Each detector uses separate preamplifiers, which are linear by design. The InSb detector response is inherently linear; however, the HgCdTe detector response is known to exhibit nonlinear behavior. The system linearity of the longwave detector band has been characterized in each of the AERI systems using reference observations at +60°C (hot), +20°C (ambient), and near 77 K (cold). The size of the nonlinearity effect in the calibrated AERI longwave radiance is relatively small (order 1%–2% of ambient radiance), but the absolute calibration requirement of “better than 1% of ambient radiance” for all scene conditions implies that a nonlinearity correction is necessary. The goal has been to characterize the nonlinearity of each longwave detector to better than 10%, which, when that knowledge is applied as a nonlinearity correction, implies an uncertainty contribution to the final calibrated radiance of less than about 0.2% of ambient radiance.
While, in principle, both the in-band and out-of-band responsivity could be used to determine the nonlinearity coefficient, a2, in practice the uncertainty of the spectral emissivity of the LN2 cold blackbody is too large to allow the use of an in-band fit. Fortunately, the out-of-band signal for the quadratic nonlinearity provides an unambiguous determination of a2, largely independent of the issues that affect the in-band signal. Figure 1 shows a least squares regression fit for a2 to the difference of measured responsivities in the 200–460 cm−1 region using Eq. (3) to determine the quadratic nonlinearity coefficient. The large difference between the measured and modeled results between 700 and 1000 cm−1 is due to the absorption caused by the liquid water cloud (fog) that forms over the open-mouth LN2 dewar. For this and other reasons, a liquid nitrogen cold target does not make a suitable operational calibration reference. For the UW-SSEC AERI systems, the LN2 reference is used only in the determination of the instrument nonlinearity in the out-of-band region, which is relatively immune to the uncertainty in emissivity that impacts the in-band region.
The nonlinearity determined for each of the longwave detectors used in the AERI instruments is summarized in Table 1. The in-band nonlinearity correction factor, 2a2Vdc, was computed using the a2 value measured on the stated test date and dc-level values computed from Eq. (2) using I(0) equal IH(0), zero, and IC(0) to represent the hot, ambient, and LN2 blackbodies, respectively. These dc-level values span the range of nonlinearity corrections used in the calibration of atmospheric scenes. The nonlinearity corrections (as a percent of raw signal) vary from 1%–2% for the AERI-01 at the Southern Great Plains (SGP) Central Facility (CF) to 3%–6% for the Tropical Western Pacific (TWP) systems. Since the radiometric calibration is based on differences from the ambient blackbody, the effective correction is generally less than 2% of the ambient blackbody radiance. An example of an AERI nonlinearity correction is shown in Fig. 2 for the AERI-05 (Hillsboro) system. Figure 2 shows that the nonlinearity correction is important for the AERI system because the correction is of the same order of magnitude as the 1% absolute calibration specification.
2) Radiometric calibration
This section describes the methodology and error analysis associated with radiometric calibration and verification of the AERI instruments.
(i) Methodology
The AERI instruments are configured to operate on a repeating scene mirror schedule such that the scene being calibrated is bracketed by views of the onboard reference blackbodies. The standard AERI scene mirror schedule is a repeating sequence of the form HASAHS, where H, A, and S represent views to the hot blackbody, ambient blackbody, and sky positions, respectively. Multiple S views are also possible, with a practical limit imposed by the rate of drift of instrument temperatures during the calibration period. The standard view angles and dwell times for each mirror position are given in Table 2 for the AERI-01 system. There are two Michelson mirror sweeps (forward and backward) for each total scan. Dwell times in Table 2 are only approximate. The AERI calibration methodology is to define a calibration sequence composed of the scene view to be calibrated and the pair of hot and ambient blackbody views measured closest in time before and after each scene view. In order to account for changes in the instrument temperature during a calibration sequence, the blackbody temperature measurements are fit to a linear function of time, and this fit is evaluated at the center time of the sky view measurements. A similar linear interpolation to the sky view time is performed for each complex spectral element of the hot blackbody and ambient blackbodies.
(ii) Predicted calibration performance
A differential error analysis of the calibration equation was used to guide the instrument development of the AERI system. In particular, the accuracy of the reference blackbodies was chosen to ensure that the instrument measurements that enter into the calibration equation are adequate to meet the overall calibration requirements. If Nυ represents the calibrated radiance for a set of known blackbody temperature and emissivity values, then Eq. (4) can be used to write an equation for the radiance derived for a set of perturbed blackbody parameters. One can then compute the radiance perturbation ΔNυ = N′υ − Nυ for a range of scene temperatures by perturbing each parameter. Figure 4 shows the radiance errors as a percent of ambient blackbody radiance for the uncertainty estimates given in Table 3. Figure 4 also shows the combined error for this set of uncertainties as a root sum square of errors. The extrapolation due to the use of a hot blackbody (rather than a cold blackbody) causes the calibration error to increase when the scene temperature is below that of the ambient blackbody. However, the uncertainty analysis shows that the AERI system design will meet the ARM requirement of 1% of ambient radiance if the blackbodies achieve the accuracy defined in Table 3. Moreover, the scene radiance error is reduced as the ambient blackbody temperature decreases. This is particularly important for the Arctic, where the clear-sky scene radiance in the winter is close to zero in the window regions. This analysis suggests that the largest calibration error experienced by the AERI instrument is for hot, dry conditions where the ambient blackbody temperature is warm but the window radiances are very low. In contrast, the best absolute accuracy (as a percent) is obtained when scenes are close to the ambient blackbody temperature. This is the reason that the Marine-AERI system is able to provide such an accurate measurement of sea surface temperature (Minnett et al. 2001). A detailed analysis of calibration uncertainties of the AERI system under different operating conditions is deferred to a future paper.
(iii) Laboratory radiometric calibration verification
Prior to the deployment of each AERI instrument built for the ARM Program, an end-to-end calibration verification test was performed using UW-SSEC blackbodies as external reference sources. In the laboratory, the AERI hot blackbody is temperature controlled to about 333 K, while the ambient blackbody operates at room temperature (about 300 K). One of the external blackbodies is controlled to an intermediate temperature (about 318 K), while the second external blackbody reference is a cavity partially submerged in an ice slurry bath (273.15 K). The ice temperature cavity is operated with a purge of dry nitrogen to prevent condensation on the interior surfaces during the laboratory tests. The external reference sources are calibrated using the same (National Institute of Standards and Technology) NIST-traceable approach as the AERI onboard blackbody references. Figure 5 shows the typical setup for this “four body” test with the intermediate temperature external blackbody in the zenith position and the ice temperature blackbody in the nadir view position. For this test, the scene mirror is programmed to cycle through each of the internal and external blackbody view positions, with a dwell period in each position of about 100. Data are collected in a stable environment over a period of several hours in order to reduce the noise level on the mean measurement. During the test period the temperatures of the external blackbodies are recorded. The external blackbody temperatures are combined with a cavity emissivity model to predict the equivalent blackbody temperature that the AERI instrument should see. The predicted radiometric temperatures for the intermediate and ice blackbody are used as “truth” for this test.
An example calibration verification test result from 8 May 2001 for the AERI-06 instrument is shown in Fig. 6. The air path between the interferometer and the reference blackbodies is transparent for most spectral channels, with the exception of the water vapor band (1400–1900 cm−1) and the carbon dioxide bands at 667 cm−1 and near 2380 cm−1, which contaminate the measurement. For the standard AERI instruments the signal-to-noise level is also degraded at wavenumbers below about 550 cm−1 and above about 2500 cm−1. A wavenumber region in each detector spectral band was chosen to provide an estimate of the error (measured minus predicted) for each calibration verification test. A summary table containing the mean and 1σ uncertainty in the mean for each instrument test is provided in Table 4. One of the AERI systems (AERI-06) was tested in 1997 and again in 2001; however, the system was completely recalibrated for the test in 2001 with a new detector and new blackbodies, so the two tests are independent from the point of view of radiometric calibration. Variations in the laboratory test results from instrument to instrument provide a measure of the variability in the absolute calibration of the AERI instruments. Table 4 shows the mean and 3σ errors for all the calibration verification tests of the “standard” AERI instruments (AERI-02, -03, -04, -05, -06, -08). The measured 3σ errors are compared against a model prediction of the root-mean-square of absolute calibration errors based upon an uncertainty analysis of the calibration equation. This analysis shows that the measured errors are close to the predicted uncertainties at the intermediate body temperature and within the expected error at the ice body temperature. Note that the predicted longwave uncertainty shown in Table 4 is slightly underestimated because it does not include the small contribution due to the uncertainty in the nonlinearity correction. These measurements at the intermediate and ice blackbody temperatures verify the model used to predict the AERI calibration uncertainties at colder scene temperatures. This is further confirmed by the sky intercomparison data presented in the next section.
(iv) Clear-sky radiometric calibration verification
As part of the calibration verification of each AERI instrument prior to initial deployment, a clear-sky intercomparison was performed at UW-SSEC against the AERI prototype instrument (AERI-00). This test was used to verify the radiometric calibration at the cold scene temperatures in the atmospheric window region by comparison to a common reference standard. Each AERI instrument is designed to measure absolute radiance to within 1% of the true ambient blackbody radiance, so the difference of any two instruments should be zero to within the combined uncertainties. Figure 7 shows the intercomparison of the AERI-04 (Hillsboro) instrument with the AERI prototype on 7 December 1998. The difference between the AERI-04 and AERI-00 spectra is actually much better than 1% across the longwave spectral band with one notable exception. The AERI prototype was operating from an enclosure that was warmer than the outside air. This leads to a mismatch with the AERI-04 (which was operating outside) in the most opaque CO2 and H2O lines, which are sensitive to the air temperature in the first meter of atmosphere above the instrument.
The results from all of the “pre-ship” sky intercomparison tests are summarized in Table 5 using narrow window regions near the center of each detector band. The results are presented as a percent of the ambient blackbody radiance of each instrument in order to simplify comparison to expected level of agreement. The largest percentage difference is the longwave AERI-02 minus AERI-00 value of −0.96% (subsequent to this analysis a software calibration parameter was found to be in error for the AERI-02 unit). The mean difference for all of the cases relative to the AERI prototype is about −0.3% in both the longwave and the shortwave bands. Now one can take advantage of the fact that the seven independently calibrated AERI instruments were all compared to the same AERI prototype instrument under similar sky conditions. Assuming that the uncertainties in the absolute calibration of each AERI system vary randomly about the true value, one can interpret the mean error of the set of standard AERI instruments relative to the AERI prototype as an estimate of the absolute calibration error of the AERI prototype instrument. Under this assumption, an estimate of the absolute error of each AERI system at the stated scene temperature can be obtained by subtracting the mean AERI prototype offset from each row of Table 5. The result of including this offset is shown in Fig. 8, which combines the results of the clear-sky intercomparisons in Table 5 with the laboratory calibration verification results of Table 4. Figure 8 suggests that each of the AERI systems built for the ARM Program meet the radiometric calibration specification of 1% of ambient radiance, although the sky intercomparisons cannot preclude an overall systematic error in all instruments. Figure 8 also includes the calibration uncertainty prediction from Fig. 4, assuming an ambient blackbody of 300 K. This prediction does not include the small uncertainty in the longwave band induced by the nonlinearity correction. These results confirm the basic AERI calibration methodology of using high-precision cavity references at hot and ambient temperatures to accurately extrapolate to cold-sky scene temperatures.
3) Spectral coverage and instrument line shape
The standard AERI radiance data product is a continuous spectrum between 520 and 3020 cm−1 (the requirement is 550–3000 cm−1). The extended-range AERI (ER-AERI) radiance product at the North Slope of Alaska (NSA) site is a continuous spectrum between 380 and 3020 cm−1 (the requirement is 400–3000 cm−1). Since the AERI instrument is a Fourier transform spectrometer, the “unapodized” spectral resolution is given by Δυ = 1/(2 × X), where X is the maximum optical path difference (OPD) of the Bomem interferometer. The maximum OPD is defined by the effective sampling frequency of the interferometer laser sampling system and the number of points collected per interferogram. The AERI radiance data product is minimally sampled; that is, the spectral sample frequency is equal to the unapodized spectral resolution.
The real-time AERI radiance product contains a correction for the small effects of instrument self-apodization on the instrument line shape. The correction makes use of the knowledge of the field-of-view (FOV) half-angle to remove the effect of instrument self-apodization in the measured spectrum and create a product that represents an “ideal” sinc function instrument line shape. Since the field angles are small, the correction can be made quite accurately. The adjustment of the measured spectrum to that of an “ideal” Michelson interferometer on a standard wavenumber grid greatly simplifies the comparison of an AERI observation to radiative transfer calculations or to observations from other coincident AERI instruments. The AERI “finite field of view” correction is described here, and the resampling of the spectrum to a standard wavenumber grid is presented in the next section on spectral calibration.
4) Spectral calibration
The approach to wavenumber calibration of the AERI instruments is to take advantage of gaseous line center positions known to high accuracy through laboratory measurements (Rothman et al. 1992). A line-by-line radiative transfer model (LBLRTM) is used to calculate a downwelling atmospheric emission spectrum using a radiosonde profile of temperature and water vapor coincident with an AERI observation. The effective sampling frequency υeff is determined empirically by minimizing the standard deviation between observed and calculated emission spectra as the effective sampling frequency of the observation is varied. This minimization is illustrated in Fig. 10 for the regularly spaced CO2 lines in the wavenumber range 730–740 cm−1. A similar analysis is performed in the AERI shortwave band using the regularly spaced N2O lines between 2207 and 2220 cm−1. Prior to the initial deployment of each AERI system, a clear-sky observation of downwelling radiance was recorded coincident with a radiosonde launched from UW-SSEC. The LBLRTM was used with a version of the HITRAN database to calculate the downwelling emission at the surface (Clough and Iacono 1995; Rothman et al. 1992). Uncertainties in the atmospheric water vapor and temperature profiles cause the minimum in the standard deviation shown in Fig. 10 to be nonzero; however, this introduces only a small error in the determination of the effective sampling frequency. The AERI spectral calibration requirement is stated in Part I as “better than 0.01 cm−1” over the entire spectral range. At 3020 cm−1, the 0.01-cm−1 requirement translates into a knowledge of Δυ, and hence υeff, of 3.3 ppm (or better).
A detailed analysis was performed to quantify the uncertainty in this spectral calibration technique and to assess the long-term stability of the AERI spectral calibration. A fit to the effective sampling frequency of the AERI-01 system was performed using 241 cases of clear-sky AERI observations coincident with radiosonde launches at the DOE ARM SGP Central Facility. The data span the period from 11 November 1998 to 30 September 2001 (35 months). The calculations were performed using LBLRTM v6.01 with HITRAN2000. The radiosonde (Vaisala RS80-H) water vapor profiles were scaled to agree with the total precipitable water column measured by a coincident microwave radiometer (standard ARM processing). The details of this set of clear-sky observations is described in Turner et al. (2004). In February 2000, the HeNe laser in the AERI-01 system was replaced, presumably changing the relative alignment of the laser and the infrared beam. The upper panel of Fig. 11 clearly shows the abrupt change in the longwave effective sampling frequency caused by the laser replacement in what otherwise is a very stable spectral calibration. The data fall into two groups: 95 cases before January 2000 and 146 cases after the laser replacement in February 2000. A statistical analysis of the AERI-01 longwave spectral calibration has been performed on the two sets. The AERI-01 longwave band effective laser sampling frequency determined before initial deployment of the AERI-01 system (in 1995) was 15 798.80 cm−1, with an estimated uncertainty of about 0.05 cm−1. The refined analysis using the 95 coincident radiosonde cases at the ARM SGP Central Facility prior to January 2000 gives a mean value of 15 798.74 cm−1, with a 1σ standard deviation of 0.025 cm−1, that is, 1.6 ppm. The analysis using the 145 cases between March 2000 and September 2001 gives a new mean value for the period after the laser change of 15 799.40 cm−1, with a 1σ value of 0.022 cm−1, that is, 1.4 ppm. A similar analysis has been performed on the shortwave AERI spectral band using the wavenumber region 2207– 2220 cm−1. The AERI-01 shortwave band effective laser sampling frequency determined before initial deployment of the AERI-01 system (in 1995) was 15 798.62 cm−1, with an estimated uncertainty of about 0.05 cm−1. The analysis of the shortwave band prior to January 2000 gives the same mean value of 15 798.62 cm−1 but with a 1σ standard deviation of 0.015 cm−1 out of 15 799 cm−1, that is, 0.95 ppm. After January 2000, the shortwave mean value was determined to be 15 799.21 cm−1, with a 1σ standard deviation of 0.021 cm−1, that is, 1.3 ppm, after the laser replacement. In summary, the wavenumber knowledge determined from a single AERI/radiosonde comparison during the initial instrument testing before deployment should be accurate to within about 3 ppm (2σ), with 95% confidence, which meets the AERI specification. However, this analysis shows that the uncertainty in the wavenumber scale of each AERI system can be further reduced (by at least an order of magnitude) by careful comparison with a large set of coincident clear-sky radiative transfer calculations, as was demonstrated for the AERI-01 system.
Once the spectral calibration is known, the AERI radiance spectrum can be resampled from the “original” sampling interval to a standard “reference” wavenumber scale. The reference wavenumber scale for all of the AERI instruments was chosen to correspond to an effective laser sampling frequency of 15 799.0 (exact). The resampling is performed in software using an FFT, “zero padding,” and linear interpolation of an oversampled spectrum. This procedure is numerically intensive but can be performed without loss of accuracy. The advantage of providing the AERI radiance product on a standard wavenumber scale is to simplify comparison with radiative transfer model calculations and with other AERI instruments.
5) Noise
The AERI requirement on radiometric noise performance is stated as a standard deviation of observed radiance during a 2-min dwell of a hot blackbody (+60°C) over a specified wavenumber range. A separate specification is used for the longwave band of the ER-AERI. The horizontal lines in Fig. 12 show the AERI noise specification found in Part I. Note that while the ER-AERI achieves enhanced noise performance out to 425 cm−1, the noise performance from 600 to 1400 cm−1 is degraded relative to the standard AERI detectors. For this reason, a standard AERI system is preferred over an ER-AERI system for all but the driest atmospheres when the 380–500-cm−1 rotational water vapor band becomes important.
The real-time AERI software also produces an estimate of the noise on the final calibrated scene. This estimate makes use of the imaginary part of the calibration equation given in Eq. (5). A 25-cm−1 wavenumber standard deviation is performed on the imaginary part corresponding to each scene. The “forward” Michelson mirror scans are used to make the estimate so the result is divided by
6) Reproducibility
The short-term reproducibility of the AERI observations is illustrated in Fig. 14 with a time series of observations of an external UW-SSEC blackbody at 318 K for the AERI-03 (Vici) instrument. The test was the same setup shown in Fig. 5 and summarized in Table 4. In order to study the time variation of the calibration, the random noise was reduced by a factor of 20 using spectral averages in three 200-cm−1 regions of the longwave spectrum. The peak-to-peak variation of observed brightness temperature is less than ±5 mK relative to the mean over the 4-h period, which corresponds to a peak-to-peak radiance variation of less than 0.01% at 1000 cm−1. This exceptional stability is a result of the long time constant of the AERI blackbodies and the excellent blackbody temperature control and readout precision. The lower panel of Fig. 14 shows that the short-term temperature control of the AERI blackbodies is better than ±1 mK (peak to peak) relative to the mean over the test period. The AERI short-term reproducibility is well within the ARM requirement of 0.2%.
3. Conclusions
The performance of the AERI instruments designed and built at UW-SSEC meet the ARM Program requirements for downwelling infrared spectral observations at the surface. The AERI instruments built for the ARM Program have demonstrated radiometric accuracy of better than 1% of ambient radiance, with a reproducibility of better than 0.2%. A routine correction for nonlinearity of the longwave HgCdTe detector is applied to the observed data in real time. A small correction for instrument line shape is also applied to create an idealized “sinc” line shape function, and the data are resampled onto a standard wavenumber grid for convenience in comparison to model calculations. The spectral calibration is known to better than 1.5 ppm (1σ) using known spectral positions of atmospheric lines. A comprehensive error analysis of the AERI observations for tropical, midlatitude, and Arctic environments is the subject of a future paper.
Acknowledgments
This research was supported by the Office of Science (BER), U.S. Department of Energy, Grants DE-FG02-90ER61057 and DE-FG02-92ER61365. Special thanks go to Dave Turner of Pacific Northwest National Laboratory for his encouragement during the writing of this paper.
REFERENCES
Clough, S. A., and Iacono M. J. , 1995: Line-by-line calculations of atmospheric fluxes and cooling rates. 2: Applications to carbon dioxide, ozone, methane, nitrous oxide and the halocarbons. J. Geophys. Res, 100 , 16519–16535.
Knuteson, R. O., and Coauthors, 2004: Atmospheric Emitted Radiance Interferometer. Part I: Instrument design. J. Atmos. Oceanic Technol, 21 , 1763–1776.
Minnett, P. J., Knuteson R. O. , Best F. A. , Osborne B. J. , Hanafin J. A. , and Brown O. B. , 2001: The Marine-Atmospheric Emitted Radiance Interferometer (M-AERI), a high-accuracy, sea-going infrared spectroradiometer. J. Atmos. Oceanic Technol, 18 , 994–1013.
Revercomb, H. E., Buijs H. , Howell H. B. , LaPorte D. D. , Smith W. L. , and Sromovsky L. A. , 1988: Radiometric calibration of IR Fourier transform spectrometers: Solution to a problem with the High-Resolution Interferometer Sounder. Appl. Opt, 27 , 3210–3218.
Rothman, L. S., and Coauthors, 1992: The HITRAN molecular database: Editions of 1991 and 1992. J. Quant. Spectrosc. Radiat. Transfer, 48 , 469–507.
Sromovsky, L. A., 2003: Radiometric errors in complex Fourier transform spectrometry. Appl. Opt, 42 , 1779–1787.
Stokes, G. M., and Schwartz S. E. , 1994: The Atmospheric Radiation Measurement (ARM) Program: Programmatic background and design of the Cloud and Radiation Testbed. Bull. Amer. Meteor. Soc, 75 , 1201–1221.
Turner, D. D., and Coauthors, 2004: The QME AERI LBLRTM: A closure experiment for downwelling high spectral resolution infrared radiance. J. Atmos. Sci, 61 , 2657–2675.
The nonlinearity model fit (gray curve) to a measured responsivity difference between hot/ambient and ambient/LN2 reference targets. The fit region is 200–460 cm−1, which avoids the “in-band” uncertainties of the LN2 target emissivity due to a liquid water cloud that forms over the cold target. Note that the quadratic nonlinearity model agrees in both the 200–460- and 1800–2000-cm−1 out-of-band regions even though the least squares fit only uses the first region to determine a2. The test data are of AERI-05 on 8 Dec 1998 but are typical of all standard AERI systems. Units are instrument counts per radiance unit {counts per [mW (m2 sr cm−1)−1]}
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The nonlinearity model fit (gray curve) to a measured responsivity difference between hot/ambient and ambient/LN2 reference targets. The fit region is 200–460 cm−1, which avoids the “in-band” uncertainties of the LN2 target emissivity due to a liquid water cloud that forms over the cold target. Note that the quadratic nonlinearity model agrees in both the 200–460- and 1800–2000-cm−1 out-of-band regions even though the least squares fit only uses the first region to determine a2. The test data are of AERI-05 on 8 Dec 1998 but are typical of all standard AERI systems. Units are instrument counts per radiance unit {counts per [mW (m2 sr cm−1)−1]}
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The nonlinearity model fit (gray curve) to a measured responsivity difference between hot/ambient and ambient/LN2 reference targets. The fit region is 200–460 cm−1, which avoids the “in-band” uncertainties of the LN2 target emissivity due to a liquid water cloud that forms over the cold target. Note that the quadratic nonlinearity model agrees in both the 200–460- and 1800–2000-cm−1 out-of-band regions even though the least squares fit only uses the first region to determine a2. The test data are of AERI-05 on 8 Dec 1998 but are typical of all standard AERI systems. Units are instrument counts per radiance unit {counts per [mW (m2 sr cm−1)−1]}
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
This example shows the magnitude of the nonlinearity correction for calibrated radiances on a typical clear-sky observation. The observation was made using AERI-05 at UW–Madison on 7 Dec 1998. (top) An overlay of the calibrated longwave spectrum and a Planck function at the ambient blackbody temperature (290 K). (bottom) A radiance difference of the calibrated spectrum with and without a nonlinearity correction. The solid lines indicate ±1% of the ambient blackbody radiance. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
This example shows the magnitude of the nonlinearity correction for calibrated radiances on a typical clear-sky observation. The observation was made using AERI-05 at UW–Madison on 7 Dec 1998. (top) An overlay of the calibrated longwave spectrum and a Planck function at the ambient blackbody temperature (290 K). (bottom) A radiance difference of the calibrated spectrum with and without a nonlinearity correction. The solid lines indicate ±1% of the ambient blackbody radiance. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
This example shows the magnitude of the nonlinearity correction for calibrated radiances on a typical clear-sky observation. The observation was made using AERI-05 at UW–Madison on 7 Dec 1998. (top) An overlay of the calibrated longwave spectrum and a Planck function at the ambient blackbody temperature (290 K). (bottom) A radiance difference of the calibrated spectrum with and without a nonlinearity correction. The solid lines indicate ±1% of the ambient blackbody radiance. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The (top) responsivity magnitude and (bottom) calibrated radiance spectra for a calibration sequence of AERI-01 from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. The smooth curves in the lower panel correspond to Planck functions at the hot and ambient blackbody temperatures. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The (top) responsivity magnitude and (bottom) calibrated radiance spectra for a calibration sequence of AERI-01 from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. The smooth curves in the lower panel correspond to Planck functions at the hot and ambient blackbody temperatures. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The (top) responsivity magnitude and (bottom) calibrated radiance spectra for a calibration sequence of AERI-01 from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. The smooth curves in the lower panel correspond to Planck functions at the hot and ambient blackbody temperatures. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The predicted 3σ calibration uncertainty for the standard AERI system at 770 cm−1 for an ambient blackbody at 300 K and assuming the blackbody uncertainties from Table 3. Separate error estimates are shown for the contribution from the hot blackbody (Th), the ambient blackbody (Ta), the hot blackbody emissivity (Eh), the ambient blackbody emissivity (Ea), and the environment surrounding the blackbodies (Tr). The solid curve is the root sum square (RSS) of the individual contributions
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The predicted 3σ calibration uncertainty for the standard AERI system at 770 cm−1 for an ambient blackbody at 300 K and assuming the blackbody uncertainties from Table 3. Separate error estimates are shown for the contribution from the hot blackbody (Th), the ambient blackbody (Ta), the hot blackbody emissivity (Eh), the ambient blackbody emissivity (Ea), and the environment surrounding the blackbodies (Tr). The solid curve is the root sum square (RSS) of the individual contributions
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The predicted 3σ calibration uncertainty for the standard AERI system at 770 cm−1 for an ambient blackbody at 300 K and assuming the blackbody uncertainties from Table 3. Separate error estimates are shown for the contribution from the hot blackbody (Th), the ambient blackbody (Ta), the hot blackbody emissivity (Eh), the ambient blackbody emissivity (Ea), and the environment surrounding the blackbodies (Tr). The solid curve is the root sum square (RSS) of the individual contributions
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Typical UW-SSEC laboratory end-to-end calibration verification using reference targets at about 318 (zenith view) and 273.15 K (nadir view). This “four body” test was used to verify the radiometric calibration of the AERI instruments prior to deployment of the systems to the DOE ARM sites
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Typical UW-SSEC laboratory end-to-end calibration verification using reference targets at about 318 (zenith view) and 273.15 K (nadir view). This “four body” test was used to verify the radiometric calibration of the AERI instruments prior to deployment of the systems to the DOE ARM sites
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Typical UW-SSEC laboratory end-to-end calibration verification using reference targets at about 318 (zenith view) and 273.15 K (nadir view). This “four body” test was used to verify the radiometric calibration of the AERI instruments prior to deployment of the systems to the DOE ARM sites
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Laboratory radiometric calibration verification test results for the AERI-06 system conducted at UW-SSEC on 8 May 2001 before deployment to the TWP-Nauru site. (top) The result for the ice blackbody. (bottom) The result for an external blackbody at a temperature “intermediate” between the AERI ambient and hot blackbodies. The smooth line is the predicted radiometric temperature based upon the measured blackbody temperatures and the assumed cavity emissivity (cavity factor of 12.79). The measured spectrum is the mean calibrated radiance for the 3.2-h test period converted to equivalent blackbody (brightness) temperature. Strongly absorbing CO2 and H2O lines in the air path between the detector and the blackbody reference sources contaminate the measurement between 1400–1900 and 2300–2400 cm−1
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Laboratory radiometric calibration verification test results for the AERI-06 system conducted at UW-SSEC on 8 May 2001 before deployment to the TWP-Nauru site. (top) The result for the ice blackbody. (bottom) The result for an external blackbody at a temperature “intermediate” between the AERI ambient and hot blackbodies. The smooth line is the predicted radiometric temperature based upon the measured blackbody temperatures and the assumed cavity emissivity (cavity factor of 12.79). The measured spectrum is the mean calibrated radiance for the 3.2-h test period converted to equivalent blackbody (brightness) temperature. Strongly absorbing CO2 and H2O lines in the air path between the detector and the blackbody reference sources contaminate the measurement between 1400–1900 and 2300–2400 cm−1
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Laboratory radiometric calibration verification test results for the AERI-06 system conducted at UW-SSEC on 8 May 2001 before deployment to the TWP-Nauru site. (top) The result for the ice blackbody. (bottom) The result for an external blackbody at a temperature “intermediate” between the AERI ambient and hot blackbodies. The smooth line is the predicted radiometric temperature based upon the measured blackbody temperatures and the assumed cavity emissivity (cavity factor of 12.79). The measured spectrum is the mean calibrated radiance for the 3.2-h test period converted to equivalent blackbody (brightness) temperature. Strongly absorbing CO2 and H2O lines in the air path between the detector and the blackbody reference sources contaminate the measurement between 1400–1900 and 2300–2400 cm−1
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Coincident clear-sky comparison between AERI-04 (Hillsboro) and the AERI prototype (-00) at UW-SSEC on 7 Dec 1998 at Madison, WI. (top) The overlay of the AERI-04 observed downwelling radiance spectra and a Planck function at the AERI-04 ambient blackbody temperature. (bottom) The radiance difference between AERI-04 and AERI-00 averaged over the 1-h period 1624– 1724 UTC. For reference, ±1% of the AERI-04 ambient blackbody radiance is represented by the solid lines. [Radiance units: RU = mW (m2 sr cm−1)−1.] See the text for an explanation of the differences in the 660–680-cm−1 region
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Coincident clear-sky comparison between AERI-04 (Hillsboro) and the AERI prototype (-00) at UW-SSEC on 7 Dec 1998 at Madison, WI. (top) The overlay of the AERI-04 observed downwelling radiance spectra and a Planck function at the AERI-04 ambient blackbody temperature. (bottom) The radiance difference between AERI-04 and AERI-00 averaged over the 1-h period 1624– 1724 UTC. For reference, ±1% of the AERI-04 ambient blackbody radiance is represented by the solid lines. [Radiance units: RU = mW (m2 sr cm−1)−1.] See the text for an explanation of the differences in the 660–680-cm−1 region
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Coincident clear-sky comparison between AERI-04 (Hillsboro) and the AERI prototype (-00) at UW-SSEC on 7 Dec 1998 at Madison, WI. (top) The overlay of the AERI-04 observed downwelling radiance spectra and a Planck function at the AERI-04 ambient blackbody temperature. (bottom) The radiance difference between AERI-04 and AERI-00 averaged over the 1-h period 1624– 1724 UTC. For reference, ±1% of the AERI-04 ambient blackbody radiance is represented by the solid lines. [Radiance units: RU = mW (m2 sr cm−1)−1.] See the text for an explanation of the differences in the 660–680-cm−1 region
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Verification of the absolute radiometric accuracy of the AERI instruments built by UW-SSEC for the ARM Program showing that each system meets the specification of 1% of ambient radiance. Results are presented for (top) the longwave HgCdTe band near 10 μm and (bottom) the shortwave InSb band near 4 μm. The data points near 318 and 273 K are from the laboratory calibration verification tests using reference blackbodies. The data points at the cold scene temperatures are from clear-sky radiance intercomparisons with the AERI prototype system after the mean AERI prototype bias has been removed. Also shown (dashed curves) is the predicted error (3σ) contribution due to uncertainty in the blackbody parameters only, assuming a 300-K blackbody. The square, circle, diamond, star, downward triangle, and upward triangle represent the AERI-02, -03, -04, -05, -06, and -08 systems, respectively
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Verification of the absolute radiometric accuracy of the AERI instruments built by UW-SSEC for the ARM Program showing that each system meets the specification of 1% of ambient radiance. Results are presented for (top) the longwave HgCdTe band near 10 μm and (bottom) the shortwave InSb band near 4 μm. The data points near 318 and 273 K are from the laboratory calibration verification tests using reference blackbodies. The data points at the cold scene temperatures are from clear-sky radiance intercomparisons with the AERI prototype system after the mean AERI prototype bias has been removed. Also shown (dashed curves) is the predicted error (3σ) contribution due to uncertainty in the blackbody parameters only, assuming a 300-K blackbody. The square, circle, diamond, star, downward triangle, and upward triangle represent the AERI-02, -03, -04, -05, -06, and -08 systems, respectively
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Verification of the absolute radiometric accuracy of the AERI instruments built by UW-SSEC for the ARM Program showing that each system meets the specification of 1% of ambient radiance. Results are presented for (top) the longwave HgCdTe band near 10 μm and (bottom) the shortwave InSb band near 4 μm. The data points near 318 and 273 K are from the laboratory calibration verification tests using reference blackbodies. The data points at the cold scene temperatures are from clear-sky radiance intercomparisons with the AERI prototype system after the mean AERI prototype bias has been removed. Also shown (dashed curves) is the predicted error (3σ) contribution due to uncertainty in the blackbody parameters only, assuming a 300-K blackbody. The square, circle, diamond, star, downward triangle, and upward triangle represent the AERI-02, -03, -04, -05, -06, and -08 systems, respectively
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(bottom) Finite-FOV correction for (top) an AERI-01 radiance observation from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(bottom) Finite-FOV correction for (top) an AERI-01 radiance observation from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(bottom) Finite-FOV correction for (top) an AERI-01 radiance observation from the ARM SGP CF at 0146:21 UTC 18 Sep 2000. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The AERI longwave effective sampling frequency, υeff, is determined by the wavenumber scale factor that minimizes the std dev of the difference between observation and calculation for the wavenumber region 730–740 cm−1. The gamma factor is the ratio of the adjusted wavenumber scale to a reference wavenumber scale. Example is from the AERI-01 at the ARM SGP CF at 1120 UTC 30 Sep 2001. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The AERI longwave effective sampling frequency, υeff, is determined by the wavenumber scale factor that minimizes the std dev of the difference between observation and calculation for the wavenumber region 730–740 cm−1. The gamma factor is the ratio of the adjusted wavenumber scale to a reference wavenumber scale. Example is from the AERI-01 at the ARM SGP CF at 1120 UTC 30 Sep 2001. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The AERI longwave effective sampling frequency, υeff, is determined by the wavenumber scale factor that minimizes the std dev of the difference between observation and calculation for the wavenumber region 730–740 cm−1. The gamma factor is the ratio of the adjusted wavenumber scale to a reference wavenumber scale. Example is from the AERI-01 at the ARM SGP CF at 1120 UTC 30 Sep 2001. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) The time series of υeff − υref for each of the 142 cases of AERI-01 longwave observations and clear-sky calculations based upon microwave-scaled radisondes launched from the ARM SGP CF. The wavenumber reference for this figure (15 798.74 cm−1) is the mean of the 95 cases prior to Feb 2000. The abrupt change after Feb 2000 was due to the replacement of the instrument laser. (bottom) The same data plotted as a histogram. These results demonstrate an ability to determine the AERI spectral calibration to an accuracy of 1.5 ppm (1σ) using atmospheric observations
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) The time series of υeff − υref for each of the 142 cases of AERI-01 longwave observations and clear-sky calculations based upon microwave-scaled radisondes launched from the ARM SGP CF. The wavenumber reference for this figure (15 798.74 cm−1) is the mean of the 95 cases prior to Feb 2000. The abrupt change after Feb 2000 was due to the replacement of the instrument laser. (bottom) The same data plotted as a histogram. These results demonstrate an ability to determine the AERI spectral calibration to an accuracy of 1.5 ppm (1σ) using atmospheric observations
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) The time series of υeff − υref for each of the 142 cases of AERI-01 longwave observations and clear-sky calculations based upon microwave-scaled radisondes launched from the ARM SGP CF. The wavenumber reference for this figure (15 798.74 cm−1) is the mean of the 95 cases prior to Feb 2000. The abrupt change after Feb 2000 was due to the replacement of the instrument laser. (bottom) The same data plotted as a histogram. These results demonstrate an ability to determine the AERI spectral calibration to an accuracy of 1.5 ppm (1σ) using atmospheric observations
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The noise specification on the hot blackbody reference view for the longwave band of (top) an ER-AERI, (middle) the longwave band of a standard AERI, and (bottom) the shortwave band of any AERI, shown with horizontal solid lines. The two curves (solid and dotted) are the hot blackbody noise estimate from the two real-time AERI noise products. The two noise estimates agree so well that the curves fall directly on top of each other. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The noise specification on the hot blackbody reference view for the longwave band of (top) an ER-AERI, (middle) the longwave band of a standard AERI, and (bottom) the shortwave band of any AERI, shown with horizontal solid lines. The two curves (solid and dotted) are the hot blackbody noise estimate from the two real-time AERI noise products. The two noise estimates agree so well that the curves fall directly on top of each other. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
The noise specification on the hot blackbody reference view for the longwave band of (top) an ER-AERI, (middle) the longwave band of a standard AERI, and (bottom) the shortwave band of any AERI, shown with horizontal solid lines. The two curves (solid and dotted) are the hot blackbody noise estimate from the two real-time AERI noise products. The two noise estimates agree so well that the curves fall directly on top of each other. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Noise on three calibrated clear-sky scenes observed by the AERI. The curves show the real-time sky noise product from the AERI-01 system at the SGP CF on 18 Sep 2000, the AERI-06 instrument at the TWP-Nauru site on 15 Nov 1998, and the ER-AERI (-07) at the NSA-Barrow site on 10 Mar 1999. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Noise on three calibrated clear-sky scenes observed by the AERI. The curves show the real-time sky noise product from the AERI-01 system at the SGP CF on 18 Sep 2000, the AERI-06 instrument at the TWP-Nauru site on 15 Nov 1998, and the ER-AERI (-07) at the NSA-Barrow site on 10 Mar 1999. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Noise on three calibrated clear-sky scenes observed by the AERI. The curves show the real-time sky noise product from the AERI-01 system at the SGP CF on 18 Sep 2000, the AERI-06 instrument at the TWP-Nauru site on 15 Nov 1998, and the ER-AERI (-07) at the NSA-Barrow site on 10 Mar 1999. [Radiance units: RU = mW (m2 sr cm−1)−1]
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) Short-term calibration reproducibility of better than 5 mK and (bottom) temperature control better than 1 mK, illustrated using a blackbody target at 318K. Data were collected from the AERI-03 (Vici) instrument at UW-SSEC on 12 Nov 1998. This performance exceeds the ARM requirement for calibration reproducibility by an order of magnitude
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) Short-term calibration reproducibility of better than 5 mK and (bottom) temperature control better than 1 mK, illustrated using a blackbody target at 318K. Data were collected from the AERI-03 (Vici) instrument at UW-SSEC on 12 Nov 1998. This performance exceeds the ARM requirement for calibration reproducibility by an order of magnitude
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
(top) Short-term calibration reproducibility of better than 5 mK and (bottom) temperature control better than 1 mK, illustrated using a blackbody target at 318K. Data were collected from the AERI-03 (Vici) instrument at UW-SSEC on 12 Nov 1998. This performance exceeds the ARM requirement for calibration reproducibility by an order of magnitude
Citation: Journal of Atmospheric and Oceanic Technology 21, 12; 10.1175/JTECH-1663.1
Estimate of the “in-band” nonlinearity correction for each of the AERI longwave detectors as a percent of the raw signal (100 × 2a2Vdc ) for the hot (333 K), ambient (300 K), and LN2 (77 K) blackbodies. The extended-range AERI detectors for the NSA site do not exhibit a measurable nonlinearity, and no nonlinearity cor rection is applied. The systems labeled “MAERI” are the three Ma rine-AERI systems built by UW-SSEC for the University of Miami, described in Minnett et al. (2001)
Scene mirror sequence for the AERI-01 system
Parameters used in the AERI calibration uncertainty analysis
Summary of AERI laboratory calibration verification results (mK). The mean error and the 1σ uncertainty in the mean is listed for each verification test. The sample mean and sample std dev is computed for the set of seven independent instrument calibration verification tests. The variance among the tests is compared with the 3σ predicted error and the AERI calibration specification (1% of ambient radiance) at the scene temperatures and measured wavelengths
Summary of the clear-sky comparison of each AERI system to the AERI prototype at UW-SSEC prior to initial system deployment
