Downward longwave irradiance uncertainty under

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. D12, 4358, doi:10.1029/2002JD002937, 2003

Downward longwave irradiance uncertainty under arctic atmospheres: Measurements and modeling Christoph Marty,1,2 Rolf Philipona,2 Jennifer Delamere,3 Ellsworth G. Dutton,4 Joe Michalsky,5 Knut Stamnes,6 Rune Storvold,1 Tom Stoffel,7 Shepard A. Clough,3 and Eli J. Mlawer3 Received 12 September 2002; revised 27 February 2003; accepted 18 March 2003; published 21 June 2003.

[1] Measurement and modeling of downward longwave irradiance are a special challenge

in arctic winter due to its low water vapor content and the extreme meteorological conditions. There are questions about the representativeness of the instrument calibration, the consistency and uncertainty of measurements and models in these environments. The Second International Pyrgeometer and Absolute Sky-scanning Radiometer Comparison (IPASRC-II), which was conducted at Atmospheric Radiation Measurement (ARM) program’s North Slope of Alaska (NSA) site in Barrow provided a unique opportunity to compare high accuracy downward longwave irradiance measurements and radiative transfer model computations during arctic winter. Participants from 11 international institutions deployed 14 pyrgeometers, which were field-calibrated against the Absolute Sky-scanning Radiometer (ASR). Continuous measurements over a 10-day period in early March 2001 with frequent clear-sky conditions yielded downward longwave irradiances between 120 and 240 W m2. The small average difference between ASR irradiances, pyrgeometer measurements, MODTRAN and LBLRTM radiative transfer computations indicates that the absolute uncertainty of measured downward longwave irradiance under INDEX TERMS: 0360 Atmospheric Composition arctic winter conditions is within ±2 W m2. and Structure: Transmission and scattering of radiation; 1610 Global Change: Atmosphere (0315, 0325); 3309 Meteorology and Atmospheric Dynamics: Climatology (1620); 3349 Meteorology and Atmospheric Dynamics: Polar meteorology; KEYWORDS: longwave radiation, pyrgeometer, MODTRAN, LBLRTM Citation: Marty, C., R. Philipona, J. Delamere, E. G. Dutton, J. Michalsky, K. Stamnes, R. Storvold, T. Stoffel, S. A. Clough, and E. J. Mlawer, Downward longwave irradiance uncertainty under arctic atmospheres: Measurements and modeling, J. Geophys. Res., 108(D12), 4358, doi:10.1029/2002JD002937, 2003.

1. Introduction [2] Downward longwave radiation at the surface is a key indicator of an enhanced greenhouse effect related to changes in cloud cover or increased temperature and humidity of the lower atmosphere. Accurate downward longwave radiation measurements allow the detection of clear-sky situations and thus the determination of the surface cloud forcing [Charlock and Ramanathan, 1985; Marty et al., 2002]. The fact that downward longwave radiation is 1

Geophysical Institute, University of Alaska Fairbanks, Fairbanks, Alaska, USA. 2 Physikalisch-Meteorologisches Observatorium and World Radiation Center, Davos Dorf, Switzerland. 3 Atmospheric and Environmental Research, Lexington, Massachusetts, USA. 4 Climate Monitoring and Diagnostics Laboratory, NOAA, Boulder, Colorado, USA. 5 Atmospheric Science Research Center, State University of New York at Albany, Albany, New York, USA. 6 Stevens Institute of Technology, Hoboken, New Jersey, USA. 7 National Renewable Energy Laboratory, Golden, Colorado, USA. Copyright 2003 by the American Geophysical Union. 0148-0227/03/2002JD002937$09.00

ACL

directly related to the greenhouse effect and its relatively small year-to-year variation makes it one of the most promising parameters to monitor greenhouse effect related changes of the atmosphere (A. Ohmura, personal communication). Accurate measurements and modeling of the downward longwave flux is therefore an important issue in climate research [Dutton, 1993; Wild et al., 2001]. Especially in polar regions, longwave radiation dominates the surface radiation budget during much of the year due to large solar zenith angles at high latitudes. This causes the annual total downward longwave radiation to be more than double the downward shortwave radiation [Curry et al., 1996]. [3] Atmospheric longwave irradiance is usually measured with hemispherical receivers on flat horizontal surfaces. The most prominent instruments to measure longwave irradiance in climatological networks are pyrgeometers. The most widely used pyrgeometer, the Precision Infrared Radiometer (PIR) from Eppley Laboratory has been improved over the last 25 years. Pyrgeometers have always been calibrated with blackbody radiation sources. Investigations of calibration methods, instrument characterization and measurement techniques have led to a deeper understanding and dramatic improvements of pyrgeometer calibration and measure-

4-1

ACL

4-2

MARTY ET AL.: LONGWAVE IRRADIANCE UNCERTAINTY

ments [Philipona et al., 1995; Fairall et al., 1998]. The Baseline Surface Radiation Network (BSRN) [Ohmura et al., 1998] round-robin calibration experiment [Philipona et al., 1998] demonstrates that PIRs are stable and that calibration results of individual blackbody calibration sources used worldwide are within 2% of each other. Uncertainties of longwave irradiance measurements dropped due to these efforts from about 10% in the early 1990s [Dehne et al., 1993] to less than 2% in the last few years. The remaining uncertainties of pyrgeometers, mainly related to the silicon dome, arise from thermal effects, inadequate spectral transmission and cosine response errors. New instruments like the CG4 pyrgeometer from Kipp&Zonen show promising advantages with respect to dome heating and spectral transmission. However, these advantages do not answer the question of the absolute error. Therefore, until recently it was unknown how well blackbody calibrated pyrgeometer measurements represented the absolute value of the atmospheric longwave irradiance. [4] The Absolute Sky-scanning Radiometer (ASR) was developed at the Physikalisch-Meteorologisches Observatorium Davos and World Radiation Center (PMOD/WRC) in co-operation with the Swiss Federal Institute of Technology (ETH) Zurich in 1999 to overcome this deficiency [Philipona, 2001]. The ASR measures radiance in a narrow viewing angle and scans the sky in order to be able to compute hemispherical downward longwave irradiance by integration of 32 measurement points. The calibration of the ASR is based on a reference blackbody source traced to absolute temperature standards. [5] The ASR was used to calibrate 15 pyrgeometers during the BSRN-initiated First International Pyrgeometer and Absolute Sky-scanning Radiometer Comparison (IPASRC-I) at the Atmospheric Radiation Measurement (ARM) program’s Southern Great Plains (SGP) site in September 1999 [Philipona et al., 2001]. Comparisons between these pyrgeometers, the Atmospheric Emitted Radiance Interferometer (AERI), radiative transfer models and the ASR showed astoundingly good agreement of less than 2 W m2 for nighttime measurements and calculations. Since IPASRC-I observed downward longwave irradiance in the range between 260 and 420 W m2 this result is valid for midlatitude summer conditions. [6] This paper describes efforts undertaken and results of the Second International Pyrgeometer and Absolute Skyscanning Radiometer Comparison (IPASRC-II). The main goal of IPASRC-II was to reveal the uncertainties of atmospheric longwave irradiance measurements in cold and dry atmospheres. IPASRC-II was held between March 5 and 15, 2001 at the ARM program’s North Slope of Alaska (NSA) Barrow site [Stamnes et al., 1999] to compare pyrgeometer and AERI data with ASR measurements and radiative transfer model computations.

2. Rationale [7] Looking at annual mean values, downward longwave radiation is the largest atmospheric component of the energy balance equation. Earth loses part of this energy to space in order to maintain its relatively pleasant temperatures. Water vapor, as Earth’s primary greenhouse gas, controls much of this cooling. Clough et al. [1992] and Sinha and Harries

[1995] show the importance of the water vapor band from 17 to 33 mm (300 to 600 cm1) to the Earth energy balance. For typical midlatitude conditions the 17 to 33 mm band is opaque due to the high concentration of atmospheric water vapor. Emission in this spectral region is therefore described by the Planck function at near-surface temperature. At high latitude or high altitude, however, it is often dry enough that regions between spectral lines in this strong absorption band become transparent and the associated downward irradiances are representative of colder temperatures. The transparency of these narrow ‘‘microwindows’’ in the region between 17 and 33 mm (cf. Figure 6) is a special challenge for atmospheric radiative transfer. All these microwindows act together like a secondary atmospheric ‘‘window,’’ sometimes called the ‘‘dirty window’’ and are therefore critical for climate and energy balance related issues. This is true not only for downward longwave fluxes at the surface of polar regions, but also for the top-of-atmosphere fluxes and cooling rates on a global scale [Sinha and Harries, 1995; Stamnes, 1996; Stamnes et al., 1999]. [8] Hence, the significantly different longwave spectra in the arctic make not only the measurement, but also the modeling of the downward longwave irradiance a special challenge. Although clear-sky longwave modeling is quite accurate for warmer climates there is still some uncertainty for cold and dry climates due to the above mentioned microwindows [Tobin et al., 1999]. A recent comparison of downward longwave irradiances between global climate models and the best available surface measurements shows an especially large disagreement at observation sites in cold and dry climates [Wild et al., 2001]. Improvements in accurate measurements and modeling of atmospheric longwave irradiance under cloudless skies in cold and dry conditions do not only benefit the high latitudes, but also the high altitudes, which both have a high sensitivity in regard to climate change.

3. Instruments and Models 3.1. Pyrgeometers [9] A total of 14 calibration-traced pyrgeometers, chosen from a broad international community (cf. Table 1), were compared during IPASRC-II. These pyrgeometers were subdivided into two groups. The first group consisted of seven original Eppley PIR pyrgeometers. Original PIRs are instruments as available from the manufacturer. They were always used in the ‘‘nonbattery’’ mode, which means that the body temperature is not compensated by the battery, but is measured separately. The second group consisted of five modified Eppley PIR pyrgeometers and two Kipp&Zonen CG4 pyrgeometers. The modified PIRs have three instead of only one dome thermistor in order to provide a more uniform temperature measurement [Philipona et al., 1995]. The CG4 pyrgeometers use a meniscus shaped dome, which is designed to minimize temperature effects within the dome by optimizing the thermal contact to the body of the instrument. They usually come without a dome thermistor, but the ones used for the comparison were specially modified with one dome thermistor. A separate analysis showed that the dome-induced correction term (cf. Section 4) was indeed very small for CG4’s, i.e. about 20 times smaller than for a PIR. For convenience from now on, we

4-3

ACL


Table 1. Participating Pyrgeometers (Owner, Number of Dome Thermistors, and Calibration Factors per Calibration Set) USERa

CMDLa

Serial No.

Ownerb

TD

C

k3

C

k3

PIR32054F3 PIR29255F3 PIR32227F3 PIR31195F3 PIR13678F3 PIR30555F3 PIR32046F3 PIR26036mod PIR28146mod PIR30475mod PIR32205mod PIR31463mod CG4-004 CG4-005

DOE/ARM, US NOAA/SRRB, US Eppley, US NREL, US BoM, AU NOAA/CMDL, US DOE/ARM, US NASA, US AES, CA DWD, DE JMA, JP PMOD/WRC, CH PMOD/WRC, CH PMOD/WRC, CH

1 1 1 1 1 1 1 3 3 3 3 3 1 1

4.20 4.00 3.90 3.56 4.29 3.57 3.75 3.84 3.69 3.63 3.92 3.32 12.29 8.75

4.0 4.0 4.0 4.0 4.0 4.0 4.0 3.48 3.42 3.13 2.66 3.53 0.30 1.00

4.20 4.01 3.88 3.47 3.92 3.45 3.73 3.72 3.65 3.63 3.81 3.40 12.17 8.93

3.3 3.0 3.5 3.5 4.5 4.0 3.5 4.5 4.0 3.5 3.5 4.0 1.0 1.0

PMOD C

4.36 4.34 3.95 4.25 3.71 12.29 8.75

k1

0.1218 0.1510 0.0733 0.1112 0.0766 0 0

FIELD k2

1.0038 1.0240 1.0014 1.0031 0.9974 1.0000 1.0000

k3

C

k1

k2

k3

3.58 3.61 3.11 2.75 3.39 0.30 1.00

4.11 3.98 3.85 3.41 4.06 3.51 3.80 4.28 4.32 3.96 4.21 3.76 12.42 8.94

0 0 0 0 0 0 0 0.1218 0.1510 0.0733 0.1112 0.0766 0 0

1.000 1.002 0.987 0.997 0.994 0.994 1.000 0.9960 1.0028 1.0000 1.0031 0.9850 1.0004 1.0000

3.3 3.0 3.5 3.5 4.5 4.0 3.5 3.58 3.61 3.11 2.75 3.39 0.30 1.00

a

Here k1 = 0 and k2 = 1. DOE/ARM, Department of Energy/Atmospheric Radiation Measurement Program; NOAA/SRRB, National Oceanic and Atmospheric Administration/ Surface Radiation Branch; NREL, National Renewable Energy Lab; BoM, Bureau of Meteorology; NOAA/CMDL, National Oceanic and Atmospheric Administration/Climate Monitoring and Diagnostics Laboratory; NASA, National Aeronautics Space Administration; AES, Atmospheric Environmental Service; DWD, Deutscher Wetter Dienst, JMA, Japanese Meteorological Agency; PMOD/WRC, Physikalisch-Meteorologisches Observatorium Davos/ World Radiation Center. b

will refer to this second group of seven pyrgeometers as ‘‘modified pyrgeometers’’ in contrast to the first group, which is called ‘‘original pyrgeometers.’’ All instruments were ventilated with slightly preheated air to prevent frost and rime on the domes. Each pyrgeometer was mounted on a solar tracker equipped with a shading ball, so that direct solar radiation could not interfere with the longwave measurement. 3.2. Absolute Sky-Scanning Radiometer [10] The Absolute Sky-scanning Radiometer (ASR), considered the reference instrument for the experiment, was installed on the same platform as the pyrgeometers. The ASR measures radiance in a full view angle of 6 and scans the sky at four elevation angles and eight azimuthal directions. Hemispherical longwave irradiance is computed by using Gaussian quadrature to integrate over 32 measuring points. The ASR uses no window or optical components with the exception of a 90 reflecting gold mirror, which directs the narrow field of view during a measurement alternately either to the sky or into the reference blackbody source for calibration. Hence, the ASR is still blackbody calibrated and related to absolute standards of internationally accepted metrological units via absolute temperature measurements. However, unlike pyrgeometers, it is not subject to spectral and directional dome transmission problems, which produce uncertainties in pyrgeometer measurements particularly with respect to absolute measurements. Since no window or filter is installed in the beam path to the pyroelectric detector, sky scans can only be made during nighttime to prevent shortwave disturbance. The combined standard uncertainty (root sum of squares) for ASR measurements has been calculated as less than ±1 W m2 [Philipona, 2001]. 3.3. Extended-Range Atmospheric Emitted Radiance Interferometer [11] The Extended-Range Atmospheric Emitted Radiance Interferometer (AERI-ER) permanently deployed at the ARM NSA site in Barrow was developed at the University

of Wisconsin with the support of the Department of Energy ARM Program. The AERI-ER measures downward infrared radiance from 3.3 to 25 mm (400 to 3000 cm1) with a spectral resolution of 0.482 cm1 (unapodized, first zero), outputting a sky radiance spectrum every 8 minutes [Revercomb et al., 1996]. The AERI-ER is calibrated to within 1% of the ambient radiance [Knutson et al., 1998] by having the interferometer view high-emissivity blackbody cavities before and after every scene view. The AERI-ER is an ideal instrument to measure the climatologically important microwindow radiances below 600 cm1, and therefore provides an important contribution to Arctic clear-sky measurement and model intercomparisons [Tobin et al., 1999]. 3.4. Radiative Transfer Models [12] An important aspect of the IPASRC-II project is the comparison of observed irradiances with those calculated by radiative transfer models. Two commonly used radiative transfer models are used for this purpose: LBLRTM v6.12 [Clough et al., 1992] and MODTRAN v4.2 [Berk et al., 2000]. Both of these models obtain line parameters from the HITRAN 2000 database and the water vapor continuum data from the CKD 2.4 model [Clough et al., 1989]. [13] The line-by-line radiative transfer model LBLRTM calculates spectral transmittances and radiances with high accuracy at monochromatic resolution. LBLRTM radiance calculations have been extensively validated against highresolution spectral measurements for many climatic conditions [Clough et al., 2000; Tobin et al., 1999; Walden et al., 1998]. A significant feature in LBLRTM v6.12 is a modification to the carbon dioxide continuum and the carbon dioxide line shape. LBLRTM radiances calculated for this intercomparison do not include the effects of aerosols or multiple scattering. [14] MODTRAN is a moderate spectral resolution model for the calculation of transmittances, radiances, and irradiances [Berk et al., 1999]. MODTRAN has been designed to reproduce the results of line-by-line models within the constraints of increased computational efficiency. All cal-

ACL

4-4


culations for this paper were performed using the following MODTRAN options: (1) the correlated-k method of radiative transfer, utilizing 33 k values, (2) 16-stream DIScrete Ordinate Radiative Transfer (DISORT) algorithm [Stamnes et al., 1988] to solve the radiative transfer equation, and (3) irradiance output at 1 cm1 resolution. [15] The AERI-ER measures zenith atmospheric radiance over the spectral range from 400 to 3000 cm1. Since the ASR and the pyrgeometers observe irradiance, an algorithm has been constructed to convert the AERI-ER spectral radiances to a broadband irradiance. This conversion occurs in three steps: (1) the atmospheric state is defined for the time of the AERI-ER measurement, and LBLRTM is used to compute radiances from 10– 400 cm1, (2) the AERI-ER spectral domain is extended to 10 cm1 using LBLRTMcalculated radiances, and (3) a radiance-to-irradiance algorithm is applied to the AERI-ER radiances [Clough et al., 2000]. [16] The ratio of irradiance to zenith radiance depends on the transmittance of the atmosphere for a given spectral element. For opaque conditions the radiation field is isotropic and this ratio equals p. However, for nonopaque conditions the ratio is greater than p. To obtain the appropriate ratio, the rapid radiation model RRTM v3.0 [Mlawer et al., 1997; Delamere et al., 2002] is employed. Utilizing the same atmospheric profile as that for the LBLRTM calculation, RRTM computes the ratio of the downward surface irradiance (using three Gaussian quadrature angles) to the downward zenith surface radiance for each of its 16 spectral bands. For each RRTM spectral band, the AERI-ER radiances falling within the band are summed and the RRTM irradiance-to-radiance ratio applied; the 16 spectral-band irradiances are then summed to produce the broadband AERI-ER irradiance. For consistency, the LBLRTM-calculated irradiance is obtained from the LBLRTM radiances in a manner identical to that used for the AERI-ER conversion.

group of the seven modified pyrgeometers. The original pyrgeometers have USER calibration factors, which were supplied by the owners of the instruments. The dome correction factor for all of these instruments is k3 = 4, as recommended by Eppley. The modified pyrgeometers have USER and PMOD calibration factors, which were determined just after the modification of the instruments by the blackbody source of the World Radiation Center (WRC), Switzerland. The USER set consists only of C and k3 (k1 = 0, k2 = 1) forcing the use of the Albrecht et al. [1974] equation, whereas the PMOD set applies all correction factors allowing the use of the Philipona et al. [1995] equation. Furthermore, both groups make use of CMDL calibration factors, which were obtained just before or after IPASRC-II with the blackbody source of NOAA’s Climate Monitoring and Diagnostics Laboratory (CMDL). Thus, USER, CMDL and PMOD calibration sets were all determined with blackbody calibration sources. [19] A FIELD calibration set was applied to all participating instruments by comparing the pyrgeometer measurements to the ASR reference instrument. The field calibration was performed in two steps. First, on March 5 at 6:30 LT, during low cloud conditions (low net irradiance, therefore, low thermopile signal and little temperature difference between dome and body) the irradiance is mainly a function of temperature. This allows the irradiance of the individual pyrgeometers to be matched to the mean value of the group by changing the k2 factor, which is unity in the Albrecht et al. equation. Secondly, during the clear-sky night of March 12 the thermopile sensitivity C was adjusted such that the pyrgeometer irradiance value matched the ASR derived irradiance at 22:36 LT. Hence, FIELD calibration factors were determined only with two time sequences (one overcast, one clear-sky) and subsequently used for all 10 days of the campaign.

5. Measurements 4. Pyrgeometer Equation and Calibration Sets [17] The pyrgeometer longwave irradiance EL was computed from the raw data by using the following equation: EL ¼

Uemf 1 þ k1 sTB3 þ k2 sTB4 k3 s TD4 TB4 ; C

ð1Þ

with Uemf the thermopile signal, TD and TB the absolute temperature of the pyrgeometer dome and body, s the Stefan-Boltzmann constant, and C the sensitivity of the thermopile. The three k-factors, k1, k2, k3 are correction factors for the dome and body temperature measurements. This equation was introduced by Philipona et al. [1995] and would be equivalent to the Albrecht et al. [1974] formula when k1 is zero and k2 is unity. This is in fact the case for the original blackbody calibrated pyrgeometers, on which k1 and k2 are assumed to be zero and unity, respectively. [18] In order to analyze the influence of the pyrgeometer calibration on the results, different calibration factor sets were applied for the derivation of longwave irradiance. We distinguished between USER, CMDL, PMOD and FIELD calibration sets (cf. Table 1). In the following figures these names are always used together with the acronyms ‘‘org’’ or ‘‘mod’’ to separate the group of the seven original from the

[20] March was chosen for IPASRC-II because a climatological analysis revealed the best chances for cold, clear-sky conditions. The comparison started with a day of low-level stratus clouds. The remaining nine days were mostly clear with a minimum of longwave irradiance measured on March 13. These ideal conditions allowed us to compare downward longwave irradiance between 120 and 240 W m2, which covers the typical conditions of an arctic winter atmosphere. All data presented in this paper are referred to local standard time (LT), which is UTC minus 9 hours. [21] Pyrgeometer measurements were taken every second, but only 1-minute averages were stored. Nighttime and daytime pyrgeometer measurements are separated in this analysis because pyrgeometer data are generally more consistent during nighttime. Since downward longwave irradiances during clear-sky nights show a minimum just before dawn, data between 5:00 and 6:00 LT were chosen for the analysis of nighttime measurements. Daytime downward longwave irradiances show their maximum during clear-sky days usually 2 – 3 hours after solar noon. The hour between 15:00 and 16:00 LT was therefore chosen for daytime investigations. [22] Since the Absolute Sky-scanning Radiometer does not have any optical components, it was only used during


nighttime. Absolute measurements could therefore only be taken during clear-sky nights. One scan of the Absolute Sky-scanning Radiometer lasts 24 minutes and is dependent on a homogenous sky during this period. Due to strong winds encountered during the campaign, the sensitive pyroelectric detector of the ASR needed to be installed in the lee of a wind shield in order to make successful measurements. Although this shield took half of the hemispherical field of view, tests in windless conditions demonstrated that successful measurements could be done with only half the hemisphere as long as the sky was homogenous. Furthermore, the ASR could only be used during the last three days of the experiment due to technical problems. A total of 25 scans were performed during this time and used to compare to the pyrgeometer measurements, which were averaged during the time of the scan.

6. Pyrgeometer Precision [23] In order to analyze the comparability of pyrgeometer measurements, hourly daytime (15:00 – 16:00 LT) and nighttime (5:00 – 6:00 LT) averages were calculated for each individual pyrgeometer with all calibration sets. The difference between the maximum and minimum value of these hourly averages was then used as an indicator of the measurement uncertainty. The lower panels of the graphs in Figure 1 (original pyrgeometers) and Figure 2 (modified pyrgeometers) show these hMax-Mini differences for each day and the average hMax-Mini difference for the whole 10-day period. The upper panels show the difference between the hourly mean irradiances of the individual pyrgeometers and the mean value measured by all pyrgeometers of this group. This information can therefore be used to trace the behavior of individual instruments. Nighttime measurements are plotted on the left graphs, daytime measurements on the right graphs. Since three of the seven modified pyrgeometers were not operational on the first day of the campaign the hMax-Mini difference for this day is not included in the calculation of the average hMax-Mini difference for this pyrgeometer group. [24] Figure 3 summarizes the results of Figures 1 and 2 by plotting the nighttime and daytime average hMax-Mini difference for both groups of pyrgeometers and all calibration sets. Standard deviations of these calculated differences are always about 0.5 W m2. Daytime measurements show generally slightly larger differences than nighttime measurements due to increased temperature stress within the instruments. Modified pyrgeometers show about the same agreement as the original pyrgeometers. Consistency between pyrgeometer measurements generally improves by about a factor of three going from the uniform CMDL and PMOD blackbody calibration (6 W m2) to the FIELD calibration (2 W m2). 6.1. Nighttime Precision [25] The nighttime precision for the original pyrgeometers was analyzed with three different calibration sets. With the USER calibration factors the maximum difference between individual pyrgeometers was found to be a little more than 6 W m2. The average hMax-Mini difference over all 10 days was calculated as 5.4 W m2. The CMDL calibration factors show a slightly better agreement between

ACL

4-5

the individual instruments resulting in average difference of 4.7 W m2. A much larger improvement is achieved after the FIELD calibration, which decreases the hMax-Mini difference to 1.1 W m2. The maximum difference between the seven pyrgeometers during the 10 days of measurements with irradiances between 120 and 230 W m2 was only 1.7 W m2 for FIELD calibrated original pyrgeometers. [26] Similar results were found for nighttime measurements of the modified pyrgeometers. The USER calibration is not shown in Figure 2, but the results are very similar to the PMOD calibration (cf. Figure 3), since their factors were both determined in the same blackbody source. With the CMDL calibration factors the average hMax-Mini difference is 5.4 W m2, which is only slightly worse than the PMOD difference of 5.0 W m2. The characteristic small temperature difference between body and dome of the pyrgeometers during the experiment might explain why the PMOD calibration set did not perform better despite the inclusion of all the correction terms in equation (1). By far the best consistency between the modified instruments was again found for the FIELD calibration set, which shows an average hMax-Mini difference of only 1.3 W m2. 6.2. Daytime Precision [27] Daytime hMax-Mini differences are analyzed in the right graphs of Figures 1 and 2. They generally tend to be slightly less precise (on average about 0.5 W m2 larger differences) than the nighttime measurements, although the CMDL calibration for the original pyrgeometers and the PMOD calibration for the modified pyrgeometers show surprisingly slightly better agreement during daytime. With the typical standard deviation of 0.5 W m2 of the average hMax-Mini difference these differences are not significant at all. There is also no significant difference between the original and the modified group of pyrgeometers. With the FIELD calibration factors for example the maximum difference between individual pyrgeometers was found to be 2.9 W m2 for the original pyrgeometers and 2.5 W m2 for the modified pyrgeometers. The overcast conditions of March 5 generally improved the consistency of the instruments, but minimal differences were also found during clear-sky conditions.

7. Pyrgeometer Accuracy [28] This section focuses on the absolute uncertainty of pyrgeometer irradiances by comparing them to the measurements of the Absolute Sky-scanning Radiometer. Figure 4 shows the result of these analyses by plotting the mean irradiance difference between the 25 ASR scans and the corresponding pyrgeometer measurements for all seven calibration sets. The variance between the different measurements is given as the standard deviation for each calibration sets, which is typically only about 0.7 W m2. The outermost marker attached to each bar indicates the maximum bias observed between an individual pyrgeometer and the ASR measurement. With the exception of the USER calibration set the mean irradiance of all other calibration sets are within ±1 W m2 of the ASR measurement. The mean of the FIELD calibrated pyrgeometers is even within ±0.5 W m2 of the ASR measurement. The maximum biases in the graph indicate that with the exception of the

ACL

4-6


Figure 1. Downward longwave irradiance measured with the seven original pyrgeometers (serial number in legend) and the Absolute Sky-scanning Radiometer (ASR) during one nighttime hour (left) and one daytime hour (right) for each day between March 5 and 15. Different calibration factor sets (USER, CMDL, FIELD) were applied to show their influence on the precision of the instruments. The lower panel always shows the hMax-Mini differences between the seven pyrgeometers, the upper panel indicates the difference of each pyrgeometer to the hourly mean of all seven instruments.


ACL

Figure 2. Downward longwave irradiance measured with the seven modified pyrgeometers (serial number in legend) and the Absolute Sky-scanning Radiometer (ASR) during one nighttime hour (left) and one daytime hour (right) for each day between March 5 and 15. Different calibration factor sets (CMDL, PMOD, FIELD) were applied to show their influence on the precision of the instruments. The lower panel always shows the hMax-Mini differences between the seven pyrgeometers, the upper panel indicates the difference of each pyrgeometer to the hourly mean of all seven instruments.

4-7

ACL

4-8


Figure 3. Nighttime and daytime hMax-Mini differences for both groups of pyrgeometers and all calibration sets (summary of Figures 1 and 2). USER calibration for the original pyrgeometers the accuracy of an individual pyrgeometer measurement for all other calibrations sets is in the worst case within ±2.5 W m2. [29] These results are only valid for the three nights with irradiances between 120 and 150 W m2, where measurements of the ASR were available. In order to analyze all 10 days of the experiment, including daytime data, pyrgeometer measurements were compared to the mean value of the FIELD calibrated group of original pyrgeometers (FIELDorg), which yielded the highest precision in Figure 3 and high accuracy in Figure 4. As indicated in Section 4, the FIELD calibration factors have been determined with respect to one well chosen absolute sky-scan and the good agreement of FIELD derived irradiances with ASR measurements has been demonstrated in the previous paragraph. The results of this comparison are plotted in Figure 5 as the difference between the mean value of each group and the FIELDorg group for the whole measurement period. Nighttime differ-

Figure 4. Performance of pyrgeometer irradiances with different calibration sets compared to Absolute Skyscanning Radiometer (ASR) measurements. The error bars indicate the standard deviation and the maximum bias for an individual pyrgeometer.

Figure 5. Accuracy evaluation of pyrgeometer irradiances during the 10 days of the experiment derived with different calibration sets with respect to the ASR adjusted FIELD calibration of the original pyrgeometers (FIELDorg). ences (plotted in the lower part of the graph) are again within about ±1 W m2 to the FIELDorg reference group for all calibration sets excluding the USER sets. The overcast case of March 5 shows the same good agreement with the exception of the FIELDmod group, whose mean value is not really representative of this day since three of seven instruments were not operational. The upper part of Figure 5 shows a similar agreement for the daytime measurements. Only the USERorg group shows significantly higher (1 W m2) daytime than nighttime biases to the FIELDorg reference group.

8. Longwave Measurements Compared to Model Calculations [30] An important component of IPASRC-II is the examination of the radiative fields from both a broadband and spectral perspective. Comparing measured and modeled irradiances within specific spectral intervals reveals whether or not specific physical processes are correctly represented in either the models or the specification of the atmospheric state. [31] This effort closely parallels not only IPASRC-I but also the AERI/LBLRTM Quality Measurement Experiments (QMEs) underway at the ARM Southern Great Plains CART site [Clough et al., 2000; Revercomb et al., 2003]. The QMEs have been critical in the identification of a number of issues, such as a dry bias in the radiosonde relative humidity measurements, and AERI calibration issues. [32] The cases chosen for the IPASRC-II intercomparison were carefully screened for availability of coincident nighttime ASR measurements and radiosonde launches. Three times met these criteria: 2:15 LT, March 11th; 2:00 LT, March 12th; 5:00 LT, March 13th; the measured irradiances for these cases range from 120 to 150 W m2. For each case study the instrument data were chosen as follows:


the ASR scan nearest to the sonde launch was used, the AERI-ER observation centered between 5 and 10 minutes after the sonde launch was used, and the pyrgeometer data were averaged over 30 minutes after the sonde launch. [33] Radiative transfer calculations require specification of the atmospheric state, which includes the vertical profile of the pressure, temperature, water vapor, ozone, and other trace gases. The pressure, temperature, and water vapor profiles were provided by balloon-borne sounding systems. Two radiosondes packages were used during the campaign: the Vaisala RS-80 and the Meteolabor Snowwhite (chilled mirror hygrometer). The two packages were deployed on the same balloon for the March 12 and March 13 cases, thereby providing two measurements of the atmospheric state for each observation period. The known dry-bias of the RS-80 sonde [Wang et al., 2002] was not corrected, because no significant difference was found compared to the Snowwhite sonde, which is not known to have a bias. The ozone profile was constructed by scaling the sub-arctic winter climatological profile to match the daily total column ozone, measured by a Dobson Ozone spectrophotometer located nearby at the NOAA CMDL Barrow station. The carbon dioxide concentration, also measured at the CMDL Barrow station, was 375 ppm. Climatological values of the other radiatively important trace gases, including methane, nitrous oxide and chlorofluorocarbons were assumed. The baseline LBLRTM and MODTRAN calculations do not include aerosols. [34] The relative placement of the instruments is an important consideration in the measurement-model intercomparison. Since the ASR is the reference instrument, the model calculations extend from the top of the atmosphere to the ASR detector. The ASR and the pyrgeometers were located on a deck above the NSA instrument shelter. However, the AERI-ER was enclosed roughly two meters below the ASR within the NSA instrument shelter. Therefore, an effective AERI-ER radiance was constructed which represents the spectral radiance at the position of the ASR. This was accomplished by replacing the measured AERIER radiances for the most opaque spectral elements by an appropriate linear combination of the AERI-ER radiances and those calculated by LBLRTM. Resultantly, for the frequencies corresponding to the greatest opacity, e.g. the center of the 667 cm1 carbon dioxide band, the measured AERI_ER radiances are replaced with the LBLRTM-calculated radiances. 8.1. Spectral Intercomparison [35] Figure 6 presents a spectral analysis for the March 11, 2:15 LT case, for which the atmospheric state was measured by a Vaisala RS-80 sonde. The effective AERIER irradiance is plotted on the top panel for the spectral range from 400 to 1400 cm1. The bottom panel illustrates the irradiance residuals (observed-calculated) for both LBLRTM and MODTRAN. The MODTRAN and the AERI-ER/LBLRTM irradiances were initially at two different resolutions. A Gaussian window with a half-width-halfmaximum of 5 cm1 was convolved with the AERI-ER, LBLRTM, and MODTRAN irradiances. The resulting irradiances were calculated on a 1 cm1 grid to provide consistency between the models and measurements. [36] Spectral analysis of model calculations versus measurements simultaneously addresses the quality of the spec-

ACL

4-9

Figure 6. The effective AERI-ER flux spectrum for March 11, 2001 at Barrow is shown in the top panel. The lower panel shows flux residuals (observation minus model), convolved with a Gaussian window (5 cm1 HWHM), between the effective AERI-ER and model calculations with LBLRTM (solid line) and MODTRAN (dotted line).

ification of the atmospheric state, radiometric measurements, and radiative transfer modeling. As an example, the calculated irradiances are highly dependent on the specification of temperature and trace gas profiles. The LBLRTM residuals in the 400– 600 cm1 principally reflect uncertainties in the water vapor profile. The LBLRTM residuals in the 980– 1080 cm1 spectral region are primarily due to the scaled climatological ozone profile used. Residuals for the other cases in the intercomparison have similar features and magnitudes. [37] The residuals in the 800– 1200 cm1 atmospheric window are on the order of 0.005 W m2/cm1. Similar residuals were obtained for AERI-ER versus LBLRTM radiance intercomparisons at the Surface Heat Energy Budget of the Arctic Ocean (SHEBA) experiment [Tobin et al., 2002]. These residuals are not attributable to possible inaccuracy of the water vapor profiles. A sensitivity calculation performed with a large increase in water vapor did not significantly change the magnitude of the residuals in this region. The calibration of the AERI-ER is under study as a contributor to these residuals as this effect has been observed over a broad range of AERI observations (D. Turner, private communication), although there may also be contributions from aerosols and diamond dust. Residuals due to the AERI-ER calibration will primarily affect spectral elements with low radiance values. [38] The total irradiance residual between the AERI-ER and LBLRTM for March 11, 2001, 2:15 LT, is 5.1 W m2. Approximately 1 W m2 is associated with the ozone dominated region from 980 to 1080 cm1, with the remain-

ACL

4 - 10


profile and approximately 1.4 W m2 to uncertainty in the water vapor profile. The mean difference between the MODTRAN and LBLRTM results is 1.2 W m2, primarily due to the difficulties in the approach used in MODTRAN to compute the radiation associated with the wings of the CO2 band around 667 cm1. [41] To assess the possible radiative impacts of high aerosol loading for this region [Stone, 2002], a MODTRAN calculation was performed using a standard rural extinction model with 50 km visibility. As shown in Figure 7, the MODTRAN calculations with aerosols increased the downward longwave irradiance by about 0.75 W m2 relative to the nonaerosol calculations.

9. Conclusions Figure 7. Mean differences between measured longwave irradiance (PIR, AERI-ER), model calculations (LBLRTM, MODTRAN) and the ASR measurements for clear-sky nighttime conditions. The dashed lines at (±1 W m2) indicate the uncertainty of the ASR measurement. The MODTRAN mean difference is smaller than the LBLRTM mean difference due to cancellation of errors, as discussed in section 8.1.

ing irradiance residuals due to the other causes mentioned above, including the AERI-ER calibration and possible uncertainty in the water vapor profile. [39] The residuals between AERI-ER and MODTRAN (no aerosols) follow roughly the same shape as the LBLRTM residuals with the exception of the wings of the carbon dioxide band (525 – 625 cm1, 710 – 760 cm1), spectral regions which are particularly difficult to model using the MODTRAN approach (A. Berk, personal communication). The cancellation of errors between the large negative MODTRAN residuals in the carbon dioxide spectral region and the generally positive residuals elsewhere results in a value for the spectrally integrated irradiance closer to the broadband observation than LBLRTM. 8.2. Broadband Intercomparison [40] The broadband results of the measurement-model intercomparison for the described three cases are presented in Figure 7 in which the mean difference between pyrgeometer fluxes (represented with the CMDLorg-group), AERI-ER irradiances, LBLRTM and MODTRAN calculations and the ASR measurement are plotted. The standard deviation about the respective means demonstrate the consistency over the three cases but does not reflect the accuracy of the components in the intercomparison. The model calculations utilized atmospheric profiles from the Vaisala RS-80 sondes. Investigations using Meteolabor Snowwhite sondes provided similar results. For the three cases examined, the mean irradiances from the AERI-ER, the ASR and the pyrgeometers agree to within 1 W m2. The assessment of the measurement and model performance using the ASR as reference must be done in context of the ASR irradiance uncertainty of ±1 W m2. The mean difference between the LBLRTM results and the ASR measurements is 2.8 W m2. Of this difference, approximately 1 W m2 may be attributed to an error in the ozone

[42] Data obtained during IPASRC-II have been analyzed to quantify the uncertainty of current longwave irradiance measurements and model calculations in arctic winter conditions. Precision or comparability of individual pyrgeometer measurements largely improves going from USER to FIELD calibration factors. Nighttime hMax-Mini differences typically decrease from about 5 W m2 to only 1 W m2. Daytime precision is generally about 0.5 W m2 worse than nighttime measurements, which seem to be caused by thermal stress within the instruments. Consistency between individual daytime pyrgeometer measurements generally improves by factor of three going from the uniform CMDL and PMOD blackbody calibration (6 W m2) to the FIELD calibration (2 W m2). The field calibration demonstrates the limits of pyrgeometer measurement precision using only blackbody calibration factors. Improvements of precision between instruments were demonstrated on the basis of a final field calibration, comparing pyrgeometers with the Absolute Sky-scanning Radiometer. [43] Absolute uncertainty improves the most going from USER to FIELD calibration. Mean nighttime values of the CMDL and PMOD group are within less than ±1 W m2. Furthermore, the mean nighttime values of the two FIELD groups are even within less than ±0.5 W m2. Daytime accuracy was investigated by comparing all groups to the FIELDorg group, which uses field calibration factors determined with the absolute measurement of the ASR. Excluding the two USER calibration groups mean values for all other groups of seven pyrgeometers show an overall accuracy of about ±1 W m2 during the 10 days of the experiment. These results demonstrate that the ASR and pyrgeometers, despite the fact that they use two completely different measurement methods, produce statistically very similar results. This is satisfactory since it proves that the blackbody calibration does not produce offsets for extreme conditions like an arctic winter atmosphere, although pyrgeometer domes are not at all ideal with respect to spectral longwave transmissions. Nevertheless, the results also demonstrate that each blackbody pyrgeometer calibration can only be a first step towards accurate longwave radiation measurements and that the fine-tuning of k2 and C has to be done in the field with the help of an absolute reference standard. [44] In contrast to the findings of IPASRC-I this time the group of modified pyrgeometers neither showed better


precision nor less sensitivity to the thermal stress of daytime measurements compared to the group of original pyrgeometers. The reason might be that the influence of the first term in equation (1) was larger during IPASRC-II caused by the large longwave net radiation (large EMF output) typical for clear-sky arctic winter conditions compared to the temperature measurements of the second and third term. [45] Radiative transfer model calculations of clear-sky nighttime situations (aerosol loading taken into account) are in good agreement (2 W m2) with the ASR measurements. This agreement would be further enhanced with improved specification of the atmospheric state, including the water vapor and ozone profiles. These improvements along with analyses of the water vapor line parameters and continuum are topics of future study. [46] In summary, IPASRC-II has successfully demonstrated that downward longwave irradiance can be measured and modeled with the same uncertainty in arctic winter climate as in midlatitude climate despite the different meteorological and atmospheric conditions. Daytime discrepancies between pyrgeometers are even smaller in the Arctic than in midlatitude climates due to less thermal stress within the instruments. [47] The next step will be to merge the findings of IPASRC-I and IPASRC-II to find out if midlatitude determined field calibration factors can successfully be applied in the arctic or vice versa. This and further investigations should all be carried out with the final goal to establish an absolute standard for longwave radiation measurements. [48] Acknowledgments. The authors would like to thank the ARM NSA site manager Bernie Zak for all the provided support and contacts. Without the experience and endless help of the NSA operations manager Jeff Zirzow and staff Walter Brower and Jimmy Ivanoff, IPASRC-II could never have been successfully finished. A special thanks goes to Dan Endres, station manager of the nearby NOAA CMDL site, and also to Francis Schmidlin and his crew from NASA Wallops for providing and operating the Snowwhite sondes. The authors would like to acknowledge the BSRN community, which initiated the mentioned longwave irradiance uncertainty investigations. A particular thanks goes to all the institutions and laboratories that provided pyrgeometers to the comparison. The authors would also like to thank all the people who provided support for the model part of the experiment, particularly Gail Anderson, Alexander Berk, and Norm Wood.

References Albrecht, B., M. Pellot, and S. Cox, Pyrgeometer measurements from aircraft, Rev. Sci. Instrum., 45, 33 – 38, 1974. Berk, A., et al., MODTRAN4 radiative transfer modeling for atmospheric correction, SPIE Proc., 3756, 348 – 353, 1999. Berk, A., G. P. Anderson, P. K. Acharya, J. H. Chetwynd, L. S. Bernstein, E. P. Shettle, M. W. Matthew, and S. M. Adler-Golden, Modtran4 users manual, Air Force Res. Lab., Space Vehicle Dir., Air Force Mater. Command, Hanson Air Force Base, Mass., 2000. Charlock, T. P., and V. Ramanathan, The albedo field and cloud radiative forcing produced by a general circulation model with interally generated cloud optics, J. Atmos. Sci., 42, 1408 – 1429, 1985. Clough, S. A., F. X. Kneizys, L. S. Rothman, and R. W. Davies, Line shape and the water vapor continuum, Atmos. Res., 23, 229 – 241, 1989. Clough, S. A., M. J. Iacono, and J.-L. Moncet, Line-by-line calculations of atmospheric fluxes and cooling rates: Application to water vapor, J. Geophys. Res., 97, 15,761 – 15,785, 1992. Clough, S. A., A. D. Brown, C. Andronache, E. J. Mlawer, T. R. Shippert, D. D. Turner, D. C. Tobin, H. E. Revercomb, and R. O. Knuteson, A longwave broadband QME based on ARM pyrgeometer and AERI measurements, paper presented at Tenth ARM Science Team Meeting, San Antonio, Tex., 13 – 17 March 2000. Curry, J. A., W. B. Rossow, D. Randall, and J. L. Schramm, Overview of arctic cloud and radiation characteristics, J. Clim., 9(8), 1731 – 1764, 1996.

ACL

4 - 11

Dehne, K., U. Bergholter, and F. Kasten, IEA comparison of longwave radiometers in Marburg, 1989, 1990, Tech. Rep. IEA-SHCP-9F-3, Dtsch. Wetterdienst, Meteorol. Obs. Hamburg, Hamburg, Germany, 1993. Delamere, J. S., S. A. Clough, E. J. Mlawer, S. Boukabara, K. E. CadyPereira, and M. Shepard, An update on radiative transfer model development at Atmospheric and Environmental Research, Inc., in Proceedings of 12th Annual ARM Science Team Meeting, U.S. Dep. of Energy, Washington, D. C., 2002. Dutton, E. G., An extended comparison between LOWTRAN7 computed and observed broadband thermal irradiances: Global extreme and intermediate surface conditions, J. Atmos. Oceanic Technol., 10(3), 326 – 336, 1993. Fairall, C. W., P. O. G. Person, E. F. Bradley, R. E. Payne, and S. Anderson, A new look at calibration and use of Eppley Precision Infrared Radiometers: part I. Theory and application, J. Atmos. Oceanic Technol., 15, 1230 – 1243, 1998. Knutson, R. O., F. A. Best, H. E. Revercomb, D. C. Tobin, P. F. W. van Delst, and V. P. Walden, AERI instrument status and analysis results, paper presented at Eighth ARM Science Team Meeting, Atmos. Radiat. Meas. Program, Tucson, Ariz., 23 – 77 March 1998. Marty, C., R. Philipona, C. Frohlich, and A. Ohmura, Altitude dependence of surface radiation fluxes and cloud forcing in the alps: Results from the alpine surface radiation budget network, Theor. Appl. Climatol., 72, 137 – 155, 2002. Mlawer, E. J., S. J. Taubman, P. D. Brown, and M. J. Iacono, Radiative transfer in inhomogeneous atmospheres: RRTM, a validated correlatedk model for the longwave, J. Geophys. Res., 102, 16,663 – 16,682, 1997. Ohmura, A., et al., Baseline Surface Radiation Network (BSRN/WCRP): New precision radiometry for climate research, Bull. Am. Meteorol. Soc., 79, 2115 – 2136, 1998. Philipona, R., Sky-scanning radiometer for absolute measurements of atmospheric longwave radiation, Appl. Opt., 40, 2376 – 2383, 2001. Philipona, R., C. Betz, and C. Frohlich, Characterization of pyrgeometer and the accuracy of atmospheric longwave radiation measurements, Appl. Opt., 34(9), 1598 – 1605, 1995. Philipona, R., et al., The Baseline Surface Radiation Network pyrgeometer round robin experiment, J. Atmos. Oceanic Technol., 15, 678 – 696, 1998. Philipona, R., et al., Atmospheric longwave irradiance uncertainty: Pyrgeometers compared to an absolute sky-scanning radiometer, AERI and radiative transfer model calculations, J. Geophys. Res., 106, 28,129 – 28,141, 2001. Revercomb, H. E., et al., Atmospheric Emitted Radiance Interferometer (AERI), part 1, Status, basic radiometric accuracy, and unexpected errors and solutions, paper presented at Sixth ARM Science Team Meeting, Atmos. Radiat. Meas. Program, San Antonio, Tex., 4 – 7 March 1996. Revercomb, H. E., et al., The Atmospheric Radiation Measurement (ARM) program’s water vapor intensive observation periods: Overview, initial accomplishments, and initial challenges, Bull. Am. Meteorol. Soc., 84(2), 217 – 236, 2003. Sinha, A., and J. E. Harries, Water vapor greenhouse trapping: The role of far infrared absorption, Geophys. Res. Lett., 22, 2147 – 2150, 1995. Stamnes, K., Radiation in the Arctic: Why worry?, in Proceedings of International Radiation Symposium ’96: Current Problems in Atmospheric Radiation, edited by W. Smith and K. Stamnes, pp. 3 – 6, A. Deepak, Hampton, Va., 1996. Stamnes, K., S. Tsay, W. Wiscombe, and K. Jayaweera, Numerically stable algorithm for discrete-ordinate-method radiative transfer in multiple scattering and emitting layered media, Appl. Opt., 27, 2502, 1988. Stamnes, K., R. G. Ellingson, J. A. Curry, J. E. Walsh, and B. D. Zak, Review of science issues, deployment strategy, and status for the ARM north slope of Alaska-Adjacent arctic ocean climate research site, J. Clim., 12(1), 46 – 63, 1999. Stone, R. S., Monitoring aerosol optical depth at Barrow, Alaska, and South Pole: Historical overview, recent results, and future goals, in Proceedings of 9th Workshop on Italian Research on the Antarctic Atmosphere, Rome, Italy, Oct. 22 – 24, 2001, Ital. Phys. Soc., Bologna, Italy, 2002. Tobin, D. C., et al., Downwelling spectral radiance observations at the SHEBA ice station: Water vapor continuum measurements from 17 to 26 mm, J. Geophys. Res., 104, 2081 – 2092, 1999. Tobin, D. C., R. O. Knuteson, and H. E. Revercomb, Observed and calculated downwelling longwave spectral radiances at the SHEBA Ice Station: Prelude to an Arctic AERI/LBLRTM QME, in Proceedings of 10th Annual ARM Science Team Meeting, U.S. Dep. of Energy, Washington, D. C., 2002. Walden, V. P., S. G. Warren, and F. J. Murcray, Measurement of the downwelling longwave radiation spectrum over the Antarctic Plateau and comparison with a line-by-line radiative transfer model for clear skies, J. Geophys. Res., 103, 3825 – 3846, 1998.

ACL

4 - 12


Wang, J., H. L. Cole, D. J. Carlson, E. R. Miller, K. Beierle, A. Paukkunen, and T. P. Laine, Corrections of humidity measurement errors from the Vaisala RS80 radiosonde-Application to TOGA COARE data, J. Atmos. Oceanic Technol., 19, 981 – 1002, 2002. Wild, M., A. Ohmura, H. Gilgen, J. J. Morcrette, and A. Slingo, Evaluation of downward longwave radiation in general circulation models, J. Clim., 14(15), 3227 – 3239, 2001.

S. A. Clough, J. Delamere, and E. J. Mlawer, Atmospheric and Environmental Research, 131 Hartwell Ave., Lexington, MA 02421, USA.

E. G. Dutton, Climate Monitoring and Diagnostics Laboratory, NOAA, Boulder, CO 80305, USA. C. Marty and R. Philipona, Physikalisch-Meteorologisches Observatorium and World Radiation Center, Davos Dorf, Dorfstrasse 33, CH-7260 Davos-Dorf, Switzerland. ([email protected]) J. Michalsky, Atmospheric Science Research Center, State University of New York at Albany, Albany, NY 12203, USA. K. Stamnes, Stevens Institute of Technology, Hoboken, NJ 07030, USA. T. Stoffel, National Renewable Energy Laboratory, Golden, CO 80401, USA. R. Storvold, Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK 99775, USA.