All-Sky Downwelling Longwave Radiation and Atmospheric-Column Water Vapour and Temperature over the Western Maritime Arctic

Abstract

Measurements of downwelling longwave radiation and atmospheric-column variables (precipitable water, mean vapour pressure, and mean temperature) derived from microwave radiometric profiles were collected over a composite year at various locations in the Beaufort Sea–Amundsen Gulf region of the Canadian Arctic. Cloud cover was specified by a temporal fractional cloud cover derived from ceilometer measurements. A logarithmic relationship was found between downwelling longwave radiation and atmospheric-column water vapour expressed as precipitable water or mean vapour pressure. This relationship explained about 84% of the variance with a standard error of around 9%. Downwelling longwave radiation was not as well correlated with mean atmospheric temperature. The parameterization of downwelling longwave radiation as a function of atmospheric-column variables, which can be analyzed more accurately than surface variables in data sparse regions, may contribute to improved climate modelling of the western maritime Arctic region. It was shown that both the annual cycle of monthly median precipitable water and water vapour intrusions influenced the magnitude of downwelling longwave radiation, and the impact was enhanced by cloud cover.

RÉSUMÉ [Traduit par la rédaction] Des mesures du rayonnement descendant de grandes longueurs d'onde et des variables de colonne atmosphérique (eau précipitable, pression de vapeur moyenne et température moyenne) dérivées de profils radiométriques en hyperfréquences ont été recueillies au cours d'une année composite à différents sites dans la région de la mer de Beaufort et du golfe d'Amundsen dans l'Arctique canadien. La couverture nuageuse était spécifiée comme une fraction temporelle de couverture nuageuse dérivée de mesures faites par célomètre. Nous avons trouvé une relation logarithmique entre le rayonnement descendant de grandes longueurs d'onde et la vapeur d'eau dans la colonne atmosphérique exprimée sous forme d'eau précipitable ou de pression de vapeur moyenne. Cette relation explique environ 84% de la variance avec une erreur-type d'environ 9%. Le rayonnement descendant de grandes longueurs d'onde n’était pas aussi bien corrélé avec la température atmosphérique moyenne. La paramétrisation du rayonnement descendant de grandes longueurs d'onde en tant que fonction des variables de la colonne atmosphérique, qui peuvent s'analyser plus précisément que les variables de surface dans les régions où il y a peu de données, peut contribuer à améliorer la modélisation du climat de la région ouest de l'Arctique maritime. Il apparaît que tant le cycle annuel de la médiane mensuelle de l'eau précipitable que les intrusions de vapeur d'eau ont influencé l'intensité du rayonnement descendant de grandes longueurs d'onde et l'effet a été amplifié par la couverture nuageuse.

Keywords:

• Arctic
• downwelling longwave radiation
• precipitable water
• mean temperature
• mean vapour pressure

1 Introduction

In the Arctic, downwelling longwave radiation (LWd) is a major component of the surface radiation budget and thus the climate. With the long polar night, the annual total LWd is generally more than double the annual downward shortwave radiation (Briegleb & Bromwich, 1998; Curry, Rossow, Randall, & Schramm, 1996; Persson, Fairall, Andreas, Guest, & Perovich, 2002; Walsh & Chapman, 1998). At high latitudes, during the winter-to-spring transition period, changes to the surface net radiation budget are primarily a result of changes in LWd (Zhang, Bowling, & Stamnes, 1997; Zhang, Stamnes, & Bowling, 1996; Zhang, Stamnes, & Bowling, 2001). These variations are critical to the onset and timing of snow melt (Sicart, Pomeroy, Essery, & Bewley, 2006). In maritime regions, LWd at the sea-ice surface may be larger than the turbulent heat flux from either the atmosphere or the ocean, making variations in LWd a significant factor in the thermodynamic or heat processes influencing the annual break-up and formation of Arctic sea ice (Key, Silcox, & Stone, 1996; Kay & Gettelman, 2009).

Where and when measurements are not available, the most accurate estimates of LWd are derived from complex radiative transfer models that consider properties of the entire atmospheric column (Key et al., 1996) including temperature and humidity profiles, cloud composition, thickness and base height, aerosol optical depth, and the amount of atmospheric ozone. These data are often unavailable. Thus, LWd is often parameterized using surface meteorological variables such as temperature and/or vapour pressure. For example, Brutsaert (1975) developed a physically based relationship between surface vapour pressure, surface temperature, and LWd under clear skies. This parameterization, based on the US 1962 standard atmosphere, is suitable for mid-latitudes. Jin, Barber, and Papakyriakou (2006) extended this clear-sky relationship to Arctic climatic conditions. Cloud generally increases LWd (Curry et al., 1996). All-sky LWd (i.e., LWd with all cloud cover including clear skies) has been parameterized by allowing the atmospheric emissivity to exceed 1.0 in Stefan-Boltzmann's equation based on the fractional cloud cover (Sicart et al., 2006). See Key et al. (1996) for a description and evaluation of several clear- and all-sky LWd parameterizations.

In data sparse regions such as the Arctic, gridded surface data from objective analyses are input to relationships between surface variables and LWd. These gridded data may be somewhat inaccurate because of data assimilation routines which, at each analysis cycle, combine observations of the current state of the atmosphere with the output from a numerical weather prediction model, a forecast. Essentially, this process tries to balance the uncertainty in the observations with the continuity of the model. In data sparse regions, objective analyses, and even reanalyses, of variables such as surface temperature and humidity, although affected by limited and generally distant observations, are more strongly influenced by modelled or forecast output. Upper-air temperature and humidity fields are less spatially variable and are more reliably influenced by distant observations. Thus, in data sparse regions the analyses of upper-air fields are generally more accurate than the analyses of surface fields (Kalnay et al., 1996), and bulk or atmospheric-column variables, such as precipitable water, mean temperature, and mean vapour pressure, can be specified more accurately than surface weather variables. Thus, it should be advantageous for climate modellers to employ parameterizations of LWd which use atmospheric-column variables rather than surface variables.

Here we determine relationships between hourly LWd measured from the CCGS Amundsen at various sites in the data sparse southeastern Beaufort Sea–Amundsen Gulf region of the Canadian Arctic during a composite year (late November 2007 to July 2008 and August 2009 to early November 2009), and atmospheric-column variables (precipitable water, mean vapour pressure, and mean temperature) derived from microwave radiometric profiles observed at the same times and locations. This parameterization of LWd with atmospheric-column variables may contribute to improved climate modelling of the data sparse western maritime Arctic.

2 Data

a Composite Year and Study Region

A composite year was created by combining the data from the International Polar Year, Circumpolar Flaw Lead (IPY-CFL) system study (November 2007 to July 2008) and ArcticNet2009 (August–November). The IPY-CFL project was an overwintering field campaign in Amundsen Gulf supported by the CCGS Amundsen (Barber et al., 2010). The collection of field data by the CCGS Amundsen expanded into the southeastern Beaufort Sea with ArcticNet2009 (Fig. 1).

Fig. 1 Locations of the CCGS Amundsen during the 2007–08 IPY-CFL field campaign and during the ArcticNet2009 field campaign. There is one dot per day at the median ship position on that day.

The annual sea-ice cycle defines the character of the study region. In December 2007 most of the southeastern Beaufort Sea and Amundsen Gulf were covered by unconsolidated sea ice. Analysis of data contained in the Canadian Ice Service digital archive (EC, 2011) revealed that the total sea-ice cover was around 97% with the other 3% being open leads and/or polynyas. The sea-ice cover exhibited little variation from 7 January to 21 April 2008; it approached 100% from 28 April to 12 May. From 19 May onward the total sea-ice cover declined rapidly to its July 2008 minimum of about 3% (Raddatz, Galley, & Barber, 2012). In August 2009, the total sea-ice cover was <10%. From 20 August to 15 October, it dropped to 4% before increasing rapidly to >90% by 5 November. The total sea-ice cover was once again greater than 95% in December 2009, completing the annual sea-ice cycle for the composite year.

b Water Vapour Density and Temperature Profiles

The Radiometrics MP-3000A^© Microwave Radiometer (MWR) aboard the CCGS Amundsen provided high temporal (approximately 1 minute) resolution temperature and water vapour density profiles (0 to 500 m at 50 m intervals, 500 to 2000 m at 100 m intervals, and 2000 to 10000 m at 250 m intervals) at various locations in Amundsen Gulf and the southeastern Beaufort Sea during the IPY-CFL (2007–08) and ArcticNet2009 field campaigns. Vertical profiles were derived from microwave brightness temperatures using the manufacturer's neural network retrieval and radiative transfer model that had been trained using measurements from the nearest operational upper-air station, Inuvik, Canada (68.30°N, 133.47°W). For explanations of radiometric profiling of temperature and water vapour, readers are referred to Solheim et al. (1998), Guldner and Spankuch (2001), Ware et al. (2003), and Gaffard et al. (2008). The high temporal resolution soundings from the MWR profiler were averaged to generate hourly profiles. This processing produced a manageable dataset while smoothing out random errors.

Throughout the IPY-CFL and ArcticNet2009 field campaigns, weather balloons carrying Vaisala RS92-SGPD radiosondes were launched from the CCGS Amundsen providing 68 useable radiosonde profiles to validate the MWR profiles. For water vapour density, the root mean square error (rmse) below 2000 m, using the radiosondes as the standard, averaged 0.25 × 10⁻³ kg m⁻³ for winter (JFM), 0.32 × 10⁻³ kg m⁻³ for spring (AMJ), 0.74 × 10⁻³ kg m⁻³ for summer (JAS), and 0.37 × 10⁻³ kg m⁻³ for autumn (OND); in all seasons the rmse decreased with height above 2000 m. For temperature, the rmse below 4000 m averaged 1.9 K for winter, 2.0 K for spring, 2.8 K for summer, and 2.5 K for autumn; in all seasons the rmse increased with height above 4000 m to a maximum of approximately 6 K. Bias calculations indicated that the measurements from the MWR profiler were slightly colder and drier than the radiosonde data. Details of the comparison between the MWR profiles and the radiosonde observations are available in Candlish, Raddatz, Asplin, and Barber (2012). Microwave radiometric profilers have been tested at a variety of other locations, and they have been found to have acceptable accuracy (Cadeddu, Peckham, & Gaffard, 2002; Gaffard et al., 2008; Guldner & Spankuch, 2001; Westwater, Han, Shupe, & Matrosov, 2001).

The hourly IPY-CFL MWR temperatures and water vapour profiles have been used to characterize and link the winter through early summer atmospheric boundary layer over Amundsen Gulf to the sea-ice cover (Raddatz, Asplin, Candlish, & Barber, 2011; Raddatz, Galley, & Barber, 2012). The addition of the late summer and fall profiles from ArcticNet2009 to the IPY-CFL time series created a composite year of hourly water vapour profiles which have been used to determine the annual cycle of monthly median precipitable water (PW), and the maximum monthly PW resulting from water vapour intrusions over the western maritime Arctic (Raddatz, Galley, Candlish, Asplin, & Barber, 2012). The annual cycle of PW was similar to that found by Serreze, Barry, Rehder, and Walsh (1995) using a comprehensive Arctic rawinsonde dataset. The maximum monthly PW resulting from water vapour intrusions was comparable to that found by Doyle et al. (2011) using MWR profiles from Eureka (80°N, 86°W).

c Downwelling Longwave Radiation (LWd)

An Eppley Pyrgeometer (model PIR^©) with a broadband coverage from 4 to 50 μm aboard the CCGS Amundsen measured downwelling radiation with high temporal resolution (approximately 1 minute). The sensor was used in “non-battery” mode. The LWd measurements were corrected to compensate for the sensor's temperature (blackbody correction) and for a small bias associated with the solar heating of the dome (Philipona, Betz, & Frohlich, 1995). Following Marty et al. (2003), the correction factors were k ₁ = 0, k ₂ = 1, and k ₃ = 4. As with the MWR profile data, the high temporal resolution LWd data were averaged to generate hourly values and to smooth out random errors.

The view of the pyrgeometer varied somewhat with the roll of the CCGS Amundsen. The pyrgeometer was mounted >20 m above sea level, and in the worst conditions (storm state with open water), the ship would only roll about 10°. Storm conditions were episodic and relatively short lived, and ocean waves were damped by ice cover for most of the year. Designed for flux measurements, a pyrgeometer has a directional or cosine response (Epply Laboratory, Inc. 2012; Minnett, 1999). For example, full response occurs when the radiation hits the sensor perpendicularly (0° from the zenith); zero response occurs when the radiation comes from the horizon (90° from the zenith). As the pyrgeometer responds weakly to radiation at large zenith angles, potential errors caused by longwave radiation from the ocean surface were assumed to be negligibly small.

d Cloud

A Vaisala CT25K^© Ceilometer aboard the CCGS Amundsen measured cloud-base height continuously using laser backscatter from water droplets and ice crystals. The ceilometer was positioned to ensure a clear view of the sky. It has a vertical resolution of 15 m to approximately 7.6 km (Vaisala, 2002). The observation interval was set to provide a maximum of 60 cloud-base height observations per hour for the IPY-CFL study and 90 observations per hour for the ArcticNet2009 campaign. Each hour's temporal fractional cloud cover was calculated as the number of observations with a measured cloud base relative to the total number of observations.

The microwave radiometric profiler also observed cloud-base height. It uses a zenith-pointing infrared radiometer (9.6–11.5 μm) to measure cloud-base temperature (if clouds are present). Cloud-base height is determined by comparing the cloud-base temperature with the temperature profile (Ware et al., 2003).

The MWR profiler and the ceilometer differed noticeably in their cloud measurements (Candlish et al., 2012). In general, the ceilometer measured considerably more hours with low clouds (bases ≤2000 m). Because the ceilometer has long been employed by weather services as the standard instrument for detecting clouds, we used this instrument's data rather than the MWR profiler's cloud-base heights.

3 Methodology

LWd parameterizations are generally empirical relationships that require surface vapour pressure, e, air temperature, T, or e/T as input (see for example, Key et al., 1996). Vapour pressure and temperature are related through the equation of state (Ideal Gas Law) for water vapour:

where ρ is the vapour density, and R_v is the gas constant for water vapour.

This study determined relationships between LWd and atmospheric-column variables rather than surface variables because atmospheric-column variables can be specified more reliably than surface variables in data sparse regions. The PW was determined from the hourly water vapour density profiles. Water vapour was measured at specific levels; the mean water vapour density, ρ_mi , for each layer, i, was multiplied by h_i , the layer's thickness, then these values were summed from the surface to the top of the atmospheric column, h_t  = 10000 m (American Meteorological Society, 2000). That is,

The value of PW (kilograms per square metre) converts directly to millimetres because 1 kg of water spread over 1 m² would produce a layer of water 1 mm deep given that the density of water is 1 × 10³ kg m⁻³ (Oke, 1987). Thus, PW is commonly expressed as the depth of liquid water (millimetres), if the water vapour were to be condensed. Mean atmospheric-column temperature, T _mean, was derived from the hourly temperature profiles (Zhang et al., 2001):

where T_mi is the mean temperature of layer i. Mean atmospheric-column vapour pressure, e _mean, was calculated from the PW and the mean temperature using the Ideal Gas Law:

Atmospheric-column (0 to 10,000 m) PW, mean vapour pressure, and mean temperature were calculated for each hour of the composite year from the water vapour density and temperature profiles (November 2007 to July 2008 and August 2009 to November 2009). The final dataset, containing all the coincident hourly observations from the pyrgeometer, the microwave radiometer, and the ceilometer, was divided into a developmental sample (approximately 90% of the hours) and a validation sample (approximately 10% of the hours).

Using the developmental sample, a regression relationship was derived between the LWd and the atmospheric-column variables. That is,

where LWd is the hourly LWd, f indicates “a function of”, and x is the hourly atmospheric-column PW, mean vapour pressure, or mean temperature. Cloud was accounted for by the term [1 + aC] where C is the temporal fraction of each hour with a cloud-base height measured by the ceilometer, and a is an empirical constant. The hourly temporal cloud fraction, C, varied from 0 for clear skies to 1 for cloudy skies. This departs from the standard practice of defining C as the areal fraction of the sky covered by cloud.

The potential impact of spatially variable cloud cover, with sky conditions for part of the celestial dome departing from those viewed by the ceilometer, was minimized by the use of the temporal fractional cloud cover based on cloud heights measured at approximately 1 minute intervals. Calculated over each hour, C should produce a representative cloud cover as a result of the ever-changing nature of the sky conditions. The impacts of cloud composition, thickness, and cloud-base height on LWd were not considered (Key et al., 1996).

4 Results and discussion

Based on the evaluation of MWR profiles by others (Cadeddu et al., 2002; Gaffard et al., 2008; Guldner & Spankuch, 2001; Westwater et al., 2001), the verification of our profiles by Candlish et al. (2012), and the previously published interpretations of these profiles, it was concluded that the MP-3000A^© Microwave Radiometer was capable of determining atmospheric-column PW, mean vapour pressure, and mean temperature over the western maritime Arctic. The hourly atmospheric-column variables were combined with the pyrgeometer's LWd measurements and the ceilometer's temporal cloud cover fractions. Hours when data were missing from one time series were filtered from all datasets. From this all-sky dataset of 5802 hours, 5222 hours were used as a developmental sample. The developmental sample had a mean LWd of 229.9 W m⁻², and a range of 129.9 to 354.4 W m⁻². The remaining 580 hours were used as a validation sample. This sample had similar statistics; the mean LWd was 229.7 W m⁻², and the range was 132.2 to 357.3 W m⁻². The measured LWd values were divided by an adjustment term (1 + aC) to account for cloud cover during all or part of each hour. The temporal cloud cover fraction is C, and the adjustment term equalled 1 for clear skies (i.e., when C = 0). The adjusted LWd time series was plotted against each of the atmospheric-column variables, PW, e _mean, and T _mean (Fig. 2), and the best-fit regression curves were determined. For water vapour, only the LWd_{all_sky} versus PW relationship is shown because the LWd_{all_sky} versus e _mean relationship is virtually the same. Through trial and error, it was found that the highest explained variance, R ², in each case, was obtained with a seasonally invariant a = 0.15. Others have generally employed an areal cloud cover factor and an adjustment term similar to ours for parameterizing all-sky LWd. For example, Marshunova (1966) used an areal cloud cover factor with a = 0.16 for the Arctic summer, and a = 0.31 for the Arctic winter. Jacobs (1978) also used an areal cloud cover factor and derived a mean value for a of 0.26 for Baffin Island, Canada, for the June–December period. Key et al. (1996) recommended Jacob's a value for sea-ice models.

Fig. 2 Hourly all-sky LWd (W m⁻²) divided by (1 + aC) versus atmospheric-column (a) PW (mm) and (b) mean temperature (°C) for the developmental sample.

From the developmental sample, the following all-sky relationships were determined between the observed LWd and the atmospheric-column variables where C varies from zero for clear skies to one for cloudy skies:

•

Hourly all-sky LWd (W m−2) as a function of atmospheric-column water vapour expressed as (i) precipitable water, PW (millimetres), LWdall_sky = [48.75 ln(PW) + 155.12][1 + 0.15C], with R 2 (explained variance) = 0.84 and σ (standard error of the estimate) = 19.3 W m−2 (8.4% of mean); (ii) mean vapour pressure, e mean (hPa), LWdall_sky = [47.06 ln(e mean) + 261.13][1 + 0.15C], with R 2 = 0.84 and σ = 19.4 W m−2 (8.4% of mean).

•

Hourly all-sky LWd (W m−2) as a function of atmospheric-column mean temperature, T mean (°C),    LWdall_sky = [4.78 T mean + 384.59][1 + 0.15C],    with R 2 = 0.67    and σ = 27.9 W m−2 (12.1% of mean).

From the validation sample, LWd_{all_sky} was estimated with each of the atmospheric-column variables, and these estimates were compared with the observed values. For the LWd_{all_sky} versus PW relationship, the rmse was 19.5 W m⁻² (8.5% of mean); for the LWd_{all_sky} versus e _mean relationship, the rmse was 19.6 W m⁻² (8.5%); for the LWd_{all_sky} versus T _mean relationship, the rmse was 27.9 W m⁻² (12.2%).

A plot (Fig. 3) of the observed hourly LWd_{all_sky} (with the 1:1 line indicated) versus the estimated values using atmospheric-column PW (Fig. 3a) and the estimated values using atmospheric-column T _mean (Fig. 3b) illustrates that the LWd_{all_sky} values from atmospheric-column moisture slightly overestimated the observed values, and this positive bias (estimated - observed) was greater if T _mean was used_. The biases were 0.07 W m⁻² for the moisture variables and 0.13 W m⁻² for mean temperature.

Fig. 3 For the validation sample, observed hourly all-sky LWd (W m⁻²) versus (a) estimated LWd from atmospheric-column PW (mm) and (b) estimated LWd from atmospheric-column mean temperature (°C).

To illustrate how the annual cycle of monthly median PW and monthly maximum PW (water vapour intrusions) impact the magnitude of LWd over the western maritime Arctic (Doyle et al., 2011; Raddatz, Galley, Candlish, et al., 2012; Serreze et al., 1995), the monthly median and maximum PW values were input to the atmospheric-column relationship obtained in this study. About 47% of the usable hours in our composite year had clear skies, thus for this rough calculation C was set to both zero (clear skies) and to one (cloudy skies) for each month. Monthly median PW values ranged from 1.4 mm for January to 16.8 mm for August. The maximum monthly PW values, attributed to water vapour intrusions, ranged from 3.4 mm for November to 37.3 mm for August (Raddatz, Galley, Candlish, et al., 2012). It is apparent (Fig. 4) that both the annual cycle of monthly median PW and water vapour intrusions influence LWd, and this impact is enhanced by cloud cover. The impact of water vapour intrusions was generally larger in winter than in summer. In winter, the difference between LWd for the median and maximum PW values was about 82 W m⁻² for clear skies and 95 W m⁻² for cloudy skies. The difference was about 38 W m⁻² in summer for clear skies and 45 W m⁻² for cloudy skies. The larger differences in winter reflect the logarithmic form of the relationship between LWd and PW.

Fig. 4 Annual cycle of LWd (W m⁻²) over the western maritime Arctic estimated from monthly median and maximum PW values using the all-sky parameterization.

The LWd calculated from the monthly median PW values increased steeply from March to May. It increased more slowly from May to August; it then dropped precipitously from August to November. The total ice-cover timeline (Raddatz, Galley, & Barber, 2012) generally coincided with the seasonal rise and fall of LWd. In the spring of 2008 the onset of break-up, defined as the first week when the total sea-ice concentration is <80%, was 14 May. The start of the open-water season, defined as the first week when the total sea-ice cover is <20%, was 9 July. In autumn 2009, the end of the open-water season, defined as the week when the total sea-ice cover is >20% (after 1 September), was 29 October. Freeze-up, defined as the first weekly date when the total sea-ice cover is >80% (after 1 September), was 5 November. It is apparent that sea-ice break-up in 2008 was influenced by the rise of LWd during the transition from spring to summer, whereas, in 2009, freeze-up may have been strongly influenced by the steep drop in LWd during the fall. This pattern is consistent with the pattern usually seen in the southeastern Beaufort Sea–Amundsen Gulf region where the timing of spring break-up is generally controlled by dynamic forcing, and freeze-up is, primarily, a thermodynamic event (Galley, Key, Barber, Hwang, & Ehn, 2008).

5 Conclusions

Hourly LWd was parameterized using measurements made during a composite year at various sites in the data sparse southeastern Beaufort Sea–Amundsen Gulf region, and atmospheric-column variables (PW, mean temperature, and mean vapour pressure) derived from microwave radiometric profiles observed at the same times and locations. All-sky LWd measurements were adjusted by a temporal fractional cloud cover derived from ceilometer measurements. A logarithmic relationship, which explained about 84% of the variance with a standard error of about 9%, was found between LWd and atmospheric-column water vapour expressed as either PW or mean vapour pressure. A linear correlation was found between LWd and T _mean. These results were not unexpected. Water vapour is a greenhouse gas (Bony et al., 2006), and PW, e _mean, and T _mean are related by the Ideal Gas Law. Longwave radiation emissions are related to temperature by a power law (Stefan-Boltzmann relationship), and water vapour saturation is an exponential function of temperature (Clausius-Clapeyron equation). Variance not explained by our parameterizations can be attributed to factors not taken into account, in particular, cloud composition, thickness, and cloud-base height (Key et al., 1996).

Application of these results should be limited to areas where LWd is similar to the study region (range = 130–360 W m⁻²). Because the parameterizations were derived from just one year of data, more years of data are required to confirm the relationships between LWd and atmospheric-column variables.

Both the annual cycle of monthly median PW and water vapour intrusions (maximum monthly PW) influence the magnitude of LWd over the western maritime Arctic, and the impact is enhanced by cloud cover. Water vapour intrusions impact LWd more dramatically in winter than in summer, thus these events play a more important role in the climate of the region in the cold season than in the warm season.

An increase in atmospheric moisture is expected to accompany future warming caused by increasing atmospheric concentrations of CO₂. An enhanced greenhouse effect caused by increased atmospheric water vapour will further increase climate warming in the Arctic (Bony et al., 2006; Schmidt, Ruedy, Miller, & Lacis, 2010). Changes in the atmospheric moisture budget are also likely to be accompanied by altered cloud cover which will further impact the magnitude of LWd (Curry, Schramm, Serreze, & Abert, 1995). Application of the parameterization of LWd with atmospheric-column moisture variables, which can be analyzed more accurately than surface variables in data sparse regions, may contribute to improved climate modelling of the western maritime Arctic.