Abstract

A simple alternative parameterization for predicting cloud fraction in the Community Climate System Model, version 3 (CCSM3) global climate model is presented. This formula, dubbed “freeezedry,” is designed to alleviate the bias of excessive low clouds during polar winter by reducing the cloud amount under very dry conditions. During winter, freezedry decreases the low cloud amount over the coldest regions in high latitudes by over 50% locally and more than 30% averaged across the Arctic. The cloud reduction causes an Arctic-wide drop of 15 W m−2 in surface cloud radiative forcing (CRF) during winter and about a 50% decrease in mean annual Arctic CRF. Consequently, wintertime surface temperatures fall by up to 4 K on land and 2–8 K over the Arctic Ocean, thus significantly reducing the model’s pronounced warm bias. Freezedry also affects CCSM3’s sensitivity to greenhouse forcing. In a transient-CO2 experiment, the model version with freezedry warms up to 20% less in the North Polar and South Polar regions (1.5- and 0.5-K-smaller warming, respectively). Paradoxically, the muted high-latitude response occurs despite a much larger increase in cloud amount with freezedry during nonsummer months (when clouds warm the surface), apparently because of the colder modern reference climate. While improving the polar climate simulation in CCSM3, freezedry has virtually no influence outside of very cold regions and has already been implemented in another climate model, the Global Environmental and Ecological Simulation of Ecological Systems, version 1 (GENESIS1). Furthermore, the simplicity of this parameterization allows it to be readily incorporated into other GCMs, many of which also suffer from excessive wintertime polar cloudiness.

1. Introduction

Understanding the behavior of Arctic clouds is especially timely and important, given the strong influence that polar clouds have on the surface energy budget, recent evidence of trends in Arctic cloud amount, the agreement among GCMs that future global warming will be amplified in the Arctic, and the traditional difficulties of climate models in simulating high-latitude clouds. Arctic clouds reduce wintertime cooling and summertime heating of the surface by tens of watts per square meter and exert a net warming influence in the annual mean (Curry et al. 1996; Schweiger and Key 1994). Clouds thus strongly shape high-latitude conditions, including climatically sensitive cryospheric processes such as the growth and decay of ice and snow cover. Satellite evidence points to seasonal changes in Arctic cloud amount during recent decades, as cloudiness during winter (spring) has been decreasing (increasing) in conjunction with measurable impacts on surface energy fluxes (Wang and Key 2003, 2005b; Schweiger 2004; Liu et al. 2007). Whether these cloud trends are an expression of anthropogenic forcing and whether they will continue is uncertain, but climate models consistently simulate an enhanced global warming response in the Arctic, regardless of differences in the sign and magnitude of their projected cloud changes (Holland and Bitz 2003; Solomon et al. 2007). Recognition of the Arctic as the most climatically sensitive region is supported both by paleoclimatic evidence and the existence of well-accepted amplifying processes such as the ice–albedo feedback. However, a major caveat of accurate climate model projections of Arctic amplification is the uncertain response of cloudiness, stemming from the historically unrealistic representation of polar clouds in the present climate system. This bias could distort the sign and magnitude of the simulated cloud feedback.

The goals of this paper are to describe a parameterization that improves the simulation of polar cloudiness under present conditions in one GCM [the National Center for Atmospheric Research’s (NCAR) Community Climate System Model, version 3 (CCSM3)], to suggest that this alternative treatment could be readily adopted into other models that are plagued with similar cloud biases, and to demonstrate that the new parameterization noticeably affects the simulated high-latitude climate sensitivity to greenhouse forcing. The difficulty in simulating polar cloudiness has been well documented for both global models (e.g., Bromwich et al. 1994; Randall et al. 1998; Walsh et al. 2002, 2005; Vavrus 2004) and regional climate models (Jones and Wyser 2004; Inoue et al. 2006). Representing clouds in high latitudes poses particular challenges because of the region’s complex radiative and turbulent processes that require alternative theoretical treatments and extremely high model resolution (Symon et al. 2005). The largest model errors typically occur during winter, a seasonal bias that persists in the latest generation of GCMs in the World Climate Research Programme’s (WCRP’s) Coupled Model Intercomparison Project phase 3 (CMIP3; Eisenman et al. 2007). Climate models commonly produce excessive wintertime Arctic clouds, particularly at low levels, an error attributed to an insufficient treatment of cloud processes unique to polar regions (Curry et al. 1996; Beesley and Moritz 1999; Beesley 2000; Jones and Wyser 2004). As a remedy, Walsh et al. (2005) recommended the “incorporation of new levels of sophistication of polar cloud formations into global models.”

The present study takes a step in this direction by describing the influence of a modification to the existing cloud parameterization in the CCSM3. This alternative formula was originally incorporated into the Global Environmental and Ecological Simulation of Ecological Systems, version 1 (GENESIS1) climate model as an alternative to the traditional approach of assigning cloud amount as a function of prognostic relative humidity (Slingo 1987) and produced a credible annual cycle of Arctic cloudiness (Thompson and Pollard 1995). Here we describe the alternative formulation and its implementation into CCSM3 (section 2), the impact it has on simulated Arctic clouds and temperature in the present climate (section 3), and its influence on high-latitude climate sensitivity in a 2 × CO2 experiment (section 4). The paper closes with a discussion of the merits and limitations of the new scheme, possibilities for improvement, and the relevance of this approach.

2. Description of CCSM3 and new cloud parameterization

The CCSM3 is a fully coupled global climate model of the atmosphere, ocean, sea ice, and land systems, whose complete description is given in Collins et al. (2006a) and the special issue of Journal of Climate (2006, Vol. 19, No. 11). The atmospheric component is the Community Atmosphere Model, version 3 (CAM3), described in detail by Collins et al. (2006b), and the land component is the Community Land Model, version 3 (CLM3; Bonan and Levis 2006). In this study, the horizontal model resolution of the atmosphere and land is T42 (approximately 2.8° × 2.8°), while the ocean and sea ice components employ a nominal horizontal resolution of 1° that maximizes near the equator and tapers toward the poles. The vertical dimension of the ocean is treated using a depth (z) coordinate with 25 levels. The ocean model [Parallel Ocean Program (POP) version 1.4.3] and sea ice model [NCAR’s Community System Model (CSM) Sea Ice Model (CSIM)] include dynamical processes and are described fully in Smith and Gent (2002) and Briegleb et al. (2004). The atmosphere contains 26 levels in a hybrid-sigma pressure coordinate system. The land model contains 10 subsurface soil layers and exchanges energy, mass, and momentum with the atmosphere, but the seasonally varying vegetation composition at each terrestrial grid point is prescribed.

A full description of CAM3’s treatment of clouds is given in Collins et al. (2006b) and Boville et al. (2006), so only a brief summary is provided here. Clouds are categorized as either convective or stratiform and are calculated separately at three levels (low, middle, and high). Condensate varies between ice and liquid as a quadratic function of temperature, using threshold temperatures of 243 and 263 K, with different settling velocities for liquid- and ice phase as functions of particle size characterized by the effective radius. The model uses the prognostic cloud water parameterization of Rasch and Kristjánsson (1998) that was updated by Zhang et al. (2003). CAM3 includes the radiative effects of aerosols in the calculation of shortwave fluxes and heating rates, based on an aerosol assimilation for the period 1995–2000. The model employs a standard maximum-random cloud overlap scheme (Collins 2001) and separate parameterizations for shallow (Hack 1994) and deep (Zhang and McFarlane 1995) convection. Cloud fraction is determined diagnostically for convective and stratiform clouds, using separate calculations for deep and shallow convection. A marine stratus parameterization based on Klein and Hartmann (1993) gives a lower bound on stratiform cloud fraction for cells that are at least half ocean. This condition is only active over limited regions off the west coasts of continents.

In this study, the model’s formula for calculating the fraction of low-level stratiform clouds (where air pressure is 750 hPa or higher) is modified in an attempt to alleviate the excessive polar cloudiness simulated during much of the year (section 3). The model’s cloud bias is identified relative to a suite of surface-based and satellite-derived observations. The surface data include measurements from Huschke (1969), Hahn et al. (1995), Makshtas et al. (1999), and Comprehensive Ocean–Atmosphere Data Set (COADS) values compiled by Serreze and Barry (2005). The remotely sensed observations are based on eight datasets, six of which are compiled and described by C. Stubenrauch et al. (2007, personal communication)—International Satellite Cloud Climatology Project (ISCCP), Television and Infrared Observation Satellite (TIROS) Operational Vertical Sounder Pathfinder B (TOVS-Path B), High Resolution Infrared Radiation Sounder (HIRS), Moderate Resolution Imaging Spectroradiometer (MODIS)-Science Team (ST), MODIS-CE [Clouds and the Earth’s Radiant Energy System (CERES)], and Advanced Very High Resolution Radiometer Pathfinder Atmospheres (AVHRR-PATMOS)—and the other two by Wang and Key [2005a; the extended AVHRR Polar Pathfinder (APP) dataset (APP-x)] and Kato et al. (2006; CERES).

CAM3 currently assigns the low cloud fraction ( f ) to be a function of the gridbox average relative humidity (RH):

 
formula

where RHMIN is the minimum relative humidity—expressed as a fraction—at which clouds form (0.8 over land, 0.9 elsewhere). While this type of representation may be effective in regions for which it was designed, studies have shown that this type of formula is not well suited for high latitudes, particularly the extremely cold and dry atmospheric conditions typical of polar winter. Beesley (2000) notes that relative humidity in the Arctic varies little during the year, thus constraining a parameterization such as (1) to generate a muted annual cycle of cloudiness. Xu and Randall (1996) and Randall et al. (1996) used observational data from the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) and the Atlantic Stratocumulus Transition Experiment (ASTEX) to propose a reduction in predicted cloud amount when air is very dry (low cloud water content). The GENESIS1 climate model (Thompson and Pollard 1995) adopted a similar approach, reducing the relative humidity-derived low cloud fraction ( f ) when the gridbox mean specific humidity (q) falls below a threshold value (0.003 kg kg−1):

 
formula

Thus, the originally calculated cloud fraction is only adjusted under very dry atmospheric conditions, in which case the low cloud amount is reduced by as much as 85% of its relative humidity–based value. The resulting reduction in cloud coverage as a function of q is similar to the more complex empirical formula derived by Xu and Randall (1996). Although this adjustment is applied globally in the low cloud subroutine of the model, it primarily affects polar regions because the threshold value of specific humidity in (2) is realized only in extremely cold conditions. For this reason and because the ensuing reduction of f acts to dry out the atmosphere of cloud, this modification to CAM3’s cloud parameterization is dubbed “freezedry.” Although freezedry is applied only to the low cloud section of CAM3, this adjustment to the cloud fraction should also be applicable to other GCMs that do not divide clouds into three layers, as long as the threshold pressure level (750 hPa in CAM3) is specified in the code.

This parameterization has a solid physical basis supported by field data. Jones and Wyser (2004) provide plausible reasons why relative humidity–based cloud fraction schemes are not appropriate for polar regions. They argue that the assumption of partial cloudiness under subsaturated conditions averaged over a grid box may not be valid in the extremely stable Arctic atmosphere during winter. In these stable environments, there should be very little subgrid-scale heterogeneity and thus less cloudiness for a given relative humidity than in the more turbulent environments for which a model’s cloud parameterization was derived. Furthermore, these authors contend that clouds are probably limited during winter by the extremely low density of condensation nuclei in polar environments. Adopting the cloud fraction formula of Xu and Randall (1996), which is similar to (2) but derived directly from field measurements, Jones and Wyser documented that the altered equation produced an improved fit to Surface Heat Budget of the Arctic Ocean (SHEBA) data in terms of winter cloud amount and downwelling longwave radiation at very low temperatures in the Rossby Centre regional climate model.

The simulations described here use modern (year 1990) radiative forcing for the control and FREEZEDRY experiments, which apply the freezdry adjustments. The control simulation is described in detail in Collins et al. (2006a) and is used as the initial conditions for FREEZEDRY (starting from year 900 of the control run). The FREEZEDRY simulation was run 40 yr, consisting of a 30-yr adjustment interval and the remainder for climatological statistics presented here. The rapid convergence is attributable to the nearly instantaneous response of the clouds to the freezedry adjustment and the fact that the clouds respond primarily during the coldest months only. Sensitivity tests have also been performed with other configurations of the model, including a slab ocean and a finite-volume dynamical core, with very similar results. Experiment FREEZEDRYCO2 utilizes (2) under a 1% yr−1 increase in CO2 for 80 yr, the final 20 of which are used as a climatology to compare with the standard CCSM3 transient-CO2 simulation (without freezedry) around the time of CO2 doubling at year 70. FREEZEDRYCO2 was initialized from the end of the modern FREEZEDRY simulation.

3. Results

a. Modern simulation

One of the most vexing and persistent biases in the CCSM3 is its overproduction of low stratus clouds over the Arctic during winter (a similar but less severe bias also occurs over austral high latitudes). This bias was present in previous versions of the model (CCSM2, CSM) and has been a common GCM weakness over the years, even among recent models in the CMIP3 dataset. Similarly problematic in both the finite-volume and Eulerian dynamical cores (regardless of resolution), the excessive wintertime cloudiness is most pronounced over mid–high-latitude land and the Arctic sea ice pack (Fig. 1). This cloud bias is strongly associated with a pronounced warm bias in the extratropics, suggestive of a causal link due to the strong warming influence of clouds during polar winter (Schweiger and Key 1994). Compared with the observational dataset of Warren et al. (1986, 1988), low cloud fraction is more than 0.5 too large over a wide region of northern Asia and more than 0.3 too great in parts of northern North America. By contrast, the combined mid- and high-level cloud coverage is fairly realistic, as inferred from the much smaller biases in total cloud amount in Fig. 1. In addition, the excessive low cloud cover is associated with approximately 50% too much wintertime precipitation in the Arctic (not shown) (Xie and Arkin 1996), thus hindering the simulated hydrologic cycle in polar regions. In fact, the Arctic climate in the T85 version of CCSM3 is among the wettest of all GCMs documented in a recent multimodel assessment (Kattsov et al. 2007).

Fig. 1.

Boreal winter [December–February (DJF)] biases in total cloud amount (%), f (%), and surface temperature (Tsfc; K) in CCSM3’s T85 control run. Biases are referenced to the Warren et al. (1986, 1988) cloud climatology and the Legates and Willmott (1990) temperature observations.

Fig. 1.

Boreal winter [December–February (DJF)] biases in total cloud amount (%), f (%), and surface temperature (Tsfc; K) in CCSM3’s T85 control run. Biases are referenced to the Warren et al. (1986, 1988) cloud climatology and the Legates and Willmott (1990) temperature observations.

The seasonal dependence of the model’s Arctic cloud bias (land and ocean, 70°–90°N; Fig. 2) shows that the simulated low cloud cover is much more realistic during summer but greatly excessive at other times, especially during the coldest months. The observed values of total cloud amount are a composite estimate from a dataset equally weighted by the 4 surface-based and 8 satellite-derived datasets described in section 2, and the monthly low cloud observations are from Huschke (1969). The Huschke data are a 63-month collection of surface observations from ice drifting stations over the central Arctic Ocean and 17 coastal sites in the central Eurasian and Canadian Arctic (poleward of 70°N). The accuracy of the estimated vertical distribution of the clouds was enhanced by calibrating simultaneous aerial and ground-based observations. However, this dataset is based on only 5 complete years, was conducted during a somewhat colder climate interval (late 1950s), and is subject to the constraint that surface-based cloud observations are biased low during dark conditions such as polar night (Hahn et al. 1995). For comparison, Fig. 2a also displays seasonal averages of low cloud amounts based on the Warren et al. (1988) climatology that correspond to the domain used by Huschke. The two datasets compare very well except during autumn, when the Warren et al. observations indicate considerably more cloudiness.

Fig. 2.

Simulated (CCSM3) and observed annual cycle of low and total cloud amount (%) averaged over the Arctic. Low cloud observations are from Huschke (1969). Observed total cloud is a composite estimate from a dataset equally weighted by four surface-based observations (Huschke 1969; Hahn et al. 1995; Makshtas et al. 1999; and COADS measurements compiled by Serreze and Barry 2005) and eight satellite-based observations: six described by C. Stubenrauch et al. (2007, personal communication; ISCCP, TOVS-Path B, HIRS, MODIS-ST, MODIS-CE, and AVHRR-PATMOS) and two supplemental (APP-x, Wang and Key 2005a; CERES, Kato et al. 2006). The bars on the observed total cloud amount represent the upper and lower quartile of values among the 12 datasets.

Fig. 2.

Simulated (CCSM3) and observed annual cycle of low and total cloud amount (%) averaged over the Arctic. Low cloud observations are from Huschke (1969). Observed total cloud is a composite estimate from a dataset equally weighted by four surface-based observations (Huschke 1969; Hahn et al. 1995; Makshtas et al. 1999; and COADS measurements compiled by Serreze and Barry 2005) and eight satellite-based observations: six described by C. Stubenrauch et al. (2007, personal communication; ISCCP, TOVS-Path B, HIRS, MODIS-ST, MODIS-CE, and AVHRR-PATMOS) and two supplemental (APP-x, Wang and Key 2005a; CERES, Kato et al. 2006). The bars on the observed total cloud amount represent the upper and lower quartile of values among the 12 datasets.

Compared with the relatively weak annual cycle of simulated clouds, the observed seasonal variation in low cloud amount (and the model’s bias) is thus largely a function of temperature. This point was emphasized by Beesley and Moritz (1999), who attributed it to the temperature dependence of ice-phase microphysical processes that favor more rapid fallout of cloud ice condensate (primarily a wintertime phenomenon) over liquid condensate. Although subsequent analysis of measurements from SHEBA has identified that a surprising amount of supercooled liquid droplets can exist at very low temperatures, the proportion of cloud ice still exceeds cloud liquid during polar winter (Intrieri et al. 2002). The simulation of low clouds in CCSM3 is a cold-season problem that affects both polar regions during their respective winters, but it is not a systematic bias plaguing other parts of the world (Fig. 1). This result supports the contention that the formula for calculating cloud fraction as a function of relative humidity (1) is much better suited for relatively warm regions, whereas it requires adjusting in very cold climates. We also see from Figs. 1 and 2 that a similar but less pronounced cold-season bias exists in the total cloud fraction in the Arctic and that most of the vertically integrated cloud amount in the model consists of low clouds.

The wintertime cloud bias described above is not unique to CCSM3 but rather is a common problem that affects many GCMs, based on an examination of the cloud characteristics in 20 climate models from the CMIP3 archive used in the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4; Solomon et al. 2007). A typical GCM very similarly overproduces total Arctic cloudiness during the coldest months—vertically integrated low cloud amount was not available—but simulates summertime cloudiness remarkably well (Fig. 3). This seasonal variation in the accuracy of simulated cloud coverage is also apparent in terms of the models’ precision: the intermodel spread of cloud amount is up to factor of 3 larger during winter than summer. Thus, the oft-cited difficulty of climate models in reproducing Arctic cloudiness (e.g., Randall et al. 1998; Walsh et al. 2005) should be understood as being primarily a seasonal problem of capturing the behavior of clouds during the coldest periods of the year.

Fig. 3.

Simulated and observed annual cycle of total cloud amount (%) averaged over the Arctic. Modeled values are the 20-GCM ensemble mean from the CMIP3 dataset and the corresponding intermodel std dev. Observations are the same as in Fig. 2.

Fig. 3.

Simulated and observed annual cycle of total cloud amount (%) averaged over the Arctic. Modeled values are the 20-GCM ensemble mean from the CMIP3 dataset and the corresponding intermodel std dev. Observations are the same as in Fig. 2.

Incorporating the freezedry parameterization into CAM3 produces a large change in low cloud amount and surface temperature in both polar regions (Fig. 4). The most striking effect occurs during the respective winter season, in accordance with the activation of the freezedry adjustment to the calculated cloud fraction only under extremely dry (and cold) conditions (2). Low cloud fraction decreases by more than 0.5 (expressed in Fig. 4 as 50%) over the coldest regions of the Northern Hemisphere, located in Siberia, eastern Canada, and the Arctic Ocean, while widespread reductions of over 0.4 are seen across much of the Arctic. Notice that freezedry has little or no impact over ice-free maritime regions, where the atmosphere is too moist to be affected. Accompanying the large drop in low cloud amount is a pronounced cooling of the surface during boreal winter. Temperatures fall by up to 4 K on land in roughly the same areas where clouds diminish the most, while even larger cooling occurs across Arctic sea ice, which experiences temperature reductions of at least 2 K in most places and up to 8 K in the interior ice pack. Annually averaged over the Arctic (70°–90°N), FREEZEDRY is 1.2 K colder, 4% drier, and produces somewhat greater sea ice: 0.07 m thicker (+3.5%) and 2.5% higher concentration. Very similar decreases in low clouds and temperature are triggered over ice-covered portions of the Southern Ocean and parts of Antarctica during austral winter. Notice, however, that cloud reductions do not occur over highly elevated regions of Antarctica, because the freezedry adjustment only affects the formation of low clouds, which are designated in CAM3 as 750 hPa to the surface. The parameterization may need refinement to capture “low” clouds occurring above highly elevated terrain here and over the Greenland Plateau, where a similarly muted response is seen.

Fig. 4.

Changes in f (%) and Tsfc (K) produced in FREEZEDRY during (a), (b) DJF and (c), (d) June–August (JJA).

Fig. 4.

Changes in f (%) and Tsfc (K) produced in FREEZEDRY during (a), (b) DJF and (c), (d) June–August (JJA).

Freezedry’s targeting of low clouds is especially evident in a vertical cross section of cloud concentration (Fig. 5), which shows a sharp vertical break in the cloud decrease at 750 hPa and the near elimination of the prevalent wintertime cloud deck at the lowest levels of the Arctic that exists in the CCSM3 control run. Also apparent is the virtual lack of cloud changes at any height outside of high latitudes; even though the freezedry parameterization is applied globally to the lower troposphere in the model, regions where the specific humidity exceeds the threshold value of 0.003 kg kg−1 are unaffected.

Fig. 5.

Vertical cross section of cloud fraction during boreal winter in (a) FREEZEDRY, (b) control, and (c) FREEZEDRY − control.

Fig. 5.

Vertical cross section of cloud fraction during boreal winter in (a) FREEZEDRY, (b) control, and (c) FREEZEDRY − control.

In a similar vein, the annual cycle of cloudiness in the Arctic is largely unchanged in FREEZEDRY during the relatively warm and moist polar summer, whereas the coverage of low clouds plummets in other seasons (Fig. 6). As expected, this cloud reduction is mostly a function of temperature, with the most pronounced decreases during winter. The freezedry adjustment allows the model to produce a realistic annual cycle of low clouds, though the amounts are still somewhat larger than the observational estimates of Huschke (1969) used here. FREEZEDRY’s reduction of total cloud amount overshoots observed values from November to April, but the magnitude of the errors during this period is about the same as those in the CCSM3 control run.

Fig. 6.

As in Fig. 2 but with FREEZEDRY values overlain.

Fig. 6.

As in Fig. 2 but with FREEZEDRY values overlain.

The cloud changes induced by freezedry strongly affect the heating influence of Arctic clouds, expressed as the cloud radiative forcing (CRF), especially at the surface (Fig. 7). The reduced cloudiness in FREEZEDRY causes an increase in shortwave CRF in all months with daylight, both at the surface and top of atmosphere (TOA). However, these cloud losses also cause the longwave CRF to fall at both levels throughout the year. At the TOA, this competition balances annually, as the net decrease in CRF during the darkest half year (October–March) offsets the net gain during the brightest six months (time-mean change in net CRF = −0.01 W m−2). The surface, however, is more sensitive to the longwave impact of reduced cloud cover. Drops in longwave CRF at the surface across the Arctic average 15 W m−2 during winter in FREEZEDRY (with local monthly anomalies down to −30 W m−2) and reach a maximum of 20 W m−2 in April. Even though the longwave differences rebound during summer to values like those at the TOA, the much larger negative longwave changes during all other seasons dominate the overall response. Freezedry causes a decreased mean annual net CRF at the surface (9.4 W m−2) that is almost half the magnitude (48%) of the net CRF in the control run (19.7 W m−2).

Fig. 7.

Change in annual cycle of Arctic CRF (70°–90°N) at surface and top of atmosphere in FREEZEDRY (W m−2). Plotted are shortwave CRF (solid circles), longwave CRF (open circles), and net CRF (open squares).

Fig. 7.

Change in annual cycle of Arctic CRF (70°–90°N) at surface and top of atmosphere in FREEZEDRY (W m−2). Plotted are shortwave CRF (solid circles), longwave CRF (open circles), and net CRF (open squares).

b. Effect on climate sensitivity

Not only does freezedry greatly affect the polar cloud simulation under present climatic conditions, it also exerts a strong influence on high-latitude climate sensitivity, or more precisely the transient climate response of CCSM3 under greenhouse forcing. Starting from the end of the modern FREEZEDRY run, the model was integrated with freezedry for 80 yr using a 1% yr−1 increase in CO2 concentration. Because this experiment with ramped-up greenhouse forcing (FREEZEDRYCO2) reaches a doubled CO2 concentration after 70 yr, the solution averaged a decade on either side (years 60–80) was used to construct a 2 × CO2 climatology to compare with the FREEZEDRY simulation run under modern radiative forcing. To increase confidence in the results, a second ensemble member was also created by using the same procedure as for FREEZEDRYCO2 but starting from a slightly different point at the end of the FREEZEDRY control run. Because this second simulation bore such a strong resemblance to the first one, only the results from the first greenhouse experiment are presented here.

In CCSM3’s standard transient-2 × CO2 simulation (without freezedry), the model produces more low clouds annually in the tropics and the Arctic (Fig. 8a) as the climate warms. Conversely, cloud amount does not increase in the high latitudes of the Southern Hemisphere, where deep-ocean mixing causes more modest temperature increases. Whereas the tropical cloud response is similar in FREEZEDRYCO2, low-level polar clouds show bigger differences in both hemispheres. Over the Southern Ocean, low cloud amount increases by the same magnitude as the peak tropical rise, while the gain in cloud cover is greatly enhanced in the Arctic and even extends into midlatitudes: positive anomalies of low cloud occur poleward of 60°N in the standard 2 × CO2 simulation but spread to around 50°N in FREEZEDRYCO2. In addition, the low cloud decreases across the midlatitudes of both hemispheres in the standard run are less negative in FREEZEDRYCO2. Freezedry causes the largest enhancement in Arctic low clouds during the coldest months, whereas summertime anomalies are only slightly greater than those in the standard CCSM3 simulation (Fig. 8b).

Fig. 8.

Changes in Arctic low clouds and temperature (70°–90°N) in the transient greenhouse simulations: FREEZEDRYCO2 (solid circles) and CCSM3’s std 2 × CO2 run (open circles). Shown are changes in (a) annual low cloud fraction, (b) annual cycle of f, (c) annual Tsfc, and (d) annual cycle of Tsfc.

Fig. 8.

Changes in Arctic low clouds and temperature (70°–90°N) in the transient greenhouse simulations: FREEZEDRYCO2 (solid circles) and CCSM3’s std 2 × CO2 run (open circles). Shown are changes in (a) annual low cloud fraction, (b) annual cycle of f, (c) annual Tsfc, and (d) annual cycle of Tsfc.

The enhanced low cloudiness induced by freezedry under greenhouse forcing stems largely from how the parameterization changes the reference cloud amount under modern radiative conditions (FREEZEDRY versus control) and the threshold nature of the cloud adjustment as a function of specific humidity (2). Starting with much less low cloud in FREEZEDRY compared to the control run (Fig. 4) allows a larger potential increase in cloudiness as the climate warms. In addition, where greenhouse warming causes q to increase from below to above the 0.003 kg kg−1 threshold for triggering freezedry, the cloud fraction is assigned to be the larger function of relative humidity only and no longer adjusted downward.

Despite the more positive low cloud change in FREEZEDRYCO2 over polar regions, where clouds generally act as a warming mechanism, the annual warming is muted in the high latitudes of both hemispheres (Fig. 8c). The temperature increase is up to 0.5 K smaller near the South Pole and 1.5 K around the North Pole, both of which represent about a 20% reduction in the transient climate response (TCR). This tempered greenhouse warming is especially odd, given that the enhanced cloud increases in FREEZEDRYCO2 are greatest during the coldest months (Fig. 8d), when the cloud radiative forcing is most positive (Schweiger and Key 1994). The warming with freezedry is actually reduced most strongly from November to February, nearly the coldest four-month interval of the year. While it is beyond the scope of this study to provide an in-depth explanation for this difference in the TCR, the most plausible reason is that the modern climate in both polar regions is considerably colder and icier in FREEZEDRY than the control (Fig. 9). As a result, the radiative heating from rising CO2 is able to generate fewer positive feedbacks from sea ice melt off in FREEZEDRYCO2 than in the standard 2 × CO2 simulation. This expectation is in line with results of previous studies indicating that models with relatively thin ice in their control simulations produce more melt off and thus greater warming under greenhouse forcing (Rind et al. 1995, 1997; Holland and Bitz 2003). This reasoning is also consistent with a related explanation that differences in atmospheric circulation changes between the two greenhouse simulations are responsible for the muted polar warming in FREEZEDRYCO2. More sea ice remains along the Atlantic margin with freezedry, thereby inducing a relative trough in the upper air in FREEZEDRYCO2 that is associated with a more anticyclonic circulation anomaly over the Arctic Ocean and thus less Atlantic inflow reaching the interior Arctic (S. Vavrus 2007, unpublished manuscript; Fig. 10).

Fig. 9.

Differences in mean annual Tsfc and sea ice concentration between FREEZEDRY and the CCSM3 modern control simulation.

Fig. 9.

Differences in mean annual Tsfc and sea ice concentration between FREEZEDRY and the CCSM3 modern control simulation.

Fig. 10.

Differences in the circulation response induced by freezedry under transient-CO2 forcing. Maps show the mean annual changes in (a) 500-hPa geopotential heights (m) and (b) sea level pressure (hPa) using freezedry relative to the changes using the std model version: (FREEZEDRYCO2 − FREEZEDRY) − (std 2 × CO2 − control). Negative values are plotted as dashed lines.

Fig. 10.

Differences in the circulation response induced by freezedry under transient-CO2 forcing. Maps show the mean annual changes in (a) 500-hPa geopotential heights (m) and (b) sea level pressure (hPa) using freezedry relative to the changes using the std model version: (FREEZEDRYCO2 − FREEZEDRY) − (std 2 × CO2 − control). Negative values are plotted as dashed lines.

4. Discussion and conclusions

In addition to improving the simulation of Arctic clouds and temperature in this GCM, the freezedry parameterization is suitable for use in other climate models that are often plagued by excessive wintertime cloudiness in polar regions. The alteration to the predicted cloud fraction (2) is an extremely simple, one-line modification to the code that produces virtually no increase to the computation time and has a solid physical basis (section 2). The parameterization successfully targets the (low) cloud bias during polar winter in both hemispheres, while not noticeably affecting the model’s climate in the tropics and midlatitudes. In addition, similar forms of this parameterization have been successfully adopted into other global and regional climate models (Thompson and Pollard 1995; Jones and Wyser 2004) and shown to reproduce Arctic clouds better than any other RCM in a recent intermodel comparison (Inoue et al. 2006).

Aside from these favorable characteristics, a number of caveats are warranted concerning freezedry and the results of this study. First, this parameterization is not a panacea for the simulation of polar clouds, which may still suffer from other microphysical flaws [e.g., cloud water biases affecting reflectivity (Gorodetskaya et al. 2008) and the optical properties of thin ice clouds] even if their cloud abundance is realistic. For example, Girard and Stefanof (2007) used a regional climate model to show that accounting for acidic aerosols can help to dehydrate the polar atmosphere by promoting larger ice crystals, leading to a qualitatively similar decrease in winter cloudiness and surface temperature as freezedry produces. Second, if an adjustment is required to the predicted cloud fraction, modelers should strive for a relation that is not prescribed to certain heights (low-level in this case) and that has a clearer physical justification for a threshold; the specific humidity bound of 0.003 kg kg−1 used here is effective in this GCM but is arbitrary and highly tunable. Although the parameter values in (2) work well for this model and the GENESIS1 GCM, as well as a similar adjustment formula in the Rossby Centre regional model, the sensitivity of other models may differ and could require somewhat altered parameter choices. The low cloud restriction currently hinders application of freezedry over the highly elevated terrain of Greenland and Antarctica, although a refinement could specify that clouds be adjusted within a specified pressure range above the surface, rather than below a prescribed pressure value. Third, another major contributor to the overproduction of polar cloudiness in this model and others is probably excessive poleward moisture transport, related to CAM3’s bias of anomalously low pressure in the Arctic during winter. An atmosphere-only simulation with CAM3 generated too much precipitable water flux into the Arctic, a bias significantly correlated on monthly time scales with the modeled Arctic cloud amount. Excess moisture convergence and precipitation in high latitudes has been a long-standing GCM bias that is especially strong during the cold season and at lower model resolution (Walsh et al. 1998, 2002; Dixon et al. 2003). CCSM3 is no exception: the polar low cloud bias is more acute at coarser resolutions, but it remains pronounced even at the model’s higher standard resolution of T85.

Finally, a serious qualification of this study surrounds uncertainty in present-day cloud amount. Numerous investigations have identified and discussed differences between surface-based and satellite-derived estimates of cloud amount in high latitudes (Wilson et al. 1993; Curry et al. 1996; Schweiger et al. 1999; Town et al. 2007), with even a universally accepted definition of a polar cloud still elusive (Serreze and Barry 2005). In general, satellites report higher cloud amounts than surface measurements during polar night because of the difficulties of human observers detecting clouds in darkness (Hahn et al. 1995). However, detecting clouds remotely in polar regions is also problematic (Rossow and Schiffer 1999) and depends on a somewhat subjective optical depth threshold at which the existence of a cloud is defined (Wyser and Jones 2005). Given the focus of this study on low clouds, their masking by mid- and high-level clouds adds an additional obstacle in obtaining accurate measurements from satellite. Because of these various complicating factors, the observational estimates chosen here are simply a blend of available data measured from above and below clouds, without an attempt to filter based on their quality. The recent availability of global cloud-radar observations from CloudSat, launched in April of 2006, provides for a much more unambiguous and quantitative description of polar clouds. This includes vertical profiles of cloud structure–type and ice and liquid water content in both polar day and night conditions (Stephens et al. 2002).

This effort to improve simulations of polar clouds is relevant and timely for a number of reasons. Aside from the effect on high-latitude climate, the improvement in CCSM3 of wintertime polar cloudiness and thus surface temperature should produce more realistic interactions with midlatitudes, such as simulated extreme cold-air outbreaks (Vavrus et al. 2006). Also, Arctic clouds have shown changes recently, both in seasonal cloud amount and composition (Schweiger 2004; Wang and Key 2003, 2005b; Stone et al. 2005; Liu et al. 2007), but the cause of these trends is still under investigation. The summer of 2007 has set a record for minimum Arctic sea ice extent, and the rapid rate of melting has been linked to anomalously low cloud amount over the Arctic Ocean (Kay et al. 2008). Moreover, the future behavior of polar clouds under the influence of anthropogenic drivers such as enhanced greenhouse gases and aerosol concentration is uncertain and model dependent, based on an analysis of a suite of GCMs in the CMIP3 dataset used in the IPCC AR4 (not shown). Although intermodel spread in projected Arctic cloud amount is a characteristic of this ensemble, a recent assessment has demonstrated that the response of a model’s polar cloud amount to greenhouse warming is similar to its seasonal response to warming within the present-day annual cycle (S. Vavrus et al. 2008, unpublished manuscript). This result highlights the importance of creating more credible simulations of high-latitude clouds under modern climatic conditions. The freezedry parameterization has recently been implemented into the interim version of the next NCAR model (CCSM3.5) and appears to be improving the model’s Arctic cloud simulation.

Acknowledgments

This work has been supported by National Science Foundation Awards OPP-0327664, ARC-0628910, and DE-FG02-06ER64297 (Small Grant for Exploratory Research) jointly funded by DOE and NSF as part of the DOE Office of Science SciDAC-2 initiative. Computing support from the Climate Simulation Lab at NCAR is greatly appreciated. Acknowledgment is also given to the modeling groups, the Program for Climate Model Diagnosis and Intercomparison (PCMDI) and the WCRP’s Working Group on Coupled Modelling (WGCM) for their roles in making available the WCRP CMIP3 multi-model dataset. Support of this dataset is provided by the Office of Science, U.S. Department of Energy. The assistance of John Dyreby in the processing of the CMIP3 output was very helpful, as were constructive comments from Axel Schweiger and Jennifer Francis.

REFERENCES

REFERENCES
Beesley
,
J. A.
,
2000
:
Estimating the effect of clouds on the Arctic surface energy budget.
J. Geophys. Res.
,
105
,
D8
.
10103
10117
.
Beesley
,
J. A.
, and
R. E.
Moritz
,
1999
:
Toward an explanation of the annual cycle of cloudiness over the Arctic Ocean.
J. Climate
,
12
,
395
415
.
Bonan
,
G. B.
, and
S.
Levis
,
2006
:
Evaluating aspects of the Community Land and Atmosphere Models (CLM3 and CAM) using a dynamic global vegetation model.
J. Climate
,
19
,
2290
2301
.
Boville
,
B. A.
,
P. J.
Rasch
,
J. J.
Hack
, and
J. R.
McCaa
,
2006
:
Representation of clouds and precipitation processes in the Community Atmosphere Model (CAM3).
J. Climate
,
19
,
2184
2198
.
Briegleb
,
B. P.
,
C. M.
Bitz
,
E. C.
Hunke
,
W. H.
Lipscomb
,
M. M.
Holland
,
J. L.
Schramm
, and
R. E.
Moritz
,
2004
:
Scientific description of the sea ice component of the Community Climate System Model, Version Three. National Center for Atmospheric Research Tech. Rep. NCAR/TN-463+STR, 78 pp
.
Bromwich
,
D. H.
,
R-Y.
Tzeng
, and
T. R.
Parish
,
1994
:
Simulation of the modern Arctic climate by the NCAR CCM1.
J. Climate
,
7
,
1050
1069
.
Collins
,
W. D.
,
2001
:
Parameterization of generalized cloud overlap for radiative calculations in general circulation models.
J. Atmos. Sci.
,
58
,
3224
3242
.
Collins
,
W. D.
, and
Coauthors
,
2006a
:
The Community Climate System Model version 3 (CCSM3).
J. Climate
,
19
,
2122
2143
.
Collins
,
W. D.
, and
Coauthors
,
2006b
:
The formulation and atmospheric simulation of the Community Atmosphere Model version 3 (CAM3).
J. Climate
,
19
,
2144
2161
.
Curry
,
J. A.
,
W. B.
Rossow
,
D.
Randall
, and
J. L.
Schramm
,
1996
:
Overview of Arctic cloud and radiation characteristics.
J. Climate
,
9
,
1731
1764
.
Dixon
,
K. W.
,
T. L.
Delworth
,
T. R.
Knutson
,
M. J.
Spelman
, and
R. J.
Stouffer
,
2003
:
A comparison of climate change simulations produced by two GFDL coupled models.
Global Planet. Change
,
37
,
81
102
.
Eisenman
,
I.
,
N.
Untersteiner
, and
J. S.
Wettlaufer
,
2007
:
On the reliability of simulated Arctic sea ice in global climate models.
Geophys. Res. Lett.
,
34
.
L10501, doi:10.1029/2007GL029914
.
Girard
,
E.
, and
A.
Stefanof
,
2007
:
Assessment of the dehydration-greehouse feedback over the Arctic during February 1990.
Int. J. Climatol.
,
27
,
1047
1058
.
Gorodetskaya
,
I. V.
,
L-B.
Tremblay
,
B.
Liepert
,
M. A.
Cane
, and
R. I.
Cullather
,
2008
:
The influence of cloud and surface properties on the Arctic Ocean shortwave radiation in coupled models.
J. Climate
,
21
,
866
882
.
Hack
,
J. J.
,
1994
:
Parameterization of moist convection in the NCAR Community Climate Model, CCM2.
J. Geophys. Res.
,
99
,
D3
.
5551
5568
.
Hahn
,
C. J.
,
S. G.
Warren
, and
J.
London
,
1995
:
The effect of moonlight on observation of cloud cover at night, and application to cloud climatology.
J. Climate
,
8
,
1429
1446
.
Holland
,
M. M.
, and
C. M.
Bitz
,
2003
:
Polar amplification of climate change in the coupled model intercomparison project.
Climate Dyn.
,
21
,
221
232
.
Huschke
,
R. E.
,
1969
:
Arctic cloud statistics from “air-calibrated” surface weather observations. Rand Corporation Memo. RM-6173-PR, 79 pp
.
Inoue
,
J.
,
J.
Liu
,
J. O.
Pinto
, and
J. A.
Curry
,
2006
:
Intercomparison of Arctic regional climate models: Modeling clouds and radiation for SHEBA in May 1998.
J. Climate
,
19
,
4167
4178
.
Intrieri
,
J. M.
,
C. W.
Fairall
,
M. D.
Shupe
,
P. O. G.
Persson
,
E. L.
Andreas
,
P. S.
Guest
, and
R. M.
Moritz
,
2002
:
An annual cycle of Arctic surface cloud forcing at SHEBA.
J. Geophys. Res.
,
107
.
8039, doi:10.1029/2000JC000439
.
Jones
,
C.
, and
K.
Wyser
,
2004
:
The Rossby Centre Regional Atmospheric Climate Model Part II: Application to the Arctic climate.
Ambio
,
33
,
211
220
.
Kato
,
S.
,
N. G.
Loeb
,
P.
Minnis
,
J. A.
Francis
,
T. P.
Charlock
,
D. A.
Rutan
,
E. E.
Clothiaux
, and
S.
Sun-Mack
,
2006
:
Seasonal and interannual variations of top-of-atmosphere irradiance and cloud cover over polar regions derived from the CERES data set.
Geophys. Res. Lett.
,
33
.
L19804, doi:10.1029/2006GL026685
.
Kattsov
,
V. M.
,
J. E.
Walsh
,
W. L.
Chapman
,
V. A.
Govorkova
, and
T. V.
Pavlova
,
2007
:
Simulation and projection of Arctic freshwater budget components by the IPCC AR4 global climate models.
J. Hydrometeor.
,
8
,
571
589
.
Kay
,
J. E.
,
T.
L’Ecuyer
,
A.
Gettleman
,
G.
Stephens
, and
C.
O’Dell
,
2008
:
The contribution of cloud and radiation anomalies to the 2007 Arctic sea ice extent minimum.
Geophys. Res. Lett.
,
35
.
L08503, doi:10.1029/2008GL033451
.
Klein
,
S. A.
, and
D. L.
Hartmann
,
1993
:
The seasonal cycle of low stratiform clouds.
J. Climate
,
6
,
1587
1606
.
Legates
,
D. R.
, and
C. J.
Willmott
,
1990
:
Mean seasonal and spatial variability in global surface air temperature.
Theor. Appl. Climatol.
,
41
,
11
21
.
Liu
,
Y.
,
J. R.
Key
,
J. A.
Francis
, and
X.
Wang
,
2007
:
Possible causes of decreasing cloud cover in the Arctic winter, 1982-2000.
Geophys. Res. Lett.
,
34
.
L14705, doi:10.1029/2007GL030042
.
Makshtas
,
A. P.
,
E. L.
Andreas
,
P. N.
Snvashchennikov
, and
V. F.
Timachev
,
1999
:
Accounting for clouds in sea ice models.
Atmos. Res.
,
52
,
77
113
.
Randall
,
D. A.
,
K-M.
Xu
,
R. J. C.
Somerville
, and
S.
Iacobellis
,
1996
:
Single-column models and cloud ensemble models as links between observations and climate models.
J. Climate
,
9
,
1683
1697
.
Randall
,
D.
, and
Coauthors
,
1998
:
Status of and outlook for large-scale modeling of atmosphere–ice–ocean interactions in the Arctic.
Bull. Amer. Meteor. Soc.
,
79
,
197
219
.
Rasch
,
P. J.
, and
J. E.
Kristjánsson
,
1998
:
A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations.
J. Climate
,
11
,
1587
1614
.
Rind
,
D.
,
R.
Healy
,
C.
Parkinson
, and
D.
Martinson
,
1995
:
The role of sea ice in 2×CO2 climate model sensitivity. Part I: The total role influence of sea ice thickness and extent.
J. Climate
,
8
,
449
463
.
Rind
,
D.
,
R.
Healy
,
C.
Parkinson
, and
D.
Martinson
,
1997
:
The role of sea ice in 2 × CO2 climate model sensitivity. Part II: Hemispheric dependencies.
Geophys. Res. Lett.
,
24
,
1491
1494
.
Rossow
,
W. B.
, and
R. A.
Schiffer
,
1999
:
Advances in understanding clouds from ISCCP.
Bull. Amer. Meteor. Soc.
,
80
,
2261
2287
.
Schweiger
,
A. J.
,
2004
:
Changes in seasonal cloud cover over the Arctic seas from satellite and surface observations.
Geophys. Res. Lett.
,
31
.
L12207, doi:10.1029/2004GL020067
.
Schweiger
,
A. J.
, and
J.
Key
,
1994
:
Arctic Ocean radiative fluxes and cloud forcing estimated from the ISCCP C2 cloud data set, 1983-1990.
J. Appl. Meteor.
,
33
,
948
963
.
Schweiger
,
A. J.
,
R. W.
Lindsay
,
J. R.
Key
, and
J. A.
Francis
,
1999
:
Arctic clouds in multi-year satellite data sets.
Geophys. Res. Lett.
,
26
,
1845
1848
.
Serreze
,
M. C.
, and
R. G.
Barry
,
2005
:
The Arctic Climate System.
Cambridge University Press, 385 pp
.
Slingo
,
J. M.
,
1987
:
The development and verification of a cloud prediction scheme for the ECMWF model.
Quart. J. Roy. Meteor. Soc.
,
113
,
899
927
.
Smith
,
R. D.
, and
P. R.
Gent
,
2002
:
Reference manual for the Parallel Ocean Program (POP), ocean component of the Community Climate System Model (CCSM2.0 and 3.0). Tech. Rep. LA-UR-02-2484, Los Alamos National Laboratory, 75 pp. [Available online at http://www.ccsm.ucar.edu/models/ccsm3.0/pop/.]
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M.
Tignor
, and
H. L.
Miller
,
2007
:
Climate Change 2007: The Physical Science Basis.
Cambridge University Press, 996 pp
.
Stephens
,
G. L.
, and
Coauthors
,
2002
:
The Cloudsat mission and the A-train: A new dimension of space-based observations of clouds and precipitation.
Bull. Amer. Meteor. Soc.
,
83
,
1771
1790
.
Stone
,
R. S.
,
D. C.
Douglas
,
G. I.
Belchansky
,
S. D.
Drobot
, and
J. M.
Harris
,
2005
:
Cause and effect of variations in western Arctic snow and sea ice cover. Proc. Eighth Conf. on Polar Meteorology and Oceanography, San Diego, CA, Amer. Meteor. Soc., 8.3
.
Symon
,
C.
,
L.
Arris
, and
B.
Heal
,
2005
:
Arctic Climate Impact Assessment: Scientific Report.
Cambridge University Press, 1042 pp
.
Thompson
,
S. L.
, and
D.
Pollard
,
1995
:
A global climate model (GENESIS) with a land-surface transfer scheme (LSX). Part I: present climate simulation.
J. Climate
,
8
,
732
761
.
Town
,
M. S.
,
V. P.
Walden
, and
S. G.
Warren
,
2007
:
Cloud cover over the South Pole from visual observations, satellite retrievals, and surface-based infrared radiation measurements.
J. Climate
,
20
,
544
559
.
Vavrus
,
S.
,
2004
:
The impact of cloud feedbacks on Arctic climate under greenhouse forcing.
J. Climate
,
17
,
603
615
.
Vavrus
,
S.
,
J. E.
Walsh
,
W. L.
Chapman
, and
D.
Portis
,
2006
:
The behavior of extreme cold-air outbreaks under greenhouse warming.
Int. J. Climatol.
,
26
,
1133
1147
.
Walsh
,
J. E.
,
V.
Kattsov
,
D.
Portis
, and
V.
Meleshko
,
1998
:
Arctic precipitation and evaporation: Model results and observational estimates.
J. Climate
,
11
,
72
87
.
Walsh
,
J. E.
,
V. M.
Kattsov
,
W. L.
Chapman
,
V.
Govorkova
, and
T.
Pavlova
,
2002
:
Comparison of Arctic climate simulations by coupled and uncoupled models.
J. Climate
,
15
,
1429
1446
.
Walsh
,
J. E.
,
S. J.
Vavrus
, and
W. L.
Chapman
,
2005
:
Summary of a workshop on modeling of the Arctic atmosphere.
Bull. Amer. Meteor. Soc.
,
86
,
845
852
.
Wang
,
X.
, and
J. R.
Key
,
2003
:
Recent trends in Arctic surface, cloud, and radiation properties from space.
Science
,
299
,
1725
1728
.
Wang
,
X.
, and
J. R.
Key
,
2005a
:
Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part I: Spatial and temporal characteristics.
J. Climate
,
18
,
2558
2574
.
Wang
,
X.
, and
J. R.
Key
,
2005b
:
Arctic surface, cloud, and radiation properties based on the AVHRR Polar Pathfinder dataset. Part II: Recent trends.
J. Climate
,
18
,
2575
2593
.
Warren
,
S. G.
,
C. J.
Hahn
,
J.
London
,
R. M.
Chervin
, and
R. L.
Jenne
,
1986
:
Global distribution of total cloud cover and cloud type amounts over land. NCAR Tech. Note TN-273+STR, 29 pp
.
Warren
,
S. G.
,
C. J.
Hahn
,
J.
London
,
R. M.
Chervin
, and
R. L.
Jenne
,
1988
:
Global distribution of total cloud cover and cloud type amounts over the ocean. NCAR Tech. Note TN-317+STR, 42 pp
.
Wilson
,
L. D.
,
J. A.
Curry
,
T. P.
Curry
, and
T. P.
Ackerman
,
1993
:
Satellite retrieval of lower-tropospheric ice crystal clouds in the polar regions.
J. Climate
,
6
,
1467
1472
.
Wyser
,
K.
, and
C. G.
Jones
,
2005
:
Modeled and observed clouds during Surface Heat Budget of the Arctic Ocean (SHEBA).
J. Geophys. Res.
,
110
.
D09207, doi:10.1029/2004JD004751
.
Xie
,
P. P.
, and
P. A.
Arkin
,
1996
:
Analyses of global monthly precipitation using gauge observations, satellite estimates, and numerical model predictions.
J. Climate
,
9
,
840
858
.
Xu
,
K-M.
, and
D. A.
Randall
,
1996
:
A semiempirical cloudiness parameterization for use in climate models.
J. Atmos. Sci.
,
53
,
3084
3102
.
Zhang
,
G. J.
, and
N. A.
McFarlane
,
1995
:
Sensitivity of climate simulations to the parameterization of cumulus convection in the Canadian Climate Centre general circulation model.
Atmos.–Ocean
,
33
,
407
446
.
Zhang
,
M.
,
W.
Lin
,
C. S.
Bretherton
,
J. J.
Hack
, and
P. J.
Rasch
,
2003
:
A modified formulation of fractional stratiform condensation rate in the NCAR Community Atmospheric Model (CAM2).
J. Geophys. Res.
,
108
.
4035, doi:10.1029/2002JD002523
.

Footnotes

Corresponding author address: Steve Vavrus, Center for Climatic Research, University of Wisconsin—Madison, 1225 W. Dayton Street, Madison, WI 53706. Email: sjvavrus@wisc.edu