Abstract

Ice cloud properties in Northern Hemisphere winter extratropical cyclones are examined using the Atmospheric Infrared Sounder (AIRS), version 6, cloud products. The cloud thermodynamic phase product indicates that warm frontal clouds are dominated by ice, liquid-phase clouds occur outside of the warm frontal region, and supercooled or mixed-phase clouds are found in the southwestern quadrant of the cyclones. Stratiform ice clouds populate the warm frontal region and portions of the cold sector while convective ice clouds populate southeastern portions of the warm front and the southeastern quadrant. Total cloud cover is smaller in land cyclones than in ocean cyclones, especially in the southwestern quadrant and the warm frontal region. Ice cloud cover is less over land in the warm frontal region, because land cyclones are generally weaker and drier than ocean cyclones. The impact of cyclone average precipitable water (PW) and the magnitude of 850-hPa vertical ascent ω850 on the thermodynamic phase, occurrence of stratiform or convective ice cloud, ice particle effective diameter, optical thickness, and cloud-top temperature are discussed. When comparing land and ocean cyclones with similar PW and ω850, ice cloud coverage is found to be greater over land. Convective ice cloud occurs more often and is deeper over land. Supercooled cloud appears to persist to colder temperatures over ocean than over land, especially in the warm frontal region. These results suggest that, over land, ice cloud formation in warm fronts is possibly more efficient because of a greater aerosol amount from local or regional sources.

1. Introduction

The midlatitudes in winter have variable precipitation amounts; extremes in winds, rain, or snow accumulations; maxima in total cloud cover; and radiative impacts that are primarily controlled by the frequency, location, and intensity of extratropical cyclones (e.g., Stewart et al. 1998). However, there remains a large uncertainty regarding the characteristics of these systems in a warming climate (e.g., Feser et al. 2015, and references therein): some studies predict an increase in intensity and no change in occurrence frequency, while others find no changes at all, and still others predict a decrease in occurrence frequency. This lack of consensus may be caused in part by issues in the correct representation of the storm tracks in general circulation models (e.g., Chang et al. 2013; Zappa et al. 2013), although an increase in spatial resolution has helped some models improve significantly for some regions (Colle et al. 2013). Issues with the dynamics in the models have been partially attributed to cloud representation (e.g., Trenberth and Fasullo 2010; Grise and Polvani 2014). In fact, an underestimate in modeled cloud cover over the Southern Ocean is endemic especially within extratropical cyclones and across fronts (Naud et al. 2010; Booth et al. 2013; Bodas-Salcedo et al. 2014; Naud et al. 2014).

Typically, extratropical frontal systems contribute at least 70% of the midlatitude precipitation (Catto et al. 2012). Clouds and precipitation in these systems are produced along and across cold and warm fronts (Stewart et al. 1998) and are at a maximum along the warm front and warm conveyor belt (e.g., Carlson 1980). The warm conveyor belt and warm frontal zones are the primary production zones of precipitation (e.g., Browning 1986). Field and Wood (2007, hereinafter FW07) found a strong correlation between the strength of the moisture flux along the warm conveyor belt and the precipitation rates at the warm front and the cloud cover above 440 hPa. They found similar results whether the cyclones were found over the Northern or Southern Hemisphere oceans but did not explore cyclones over land.

Here we would like to complement the work of FW07 and further explore 1) the microphysical properties of ice clouds that form in extratropical cyclones and 2) the potential differences between land and ocean cyclones. Cloud-top properties are important because they impact the radiation fluxes at the top of the atmosphere (Tselioudis et al. 2000) and can help inform the microphysical processes at work within the clouds (Stewart et al. 1998). To our knowledge, extratropical cyclone–centered composites of cloud-top microphysical properties have not been explored in detail, although a vast array of research has looked into cloud microphysics in individual cyclones (e.g., Hobbs 1978; Ryan 1996; Stewart et al. 1998, and references therein). Compositing allows us to extract the most salient features of cloud-top properties in extratropical cyclones. Therefore, composites can be used to help diagnose and evaluate model performance in dynamical systems where subgrid-scale processes are important (Katzfey and Ryan 2000). Consequently, they can help pinpoint the parameterizations that may need improvement (Klein and Jakob 1999; Field et al. 2011).

For this compositing study, we use the Atmospheric Infrared Sounder (AIRS; Chahine et al. 2006), version 6, cloud products (Kahn et al. 2014). AIRS hyperspectral measurements have been shown to be very accurate at detecting and characterizing ice clouds (Kahn et al. 2014 and references therein), as demonstrated when compared with Cloud–Aerosol Lidar with Orthogonal Polarization (CALIOP) cloud phase (Jin and Nasiri 2014). Cloud thermodynamic phase, ice cloud-top temperature, effective radius, and optical thickness are composited in a cyclone-centered grid as in FW07, although all of the cyclones are rotated such that the warm fronts are aligned along the east–west direction, as in Naud et al. (2012). These composites will help 1) determine how cloud-top thermodynamic phase and ice cloud properties change with cyclone strength and moisture availability, and 2) quantify differences and similarities in cloud-top properties of land and ocean cyclones.

2. Data and method

The AIRS cloud products that are used in this study are described in this section, along with the compositing method, which is identical to the method used in Naud et al. (2012). In addition, we use the same cyclone and warm front database as Naud et al. (2012), which covers the period from 2006 to 2010.

a. Description of AIRS cloud products

The AIRS, version 6, cloud products include cloud-top thermodynamic phase discrimination, where clouds are differentiated into ice, liquid, or an “unknown” category based on a series of midinfrared (MIR) spectral brightness temperature Tb and ΔTb thresholds (Nasiri and Kahn 2008; Jin and Nasiri 2014). There are four different ice-phase tests and two different liquid-phase tests. The unknown-phase category is intended to capture a variety of supercooled liquid- and mixed-phase clouds at high latitudes, extending into the midlatitudes during local wintertime, and liquid shallow cumulus clouds at low latitudes, extending into the midlatitudes during local summertime (Jin and Nasiri 2014; Kahn et al. 2014). Further research is necessary to determine if supercooled and mixed-phase clouds can be discriminated from each other using combinations of MIR channels, which is the reason for the all-encompassing unknown category.

In single-layered clouds, Jin and Nasiri (2014) showed that the AIRS phase indicates ice about 99% of the time whenever a portion or full transect through the AIRS footprint contains ice according to the CALIOP phase feature product (Hu et al. 2010). Likewise, in single-layered clouds, liquid water clouds are reported 95% of the time that CALIOP reports liquid, and that percentage increases to 99% when mixed-phase from CALIOP is considered within the transect. In short, AIRS is very unlikely to systematically mischaracterize ice as liquid and liquid as ice. However, much of the liquid cloud detected by CALIOP falls within the unknown AIRS phase category: specifically, 13.6% is ice, 9.4% is mixed phase, and 76.9% is liquid according to CALIOP (Jin and Nasiri 2014). The effect of multilayered clouds and broken clouds within the AIRS footprint is to “blur” the sharpness of these categorizations by not much more than several percent from the single-layered clouds.

For the 27% of AIRS footprints that have been identified as containing ice over the lifetime of the Aqua mission, an a posteriori optimal estimation retrieval is performed on the AIRS footprint with three variables in the state vector: ice cloud effective diameter De, optical thickness τice, and cloud-top temperature CTTice (Kahn et al. 2014). Error estimates and scalar averaging kernels, as well as a quality control flagging approach that is analogous to the AIRS Standard cloud clearing products, are provided along with the retrieved ice cloud products. All prior estimates of temperature and water vapor profiles, surface temperature, and emissivity are held fixed and are taken from the AIRS L2 Standard and Support product files (http://disc.sci.gsfc.nasa.gov/AIRS). The upper-layer AIRS cloud-top temperature CTTupper is used as the prior guess for the CTTice retrieval; more detailed discussion on prior guesses, error covariances, and validation of the approach is found in Kahn et al. (2014).

In addition to the aforementioned cloud fields, AIRS provides a two-layer cloud-top temperature (CTTupper and CTTlower) and effective cloud fraction product that is retrieved after the cloud clearing retrieval and before the ice cloud optimal estimation retrieval (Kahn et al. 2014). Extensive validation of these products against the 94-GHz CloudSat radar (Stephens et al. 2008), Atmospheric Radiation Measurement Program surface sites (Ackerman and Stokes 2003), and CALIOP lidar are found in Kahn et al. (2007a, 2008, 2014). Furthermore, pixel-scale comparisons between two-layer AIRS cloud products and single-layered MODIS MIR cloud products are described and reconciled by Kahn et al. (2007b) and Nasiri et al. (2011). The AIRS and MODIS cloud products show remarkable consistency when the pixel-matched cloud products are collapsed into an “effective” Tb space, suggesting that the differences between AIRS and MODIS are largely driven by algorithm assumptions and pixel-scale surface, atmospheric, and cloud heterogeneity.

b. Method

For the period December–February (DJF) 2006–10, we used the Modeling, Analysis, and Prediction (MAP) Program Climatology of Midlatitude Storminess database (MCMS; Bauer and Del Genio 2006) to locate extratropical cyclone centers with a temporal frequency of 6 h between 30° and 60°N. For the identifiable low pressure centers, warm fronts were detected as described in Naud et al. (2012). Herein, we added to the ocean-only database of the Naud et al. (2012) paper with land cases. We used the same warm frontal detection technique (Hewson 1998) applied to 6-hourly, 0.5° × 0.667°, NASA Modern-Era Retrospective Analysis for Research and Applications (MERRA; Rienecker et al. 2011) potential temperatures extracted at 1 km above the surface (to alleviate issues with highly varying topography). For this time period, there were 9936 and 7688 low pressure centers with a successful warm front detection available over Northern Hemisphere (NH) ocean and land, respectively. Next, we extracted AIRS, level 2, swath data that intersect within ±3 h and 25° of the low pressure centers along a radial direction. These conditions were met for 7562 and 5382 low pressure centers over the oceans and land, respectively. Partial coverage by the AIRS swath over the midlatitude cyclone domain was permitted. The AIRS datasets were projected into a stereographic grid centered on the low. Cyclone-centered composites of these datasets were then formed by averaging all data only after a rotation was applied for each individual cyclone resulting in every warm front being aligned in the east–west direction. Separate composites of each AIRS product were obtained for all cyclones over land and over ocean.

The large-scale properties of the cyclones were obtained by applying a similar gridding method onto 6-hourly, 0.5° × 0.667°, MERRA column-integrated precipitable water (PW) and 850-hPa vertical velocities ω850. We chose the 850-hPa level because it captures the strength of cyclone ascent in close proximity to the location of the warm front where it intersects the surface. We found no significant differences between the cyclone population sorted with the 850-hPa or the 500-hPa vertical velocities. For each cyclone, the mean PW extracted in a 25° radius centered on the low and the mean ω850 in the area of the cyclone (within 25° of the low) where ω850 is negative, that is, in the ascending region, were calculated. These two governing cyclone characteristics were then used to sort the individual cyclone “snapshots” into nine categories, defined by three range bins each of PW and ascent strength of ω850. The bins were defined by sorting the cyclone population into three equal size subsets separately for PW and ω850. This entails some slight differences in the number of members in each of the nine categories and slight variations in the average PW (ω850) per category for a given range of PW (ω850) as the ω850 (PW) range changes.

3. Ice cloud properties in winter extratropical cyclones in the Northern Hemisphere: Ocean and land

With an initial focus on the average properties of all ocean and land winter cyclones, Fig. 1 shows the cyclone-centered distribution of total cloud cover frequency as observed with AIRS. Here, and in the rest of the study, we divide the cyclone-centered composites into four quadrants: northeastern, northwestern, southwestern, and southeastern. We sometimes refer to the southeastern quadrant as the warm sector, since the warm fronts have been intentionally aligned along a horizontal line east of the low.

Fig. 1.

Cyclone-centered composites of total cloud coverage for all cyclones during NH winter (DJF) in 2006–10 over (a) oceans and (b) land. The total cloud cover frequency is the sum of liquid, unknown, and ice phases retrieved at the AIRS footprint scale, which requires a minimum effective cloud fraction >0.01 according to Jin and Nasiri (2014).

Fig. 1.

Cyclone-centered composites of total cloud coverage for all cyclones during NH winter (DJF) in 2006–10 over (a) oceans and (b) land. The total cloud cover frequency is the sum of liquid, unknown, and ice phases retrieved at the AIRS footprint scale, which requires a minimum effective cloud fraction >0.01 according to Jin and Nasiri (2014).

Figure 1a resembles similar total cloud cover distributions obtained with other instruments (e.g., Lau and Crane 1995; FW07; Naud et al. 2013): large oceanic cloud cover (>0.93) in an area comprising the low and extending eastward along and northward in advance of the warm front and relatively low cloud cover on the polar edge and in the eastern edge of the warm sector. The cyclone-centered composites show a strong signature of the warm frontal zone while a signature of the cold frontal zone is not emerging. This is principally caused by the very localized features associated with cold fronts that become blurred from significant changes in cold front location with respect to the low pressure center from one cyclone to another, and from rotating the warm front to a common east–west orientation. Figure 1b reveals that land cyclones display a qualitatively similar cloud cover distribution. However, the cloud amount is much lower than over oceans, with a peak value in the warm frontal zone less than 0.9 and a more compact area of large cloud cover in the eastward radial direction [consistent with Lau and Crane (1997)]. In addition, the asymmetry in land cloud cover between the southwestern and southeastern quadrants is inverted compared with over the ocean. This apparent asymmetry is principally caused by the much larger oceanic cloud cover in the southwestern quadrant.

Next, we explore the cloud thermodynamic phase that might govern the discrepancies between land and ocean. Figure 2 shows separate composites of cloud cover for ice, liquid, and unknown phase. For both land and ocean cyclones, ice occurs predominantly in the warm frontal region, decreases toward the pole and into the warm sector, and is at a minimum in the southwestern quadrant (Figs. 2a,b). Ice cloud cover is larger in the warm frontal zone over oceans than land but generally smaller in the southwestern quadrant. Liquid-topped clouds are detected much less often than ice clouds in both land and ocean cyclones in the warm frontal region (less than 15% of the time) but reach a maximum in the southwestern quadrant (Figs. 2c,d). Except in the warm frontal zone where land and ocean cyclones display very similar liquid cloud cover frequencies, liquid clouds are more often found over ocean than land. Finally, the unknown-phase clouds tend to occur most frequently in the southwestern quadrant of the cyclones. Over oceans, unknown phase is much more prevalent than over land (Figs. 2e,f) and in the southwestern quadrant occurs more than 50% of the time.

Fig. 2.

Cyclone-centered composites of cloud coverage for each thermodynamic phase, with (a),(b) ice, (c),(d) liquid, and (e),(f) unknown phases over (left) ocean and (right) land for the wintertime NH.

Fig. 2.

Cyclone-centered composites of cloud coverage for each thermodynamic phase, with (a),(b) ice, (c),(d) liquid, and (e),(f) unknown phases over (left) ocean and (right) land for the wintertime NH.

According to Jin and Nasiri (2014), the AIRS thermodynamic phase retrieval algorithm is not able to distinguish liquid and ice in three main scenarios: 1) the clouds are overcast but contain a mixture of ice and liquid interdispersed horizontally at small spatial scales (Chylek and Borel 2004) and/or layered vertically (Shupe et al. 2006), such that a weak MIR spectral signature is imparted; 2) the clouds are liquid with CTT > 273 K but are scattered and broken in the AIRS field of view (FOV), for instance, in the case of trade cumulus; or 3) the CTTs between 250 and 265 K are known to exert weak MIR spectral signatures (Nasiri and Kahn 2008). Over the oceans, the southwestern quadrant frequently encompasses the post–cold frontal region that is often populated by broken stratocumulus or open cellular cumulus clouds (e.g., Naud et al. 2014). As a result, it should come as no surprise that the AIRS phase algorithm identifies a preponderance of unknown phase in this region of the cyclone. Figure 2f demonstrates that over land, the southwestern quadrant contains a much lower frequency of unknown phase, with similar frequencies of liquid phase and slightly greater frequencies of ice phase.

The subtle land and ocean differences in ice phase are more challenging to explain than the differences in unknown phase. To further investigate the ice cloud differences, we take advantage of the various spectral tests that are available as “phase bits” in the AIRS thermodynamic phase retrieval product set. Out of the four tests used to detect ice, passing only one or two of the tests indicate a sensitivity to optically thinner stratiform-like clouds, while passing three or four of the tests indicates a sensitivity to optically thicker convective-like clouds. In particular, Kahn et al. (2014, their Fig. 6) show that regions where only one or two tests are passed correlate well with regions known for their predominantly stratiform nature, while regions where three or four tests are passed correspond to regions of active convection. Figure 3 shows the cyclone-centered composites of the “stratiform” (one or two tests passed) and “convective” (three or four tests passed) ice cloud cover over ocean and land. Stratiform clouds dominate the overall ice cloud frequency and are at a maximum in the warm frontal zone especially on the poleward side of the low. Convective clouds are most frequent in the southeast portion of the warm frontal zone extending into the southeastern quadrant, that is, where the temperatures and static stability are more conducive to convection, especially in the vicinity of the cold fronts. In view of the qualitatively realistic spatial distributions of stratiform and convective ice clouds, for the remainder of the manuscript the number of ice-phase tests triggered will be used as a means of ice cloud typing. Although the two types are referred to as stratiform and convective, bear in mind that this categorization is not as physically based as the cloud-type classification available from the CloudSat 94-GHz radar (Wang et al. 2012) for example, and is more analogous to the categories specified by the International Satellite Cloud Climatology Project (ISCCP) classification (Rossow and Schiffer 1999). Although the areal coverage of stratiform ice clouds show slight disparities in shape between ocean and land cyclones (Figs. 3a,b), the maximum cloud covers over ocean and land are within 3% of each other. In contrast, the areal coverage and frequency of convective ice clouds are larger over the ocean throughout the warm frontal region than over land (Figs. 3c,d). The convective ice clouds over land are more localized in the southeastern quadrant and along the southeast extent of the warm front compared with the ocean.

Fig. 3.

Cyclone-centered composites of ice (a), (b) stratiform and (c), (d) convective clouds that are based on the number of ice-phase tests passed (see text in section 3) over (left) ocean and (right) land for all cyclones in the wintertime NH.

Fig. 3.

Cyclone-centered composites of ice (a), (b) stratiform and (c), (d) convective clouds that are based on the number of ice-phase tests passed (see text in section 3) over (left) ocean and (right) land for all cyclones in the wintertime NH.

The primary sources of differences in ice cloud coverage between land and ocean originate from a greater probability of optically thicker, convective-like cloud over oceans and a more compact stratiform warm frontal region over land. These two features point to differences in the cyclone properties: ocean cyclones are more vigorous (Fig. 4a) and tend to contain more moisture (Fig. 4b). Therefore, the larger convective ice coverage over oceans is likely explained in part by the greater amount of moisture, while a more spatially constrained area of stratiform ice clouds over land in the warm frontal zone is consistent with a more spatially constrained and overall weaker ω850 ascent.

Fig. 4.

Cyclone-centered composites of the difference in MERRA (a) ω850 and (b) PW between wintertime NH ocean and land cyclones.

Fig. 4.

Cyclone-centered composites of the difference in MERRA (a) ω850 and (b) PW between wintertime NH ocean and land cyclones.

Before we can test whether these two cyclone characteristics largely govern observed land–ocean contrasts, in the following section the availability of moisture and variable ascent impacts on ice cloud properties are investigated.

4. Ice properties in extratropical cyclones: Impact of ω850 and PW

Following the work of FW07, the ocean cyclones are divided into nine categories: any combination of low, medium, and high PW and weak, moderate, and strong ω850. The work of FW07 is expanded by quantifying the sensitivity of additional ice cloud parameters in an attempt to better understand the impacts of PW and ω850 on cloud-top microphysics. In addition, the same analysis is performed on land cyclones to determine if similar sensitivities of ice cloud properties to large-scale conditions are displayed.

a. Cyclone properties

Table 1 contains the number of cyclones that fall within each category that are coincident in time (±3 h) and space (25° radial distance from the low) to the AIRS swath data. The range of PW and ω850 in the region of ascent is also shown for each category. Note that there are not lower/upper thresholds for the extreme lowest/largest categories. The land cyclone subsets use the same thresholds used to separate the categories; however, the numbers of cyclones per category are not as uniform as the numbers for ocean cyclones (cf. Table 1). In addition, for the extreme categories, the distributions will differ for the most extreme values. The number of cyclones per category confirms that land cyclones are on average drier and weaker than their oceanic counterparts. Table 1 shows that the high PW category is much less populated over land than the low or medium PW categories.

Table 1.

Number of ocean (land) cyclones that were within AIRS FOV within ±3 h and 25° from the low pressure center per mean PW–ω850 category combination.

Number of ocean (land) cyclones that were within AIRS FOV within ±3 h and 25° from the low pressure center per mean PW–ω850 category combination.
Number of ocean (land) cyclones that were within AIRS FOV within ±3 h and 25° from the low pressure center per mean PW–ω850 category combination.

Figure 5 shows the locations of the cyclones for each category. Over the ocean, the strongest cyclones are typically found in the western portions of the Pacific and Atlantic Ocean basins, while the weakest cyclones are predominantly found to the east, which is consistent with the locations of genesis, peak intensity, and lysis of the Northern Hemisphere storm tracks (Hoskins and Hodges 2002). The stronger land cyclones are typically closer in proximity to the coastlines or in the lee of major mountain ranges such as the Rockies, which is also consistent with Hoskins and Hodges’s (2002) analysis. Figure 5 also demonstrates that for a given PW category, winter cyclones over oceans are found farther north than their land counterparts.

Fig. 5.

Location of cyclones for each of the nine category combinations over land and ocean (see Table 1), with the strength of ω850 increasing from left to right, and PW increasing from top to bottom.

Fig. 5.

Location of cyclones for each of the nine category combinations over land and ocean (see Table 1), with the strength of ω850 increasing from left to right, and PW increasing from top to bottom.

For the ocean cyclones, Fig. 6 shows the cyclone-centered composites of PW and ω850 for each PW–ω850 category. Figure 6 clearly shows how the ω850 ascent evolves from a small region close to the low pressure center into a comma-shaped region that encompasses the southeastern quadrant (e.g., Houze 1993, 467–469). PW is elevated in the warm sector (southeastern quadrant) and as the cyclonewide PW increases, the maximum in the warm sector also increases. In contrast, as cyclonewide PW increases, the minimum PW in the cold sector further decreases. As the ω850 ascent strengthens, the maximum in PW increases in the warm sector. The correlation between the absolute value of the ascent strength and cyclonewide PW for each cyclone is less than R = 0.01. However, when the distribution of PW is examined within the cyclones (Fig. 6), the contrast in PW between the warm and cold sector significantly increases with the strength of the ω850 ascent. In other words, the cold sector becomes drier and the warm sector becomes moister as ω850 ascent becomes more vigorous. FW07 found a negative correlation between mean PW and mean wind speed for their subset of cyclones and argued that this was caused by the cyclones reaching peak intensity farther away from the equator. In our subsets, a wider range of longitudes and latitudes is included such as regions of cyclone genesis along the east coast of the continents, unlike in FW07’s study. As a consequence, there are likely to be more vigorous cyclones at lower latitudes in the present study than used in FW07.

Fig. 6.

Cyclone-centered composites of MERRA PW (in color) and ω850 (black contours) for ocean cyclones in the nine category combinations that are defined in Table 1.

Fig. 6.

Cyclone-centered composites of MERRA PW (in color) and ω850 (black contours) for ocean cyclones in the nine category combinations that are defined in Table 1.

Over land, a similar evolution of ω850 ascent and PW distributions is found when compared with over the ocean, albeit with lower values of both ω850 and PW (not shown).

b. Ice cloud sensitivity to PW and ω850 over oceans

Over oceans, ice cloud coverage in the warm frontal zone increases with ω850 strength (Fig. 7). This is consistent with FW07 who examined high cloud cover defined by their cloud-top pressure (CTP < 440 hPa) rather than the thermodynamic phase. Similar to the FW07 findings, as cyclone strength increases, an overall comma shape within the region of maximum frequency becomes apparent (cf. Fig. 6). A weak correlation between high cloud cover and PW was shown by FW07. The ice cloud cover used here can include clouds at levels below 440 hPa that may be more sensitive to PW. Also, while clouds with CTP < 440 hPa are expected to be overwhelmingly ice-phase, supercooled, and mixed-phase clouds can occur with some considerable frequency up to −40°C (Hu et al. 2010; Jin and Nasiri 2014). We point out that ice cloud coverage changes consistently with PW in all three cyclone strength categories: the drier cases tend to have a maximum in ice cloud coverage on the poleward side of the low, with the comma shape barely visible at weak and moderate ω850. As PW increases, ice clouds are found in a tighter zone north of the low that extends eastward along the warm front. In the high PW cases, the spatial distribution of the region of maximum ice cloud coverage becomes more annular than comma shaped. The maximum cloud coverage is confined to the east of the low except for the strongest cyclones where a very weak comma shape is found and the maximum is to the northeast.

Fig. 7.

Cyclone-centered composites of ice cloud fraction in wintertime for NH oceans as a function of cyclonewide mean PW (from top to bottom) and ω850 (from left to right).

Fig. 7.

Cyclone-centered composites of ice cloud fraction in wintertime for NH oceans as a function of cyclonewide mean PW (from top to bottom) and ω850 (from left to right).

When distinguishing between the stratiform and convective ice clouds using the four different ice-phase tests available (section 3), a clearer picture is seen between the ice clouds as a function of changing PW and/or ω850. In the polar half of the cyclones, stratiform ice clouds dominate and clearly depend on PW (temperature) (Fig. 8). At the warm front, stratiform ice clouds depend on the strength of ω850 and not on PW, while convective clouds depend more clearly on PW (Fig. 9). In the southeastern quadrant, where ice cloud cover dependency on PW and ω850 is less clear, stratiform ice cloud coverage increases with ω850 strength but decreases with PW. The convective ice cloud coverage expands into the southeastern quadrant as ω850 increases in magnitude and areal coverage but does not have a clear dependency on PW. That said, as PW increases, convective ice cloud cover in the warm frontal zone increases but over a more limited area that is approximated by the areal coverage of strongest ascent. As for ice cloud coverage in the warm conveyor belt, these results indicate that the strength of ω850 plays a large role for the probability of ice to be detected at cloud top, whereas for weaker cyclones, drier conditions favor the probability of ice formation through their correlation with low temperatures. The transition from a maximum in stratiform ice in the warm frontal zone to convective ice in the southeastern quadrant (where the cold front and warm conveyor belt might be present) where the PW is higher is also reassuring and lends further confidence in this approximate cloud typing approach.

Fig. 8.

As in Fig. 7, but for stratiform ice cloud coverage.

Fig. 8.

As in Fig. 7, but for stratiform ice cloud coverage.

Fig. 9.

As in Fig. 7, but for convective ice cloud coverage.

Fig. 9.

As in Fig. 7, but for convective ice cloud coverage.

Next, we explore how the changes in ice cloud coverage and type with PW and ω850 relate to the ice cloud properties retrieved from AIRS (De, CTTice, and τice). Table 2 shows the correlation coefficient R for all radial and angular points within the cyclone composite between these three cloud-top properties and convective and stratiform ice cloud coverage.

Table 2.

Correlation coefficients between τice, CTTice, and De and both convective and stratiform ice cloud coverage for all angular and radial bins within the composites of all cyclones over oceans (land).

Correlation coefficients between τice, CTTice, and De and both convective and stratiform ice cloud coverage for all angular and radial bins within the composites of all cyclones over oceans (land).
Correlation coefficients between τice, CTTice, and De and both convective and stratiform ice cloud coverage for all angular and radial bins within the composites of all cyclones over oceans (land).

The values of De are relatively larger in the southeastern quadrant and a portion of the southwestern quadrant when compared with the cold sector, with a difference of about 10 μm (Fig. 10). Upon first inspection, the strength of ω850 has little obvious impact on De at cloud top but a clear increase in De is observed as PW (and temperature) increases. Table 2 shows that there is little correlation between De and convective ice cloud coverage. However, a nonnegligible anticorrelation between De and stratiform ice cloud coverage is found (R = −0.58; i.e., as stratiform ice cloud coverage increases, De decreases), also apparent when comparing Figs. 8 and 10. This suggests that somewhat larger ice particles in the warm front and southeastern quadrant must originate from convective lofting in regions of stronger ascent. However, the temperature/moisture impacts on De appear to dominate at cloud top.

Fig. 10.

As in Fig. 7, but for ice cloud De at cloud top.

Fig. 10.

As in Fig. 7, but for ice cloud De at cloud top.

In the case of CTTice (Fig. 11), not to be confused with the two-layer CTT as part of the AIRS Standard L2 processing, is found to have a minimum value (CTTice < 230–235 K) in the warm frontal zone extending more radially outward than the ice cloud fraction maps (Fig. 7). This area of relative minimum extends into the cold sector for drier PW subsets and into the southeastern quadrant as ω850 becomes more vigorous (Fig. 11). A region of maximum values (CTTice > 245 K) found in the southwestern quadrant displays little relation with cyclone strength but tends to diminish in size and migrate northward as PW increases. The lower layer of the two-layer standard CTT products shows even higher values of CTTlower in the cold sector (not shown), indicative of frequent shallow liquid- and mixed-phase clouds that are found below a smaller population of transparent ice clouds (e.g., Fig. 2). There is a tendency for CTTice to decrease as both convective and stratiform cloud coverage increase, but the correlation is stronger for convective ice cloud cover (Table 2). In the warm frontal region, where the strength of ω850 is a maximum, CTTice is low enough to allow ice to form either through homogeneous nucleation or processes such as the Bergeron–Findeisen growth of ice crystals at the expense of supercooled water droplets (e.g., Naud et al. 2006). This is the region of maximum occurrence of stratiform clouds and average De is relatively small. As PW is reduced, stratiform ice clouds begin to fill in the cold sector and poleward, as midlevel altostratus or altocumulus clouds are found at temperatures low enough for glaciation to take place at cloud top. The clouds in the cold sector and poleward tend to be less glaciated as PW (temperature) increases, reflected in the reduction of stratiform ice clouds (Fig. 8).

Fig. 11.

As in Fig. 7, but for CTTice.

Fig. 11.

As in Fig. 7, but for CTTice.

Also, τice is at a maximum in the southeastern quadrant for low and moderate PW (Fig. 12) and does not spatially correspond to the ice cloud fraction (Fig. 7). As PW increases, the region of maximum τice encompasses the low resembling more closely ice cloud fraction, but no longer extends equatorward (Fig. 12). As ω850 increases, the maximum τice increases in proximity of the low and in the warm frontal zone, while ice clouds overall throughout the entire cyclone area become optically thicker. There is little correlation between τice and stratiform ice cloud coverage (Table 2), although a calculation of the correlation coefficient for each individual category reveals that there is a strong correlation between the two for the strong ω850 category (R > 0.75). As expected, the correlation with convective ice cloud coverage is more evident with larger τice as deep convective clouds occur with increasing frequency. This correlation is found to be strongest for the strong ω850 cases (R > 0.56).

Fig. 12.

As in Fig. 7, but for τice.

Fig. 12.

As in Fig. 7, but for τice.

5. Differences in ice cloud properties between land and ocean for similar large-scale conditions

Before we restrict the comparison to similar ω850/PW conditions, the dependency of ice cloud properties on PW and ω850 over land were examined. Most of the previously described sensitivities observed in ocean cyclones are also true for land cyclones, but some notable differences were found. For instance, the maximum stratiform cloud coverage decreases with increasing PW, while the maximum convective cloud coverage increases (not shown). The impact of ω850 on the spatial distributions and magnitudes of the cloud fields is as evident as for the ocean cases. One important difference is that when the same range of PW and ω850 is used for both land and ocean cyclones, the more vigorous storms are relatively uncommon over land, and a significant sample size issue arises for the more vigorous and humid cyclones (cf. Table 1).

While the range of PW and ω850 for each category are the same for land and ocean, the sample sizes and distributions within each category are not in nature. Thus, the differences between ω850 and PW still play a role in the ice cloud coverage differences in each category. Fortunately, the middle category (medium PW, moderate ω850) has a reasonably similar sample size for both land and ocean cyclones and very similar PW and ω850 distributions. Other categories either show a large disparity in the land and ocean sample size and/or noticeable differences in the distributions of PW or ω850. It follows that this category is selected for more detailed comparison between land and ocean cyclones. That said, we verified that the conclusions obtained for this category remain consistent when combining multiple categories to increase the cyclone population.

With regard to the cloud occurrence for each phase category (ice, liquid, and unknown) or type (stratiform versus convective), a two-tailed binomial test is applied to determine whether the differences are significant (p < 0.025). The differences in Fig. 13 are masked (white pixels) if 1) the binomial test results in no significance (p > 0.025), and 2) the differences are less than 0.04, an assumed uncertainty in the observations based on previous work with different satellite instruments and retrieval methodologies (Naud et al. 2013).

Fig. 13.

Cyclone-centered composites of ocean–land differences when statistically significant for (a) ice-, (b) liquid-, and (c) unknown-phase cloud coverage and (d) stratiform and (e) convective ice cloud coverage restricted to the medium PW and moderate ω850 category.

Fig. 13.

Cyclone-centered composites of ocean–land differences when statistically significant for (a) ice-, (b) liquid-, and (c) unknown-phase cloud coverage and (d) stratiform and (e) convective ice cloud coverage restricted to the medium PW and moderate ω850 category.

When the composites are constrained to moderate ω850 and medium PW, a greater ice cloud coverage is found over land in the area near the low pressure center and in the cold sector on the northernmost edge of the warm frontal zone (Fig. 13a). The differences in other thermodynamic phases are also investigated to determine whether the remaining cloud retrievals over ocean are either clear, liquid, or unknown. For clear-sky retrievals (not shown), they are more or less as frequent over land and ocean, implying that less ice cloudiness does not result in less overall cloudiness over oceans. Away from the low pressure center, liquid clouds occur slightly more often over ocean than land (Fig. 13b) and unknown phase is also significantly larger over ocean than over land (Fig. 13c). The largest differences in unknown phase occur where liquid clouds are frequently located but are horizontally broken (southwestern quadrant) or where mixtures of liquid and ice phase may be present such as the cold sector/warm frontal zone (Jin and Nasiri 2014). The latter issue, along with the difference in liquid cloud cover in Fig. 13b, suggests that oceanic warm frontal regions contain more liquid cloud tops than over land for similar PW and ω850. Over land, more frequent stratiform ice clouds are found on the polar edge of the warm front (Fig. 13d), while more frequent convective ice clouds are found in proximity of the cyclone center (Fig. 13e).

Figure 14a shows the ocean–land difference in CTTice and demonstrates that where convective clouds are more frequent over land, CTTice is generally colder over land by ~2 K (the differences are significant at the 99% level according to a z test). In short, ice clouds are somewhat optically thicker and colder around the cyclone center over land when compared with ocean cyclones. Given that these optically thicker ice clouds are of a more convective nature based on the number of ice-phase tests passed, our results suggest that convective clouds are slightly deeper over land than ocean around the cyclone center. This is consistent with independent observations made in the tropics that convective systems are deeper over land than ocean (e.g., Futyan and Del Genio 2007).

Fig. 14.

Cyclone-centered composites of ocean–land differences in temperature: (a) CTTice (K), (b) 300-hPa MERRA temperature (K), and (c) all cloud tops (CTTupper) between ocean and land cyclones for the medium PW and moderate ω850 category only.

Fig. 14.

Cyclone-centered composites of ocean–land differences in temperature: (a) CTTice (K), (b) 300-hPa MERRA temperature (K), and (c) all cloud tops (CTTupper) between ocean and land cyclones for the medium PW and moderate ω850 category only.

For the stratiform ice clouds along the northeastern edge of the warm frontal region, CTTice is slightly colder over ocean than land by 1–1.5 K, and 300-hPa MERRA temperatures are on average slightly colder as well (Fig. 14b). However, the CTT of all clouds regardless of phase (Fig. 14c) is much greater over ocean and exceeds 5 K differences in some portions of the cold sector. The clouds in the cold sector of ocean cyclones are found at lower and warmer altitudes where supercooled liquid is more likely. When the ice phase is detected, the CTTice is on average colder over ocean by up to 3 K in the northwestern quadrant.

Using the individual cyclones that fall in the medium PW and moderate ω850 category, we calculated the fraction of ice and unknown clouds in proportion to the sum of all three possible phases of cloud for 2-K CTTupper bins between 220 and 280 K. This was performed separately for ocean and land cyclones. Figure 15 shows the relationship between the ratio of ice-phase cloud or unknown-phase cloud to the total cloud as a function of CTTupper. The ratio of ice to total cloud is slightly larger over land (up to 0.05 greater than over ocean) at a given temperature in the range where supercooled liquid clouds are possible down to ~238 K, below which land and ocean relationships are slightly reversed (Fig. 15). In addition, Fig. 15 shows that the ratio of unknown to total cloud as a function of CTTupper differs substantially between land and ocean, with the ocean ratio up to 0.1 larger, down to 235 K, below which the relationship is between land and ocean. Because cyclone gridded AIRS data are used, the same grid box may contain different phase AIRS pixels for the same average CTTupper, especially in scenes with mixtures of cloud types at small scales. Consequently, Fig. 15 shows that at the coldest temperatures the unknown phase is also represented by a small frequency of occurrence, which explains why at 233 K the ice ratio has not reached 1. The greater chance of ice over land in the 233–273-K temperature range of heterogeneous ice nucleation and the even greater chance of mixed/supercooled phase over ocean both suggest a greater propensity for supercooled liquid to persist in ocean cyclones for similar moisture availability and ascent strength. This is consistent with other studies that find a higher occurrence of supercooled clouds over oceans compared to land and also show a greater frequency of supercooled clouds at colder temperatures (e.g., Hu et al. 2010).

Fig. 15.

The relationship between the ratio of either ice or unknown phase to total cloud amount (number of ice or unknown cloud pixels divided by the sum of ice, liquid, and unknown cloud pixels) vs CTTupper for all radial and angular bins within individual cyclones that fall into the medium PW and moderate ω850 category. The black lines are for the ratio of ice to total cloud, and the red lines are for the ratio of unknown to total cloud. The solid lines are for oceanic cyclones, and the dashed lines are for land cyclones. The two vertical dotted lines mark the approximate temperature range for heterogeneous ice nucleation (233–273 K).

Fig. 15.

The relationship between the ratio of either ice or unknown phase to total cloud amount (number of ice or unknown cloud pixels divided by the sum of ice, liquid, and unknown cloud pixels) vs CTTupper for all radial and angular bins within individual cyclones that fall into the medium PW and moderate ω850 category. The black lines are for the ratio of ice to total cloud, and the red lines are for the ratio of unknown to total cloud. The solid lines are for oceanic cyclones, and the dashed lines are for land cyclones. The two vertical dotted lines mark the approximate temperature range for heterogeneous ice nucleation (233–273 K).

There is a distinct possibility of aerosols playing an important role that may explain the observed ice cloud differences over land and ocean. Based on the experiment of Igel et al. (2013), aerosols act to suppress the persistence of supercooled liquid at cloud top in warm fronts but enhance the formation of ice. Thus, it is entirely possible that on average warm fronts over land are more readily glaciated at cloud top than their oceanic counterparts (for very similar ω850 and PW values) because of greater aerosol number concentrations, an increased availability of ice nuclei, or an ability to impact clouds by changing the proportion of supercooled droplets that rime onto graupel and snow versus the formation of small pristine ice crystals that are more likely to be lofted to cloud top.

Simulation studies by Fan et al. (2014) isolated the effect of long-range dust and local pollution aerosol sources on winter cyclones in California, and the results strongly suggest an increase in the mass-mixing ratio of lofted ice particles due to aerosols. The mixing ratio of lofted ice is increased by almost a factor of 10 (cf. Table 2 of Fan et al. 2014). These hydrometeors rather than snow or graupel hydrometeors are more likely to be detected by AIRS because of the MIR sensitivity to small ice particles located near cloud top. Controlling for local air pollution sources, Fan et al. (2014) showed more mixed impacts, with a dependence on whether dust is already present in the cyclone, and also on storm dynamics and environment, but a possible increase in lofted ice is also still possible in this instance. A similar argument might be made from the point of view of convective ice clouds as described by Fan et al. (2013). Rather than “dynamic invigoration” from heavy aerosol loading leading to increased cloud-top height, smaller particle size, and increased anvil area, the primary driver might be microphysics.

Therefore, the somewhat colder cloud-top temperatures and increased convective ice cloud frequency that are found in AIRS-based cyclone composites over land are entirely consistent with the argument for aerosol impacts on ice clouds.

6. Conclusions

We investigated observationally based land–ocean contrasts in extratropical cyclone cloud thermodynamic phase and ice cloud properties obtained with Atmospheric Infrared Sounder, version 6, cloud products. Oceanic Northern Hemisphere winter cyclones are on average more vigorous and contain more moisture than their land counterparts using precipitable water vapor and 850-hPa ascent ω850 from the MERRA reanalysis as proxies for large-scale thermodynamic and dynamic controls on cyclone strength. The total cloud cover within extratropical cyclones is 5%–25% greater over ocean than land. The cloud cover is decomposed into liquid, ice, and unknown thermodynamic phases as obtained with AIRS. In the southwestern quadrant where the largest land and ocean differences are found, the unknown-phase differences dominate, suggesting much more frequent broken liquid phase or mixed phase. The difference in liquid cloud cover between ocean and land is less than 4%, while the difference in ice cloud cover exceeds 4%, in particular in the warm frontal region. To examine if the differences in cyclone properties alone can explain the land–ocean contrast in ice cloud coverage, the impact of cyclone average PW and ω850 on ice cloud properties is determined. While FW07 found a strong impact of cyclone strength (using surface winds as their metric) on high-level cloud cover but little impact from PW, we find that both PW and ω850 impact ice cloud cover. This is expected because the magnitude of PW and air temperature is highly correlated, and cyclones with low PW will tend to be colder and thus favor the production and persistence of the ice phase. By using the individual ice-phase tests of the AIRS phase retrieval, we distinguish between ice clouds of a stratiform nature (large extent but optically thin) and those of a convective nature (more isolated but optically thick). This distinction provides a more detailed understanding of the relative role of cyclone PW and ω850 on ice cloud coverage. The stratiform cloud coverage increases with ω850 in the warm frontal zone and decreases in both the cold and warm sectors as PW increases. In contrast, the impact of PW on convective clouds is less clear, while ω850 is correlated with warm frontal and southeastern quadrant cloud coverage. As PW increases, the area of convective ice cloud coverage decreases but the maximum itself increases. This is seen by examining the impact of PW and ω850 on τice; clouds are thicker with increasing PW but over a smaller area of the southeastern quadrant. The stratiform clouds are correlated with De at cloud top, with smaller values where ice cloud is most frequent, and relatively smaller values in the cold sector compared to the southeastern quadrant, presumably where the effect of lofting of larger ice particles is observed. Finally, as ω850 increases, the cloud-top temperature decreases in the warm front and southeastern quadrant, while the effect of PW is unclear (consistent with FW07 results).

When the strength and moisture in both ocean and land cyclones is constrained, a larger ice coverage over land cyclones is found around the cyclone centers and along the northern edge of the warm fronts. The clouds are more convective around the cyclone center, while the warm front is more stratiform. Around the cyclone center, cloud-top temperatures are colder over land, which suggests that if convection is occurring, it is deeper than over oceans. On the northern edge of warm fronts, the cloud-top temperature is colder over ocean than land. The difference may be related to slightly colder temperatures over ocean than land, as suggested by 300-hPa temperatures. Cloud-top temperatures for all clouds, regardless of thermodynamic phase, are warmer over oceans, suggesting that supercooled clouds persist to colder temperatures in oceanic warm fronts.

Although we have not established a causal relationship between aerosols and warm frontal cloud structures, our results are consistent with previously published work that suggests that greater amounts of aerosols over land may enhance ice formation in contrast with the more pristine oceanic warm frontal clouds.

Acknowledgments

The NASA MAP Climatology of Midlatitude Storminess database can be obtained at http://gcss-dime.giss.nasa.gov/mcms. The authors thank Mike Bauer for providing the ERA-Interim-based cyclone tracks. The MERRA files were obtained from the Goddard Earth Sciences Data and Information Services Center. The AIRS, version 6, datasets were processed by and obtained from the Goddard Earth Services Data and Information Services Center (http://daac.gsfc.nasa.gov/) and the AIRS Project Science and Computing Facility at the Jet Propulsion Laboratory (JPL). CMN was supported by the NASA Science of Terra and Aqua Grant NNX11AH22G and NASA CloudSat Science team recompete Grant NNX13AQ33G. BHK was supported by the NASA Science of Terra and Aqua program Grant NNN13D455T and the AIRS Project at JPL. A portion of this research was carried out at the JPL, California Institute of Technology, under a contract with the National Aeronautics and Space Administration. Government sponsorship is acknowledged.

REFERENCES

REFERENCES
Ackerman
,
T. P.
, and
G. M.
Stokes
,
2003
:
The Atmospheric Radiation Measurement program
.
Phys. Today
,
56
,
38
44
, doi:.
Bauer
,
M.
, and
A. D.
Del Genio
,
2006
:
Composite analysis of winter cyclones in a GCM: Influence on climatological humidity
.
J. Climate
,
19
,
1652
1672
, doi:.
Bodas-Salcedo
,
A.
, and Coauthors
,
2014
:
Origins of the solar radiation biases over the Southern Ocean in CFMIP2 models
.
J. Climate
,
27
,
41
56
, doi:.
Booth
,
J. F.
,
C. M.
Naud
, and
A. D.
Del Genio
,
2013
:
Diagnosing warm frontal clouds formation in a GCM: A novel approach using conditional subsetting
.
J. Climate
,
26
,
5827
5845
, doi:.
Browning
,
K. A.
,
1986
:
Conceptual models of precipitation systems
.
Wea. Forecasting
,
1
,
23
41
, doi:.
Carlson
,
T. N.
,
1980
:
Airflow through midlatitude cyclones and the comma cloud pattern
.
Mon. Wea. Rev.
,
108
,
1498
1509
, doi:.
Catto
,
J. L.
,
C.
Jakob
,
G.
Berry
, and
N.
Nicholls
,
2012
:
Relating global precipitation to atmospheric fronts
.
Geophys. Res. Lett.
,
39
, L10805, doi:.
Chahine
,
M. T.
, and Coauthors
,
2006
:
The Atmospheric Infrared Sounder (AIRS): Improving weather forecasting and providing new insights into climate
.
Bull. Amer. Meteor. Soc.
,
87
,
911
926
, doi:.
Chang
,
E. K. M.
,
Y.
Guo
,
X.
Xia
, and
M.
Zheng
,
2013
:
Storm-track activity in IPCC AR4/CMIP3 model simulations
.
J. Climate
,
26
,
246
260
, doi:.
Chylek
,
P.
, and
C.
Borel
,
2004
:
Mixed phase cloud water/ice structure from high spatial resolution satellite data
.
Geophys. Res. Lett.
,
31
, L14104, doi:.
Colle
,
B. A.
,
Z.
Zhang
,
K. A.
Lombardo
,
E.
Chang
,
P.
Liu
, and
M.
Zhang
,
2013
:
Historical evaluation and future prediction of eastern North American and western Atlantic extratropical cyclones in the CMIP5 models during the cool season
.
J. Climate
,
26
,
6882
6903
, doi:.
Fan
,
J.
,
L. R.
Leung
,
D.
Rosenfeld
,
Q.
Chen
,
Z.
Li
,
J.
Zhang
, and
H.
Yan
,
2013
:
Microphysical effects determine macrophysical response for aerosol impacts on deep convective clouds
.
Proc. Natl. Acad. Sci. USA
,
110
, E4581–E4590, doi:.
Fan
,
J.
, and Coauthors
,
2014
:
Aerosol impacts on California winter clouds and precipitation during CalWater 2011: Local pollution versus long-range transported dust
.
Atmos. Chem. Phys.
,
14
,
81
101
, doi:.
Feser
,
F.
,
M.
Barcikowska
,
O.
Krueger
,
F.
Schenk
,
R.
Weisse
, and
L.
Xia
,
2015
:
Storminess over the North Atlantic and northwestern Europe—A review
.
Quart. J. Roy. Meteor. Soc.
,
141
,
350
382
, doi:.
Field
,
P. R.
, and
R.
Wood
,
2007
:
Precipitation and cloud structure in midlatitude cyclones
.
J. Climate
,
20
,
233
254
, doi:.
Field
,
P. R.
,
A.
Bodas-Salcedo
, and
M. E.
Brooks
,
2011
:
Using model analysis and satellite data to assess cloud and precipitation in midlatitude cyclones
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1501
1515
, doi:.
Futyan
,
J. M.
, and
A. D.
Del Genio
,
2007
:
Deep convective system evolution over Africa and the tropical Atlantic
.
J. Climate
,
20
,
5041
5060
, doi:.
Grise
,
K. M.
, and
L. M.
Polvani
,
2014
:
Southern Hemisphere cloud–dynamics biases in CMIP5 models and their implications for climate projections
.
J. Climate
,
27
,
6074
6092
, doi:.
Hewson
,
T. D.
,
1998
:
Objective fronts
.
Meteor. Appl.
,
5
,
37
65
, doi:.
Hobbs
,
P. V.
,
1978
:
Organization and structure of clouds and precipitation on the mesoscale and microscale in cyclonic storms
.
Rev. Geophys. Space Phys.
,
16
,
741
755
, doi:.
Hoskins
,
B. J.
, and
K. I.
Hodges
,
2002
:
New perspectives on the Northern Hemisphere winter storm tracks
.
J. Atmos. Sci.
,
59
,
1041
1061
, doi:.
Houze
,
R. A.
,
1993
: Cloud Dynamics. International Geophysics Series, Vol. 53, Academic Press, 570 pp.
Hu
,
Y.
,
S.
Rodier
,
K.
Xu
,
W.
Sun
,
J.
Huang
,
B.
Lin
,
P.
Zhai
, and
D.
Josset
,
2010
:
Occurrence, liquid water content, and fraction of supercooled water clouds from combined CALIOP/IIR/MODIS measurements
.
J. Geophys. Res.
,
115
, D00H34, doi:.
Igel
,
A. L.
,
S. C.
van den Heever
,
C. M.
Naud
,
S. M.
Saleeby
, and
D. J.
Posselt
,
2013
:
Sensitivity of warm-frontal processes to cloud-nucleating aerosol concentrations
.
J. Atmos. Sci.
,
70
,
1768
1783
, doi:.
Jin
,
H.
, and
S. L.
Nasiri
,
2014
:
Evaluation of AIRS cloud-thermodynamic-phase determination with CALIPSO
.
J. Appl. Meteor. Climatol
,
53
,
1012
1027
, doi:.
Kahn
,
B. H.
,
A.
Eldering
,
A. J.
Braverman
,
E. J.
Fetzer
,
J. H.
Jiang
,
E.
Fishbein
, and
D. L.
Wu
,
2007a
:
Toward the characterization of upper tropospheric clouds using Atmospheric Infrared Sounder and Microwave Limb Sounder observations
.
J. Geophys. Res.
,
112
, D05202, doi:.
Kahn
,
B. H.
,
E.
Fishbein
,
S. L.
Nasiri
,
A.
Eldering
,
E. J.
Fetzer
,
M. J.
Garay
, and
S.-Y.
Lee
,
2007b
:
The radiative consistency of Atmospheric Infrared Sounder and Moderate Resolution Imaging Spectroradiometer cloud retrievals
.
J. Geophys. Res.
,
112
, D09201, doi:.
Kahn
,
B. H.
, and Coauthors
,
2008
:
Cloud-type comparisons of AIRS, CloudSat, and CALIPSO cloud height and amount
.
Atmos. Chem. Phys.
,
8
,
1231
1248
, doi:.
Kahn
,
B. H.
, and Coauthors
,
2014
:
The Atmospheric Infrared Sounder version 6 cloud products
.
Atmos. Chem. Phys.
,
14
,
399
426
, doi:.
Katzfey
,
J. J.
, and
B. F.
Ryan
,
2000
:
Midlatitude frontal clouds: GCM-scale modeling implications
.
J. Climate
,
13
,
2729
2745
, doi:.
Klein
,
S. A.
, and
C.
Jakob
,
1999
:
Validation and sensitivities of frontal clouds simulated by the ECMWF model
.
Mon. Wea. Rev.
,
127
,
2514
2531
, doi:.
Lau
,
N.-C.
, and
M. W.
Crane
,
1995
:
A satellite view of the synoptic-scale organization of cloud properties in midlatitude and tropical circulation systems
.
Mon. Wea. Rev.
,
123
,
1984
1006
, doi:.
Lau
,
N.-C.
, and
M. W.
Crane
,
1997
:
Comparing satellite and surface observations of cloud patterns in synoptic-scale circulation systems
.
Mon. Wea. Rev.
,
125
,
3172
3189
, doi:.
Nasiri
,
S. L.
, and
B. H.
Kahn
,
2008
:
Limitations of bispectral infrared cloud phase determination and potential for improvement
.
J. Appl. Meteor. Climatol.
,
47
,
2895
2910
, doi:.
Nasiri
,
S. L.
,
V. T.
Dang
,
B. H.
Kahn
,
E. J.
Fetzer
,
E. M.
Manning
,
M. M.
Schreier
, and
R. A.
Frey
,
2011
:
Comparing MODIS and AIRS infrared-based cloud retrievals
.
J. Appl. Meteor. Climatol.
,
50
,
1057
1072
, doi:.
Naud
,
C. M.
,
A. D.
Del Genio
, and
M.
Bauer
,
2006
:
Observational constraints on the cloud thermodynamic phase in midlatitude storms
.
J. Climate
,
19
,
5273
5288
, doi:.
Naud
,
C. M.
,
A. D.
Del Genio
,
M.
Bauer
, and
W.
Kovari
,
2010
:
Cloud vertical distribution across warm and cold fronts in CloudSat data and a general circulation model
.
J. Climate
,
23
,
3397
3415
, doi:.
Naud
,
C. M.
,
D. J.
Posselt
, and
S. C.
van den Heever
,
2012
:
Observational analysis of cloud and precipitation in midlatitude cyclones: Northern versus Southern Hemisphere warm fronts
.
J. Climate
,
25
,
5135
5151
, doi:.
Naud
,
C. M.
,
J. F.
Booth
,
D. J.
Posselt
, and
S. C.
van den Heever
,
2013
:
Multiple satellite observations of cloud cover in extratropical cyclones
.
J. Geophys. Res. Atmos.
,
118
,
9982
9996
, doi:.
Naud
,
C. M.
,
J. F.
Booth
, and
A. D.
Del Genio
,
2014
:
Evaluation of ERA-Interim and MERRA cloudiness in the Southern Ocean
.
J. Climate
,
27
,
2109
2124
, doi:.
Rienecker
,
M. M.
, and Coauthors
,
2011
:
MERRA: NASA’s Modern Era Retrospective Analysis for Research and Applications
.
J. Climate
,
24
,
3624
3648
, doi:.
Rossow
,
W. B.
, and
R. A.
Schiffer
,
1999
:
Advances in understanding clouds from ISCCP
.
Bull. Amer. Meteor. Soc.
,
80
,
2261
2287
, doi:.
Ryan
,
B. F.
,
1996
:
On the global variation of precipitating layer clouds
.
Bull. Amer. Meteor. Soc.
,
77
,
53
70
, doi:.
Shupe
,
M. D.
,
S. Y.
Matrosov
, and
T.
Uttal
,
2006
:
Arctic mixed-phase cloud properties derived from surface-based sensors at SHEBA
.
J. Atmos. Sci.
,
63
,
697
711
, doi:.
Stephens
,
G. L.
, and Coauthors
,
2008
:
CloudSat mission: Performance and early science after the first year of operation
.
J. Geophys. Res.
,
113
, D00A18, doi:.
Stewart
,
R. E.
,
K. K.
Szeto
,
R.
Rainking
,
S. A.
Clough
, and
S. P.
Ballard
,
1998
:
Midlatitude cyclonic cloud systems and their features affecting large scales and climate
.
Rev. Geophys.
,
36
,
245
273
, doi:.
Trenberth
,
K. E.
, and
J.
Fasullo
,
2010
:
Simulation of present-day and twenty-first-century energy budgets of the southern oceans
.
J. Climate
,
23
,
440
454
, doi:.
Tselioudis
,
G.
,
Y.
Zhang
, and
W. B.
Rossow
,
2000
:
Cloud and radiation variations associated with northern midlatitude low and high sea level pressure regimes
.
J. Climate
,
13
,
312
327
, doi:.
Wang
,
Z.
,
D.
Vane
,
G.
Stephens
, and
D.
Reinke
,
2012
: Level-2 combined radar and lidar cloud scenario classification product process description and interface control document. JPL Rep., 22 pp. [Available online at http://www.cloudsat.cira.colostate.edu/sites/default/files/products/files/2B-CLDCLASS-LIDAR_PDICD.P_R04.20120522.pdf.]
Zappa
,
G.
,
L. C.
Shaffrey
, and
K. I.
Hodges
,
2013
:
The ability of CMIP5 models to simulate North Atlantic extratropical cyclones
.
J. Climate
,
26
,
5379
5396
, doi:.