SEARCH

SEARCH BY CITATION

Keywords:

  • aerosol cloud interactions;
  • cloud regime changes;
  • indirect effect;
  • radiative forcing;
  • ship tracks

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[1] Documentation of the evolution of ship tracks during 42 h demonstrated that ship emissions are able to convert a marine stratocumulus regime of open cells into closed cells, along with significant negative radiative forcing. This was possible by examining continuous day and night geostationary satellite data that allowed for an uninterrupted documentation of the full life cycle of ship tracks. After nearly one day the ship tracks lost their linear appearance and expanded to cover large areas. These areas, when viewed out of sequence and context, would not be attributable to aerosol perturbations. A rejuvenation of previously dissipated ship tracks in the form of extensive closed cells was also observed. It is suggested that ship emissions may undergo photochemical reactions which nucleate new aerosols that, along with remaining ultrafine particles, grow to CCN that are activated hours later and close the open cells. The added radiative forcing from the closed cells that can be related to the ship tracks, which is mainly coming from the cloud cover effect, may exceed −100 Wm−2, depending on the season and latitude. This implies that anthropogenic aerosols that can be transported from continents through the boundary layer, or travel in the free troposphere and mix with the boundary layer from above, may explain the formation of large closed cells areas that are presently not recognized as originated by aerosol perturbations. The observations reported here pose a demanding test of the ability of cloud resolving models to replicate cloud-aerosol interactions.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[2] Clouds are responsible for 2/3 of the planetary albedo and hence play a dominant role in determining the Earth's energy budget and the global temperature [Trenberth, 2009]. Aerosols can change the microstructure of Marine Stratocumulus clouds (MSC) and lead to an increase in the solar radiation reflected by them back to space, and thus cool the climate system. The most straight forward aerosol impact on cloud albedo is the Twomey effect [Twomey, 1977], whereby for a given liquid water content (LWC), more droplets are nucleated due to higher aerosol concentration that increase the droplets surface area and thus the albedo. The decrease in cloud drop size for the same LWC slows the rate of droplet coalescence and conversion of cloud water into precipitable water, which results in an increase of the liquid water path (LWP), cloud cover and cloud life time [Albrecht, 1989]. The cloud cover effect is in fact an increase of the LWC in the horizontal dimension, while the increase of LWC in the vertical dimension is the so-called LWP effect. It should be mentioned that in non or weakly precipitating closed cells the LWP may decrease with increased droplet concentration, because the smaller droplets evaporate faster and the added cooling induces stronger mixing with the dry air above the inversion [Ackerman et al., 2004; Hill et al., 2009; Xue et al., 2008].

[3] The cloud cover effect is the focus of this study in which we demonstrate by observational means that aerosols can convert partially cloudy areas of MSC into fully cloudy areas. Previous studies have already shown a positive correlation between cloud cover and aerosol optical depth [Kaufman and Koren, 2006; Kaufman et al., 2005; Loeb and Manalo-Smith, 2005; Matheson et al., 2006; Menon et al., 2008; Myhre et al., 2007; Rosenfeld et al., 2006; Sekiguchi et al., 2003], and further the dominance of the cloud cover and LWP effects over the Twomey effect [George and Wood, 2010; Kaufman et al., 2005; Lebsock et al., 2008; Sekiguchi et al., 2003]. However, it is debated whether the larger cloud cover is due to the aerosol effect, or rather due to other causes. Convergence and thickening of the aerosol layer [Mauger and Norris, 2007; Loeb and Schuster, 2008], artifact retrieval of high aerosol optical depth due to swelling of aerosols in more humid air surrounding clouds [Charlson et al., 2007; Koren et al., 2007; Myhre et al., 2007; Twohy et al., 2009; Quaas et al., 2010], and satellite retrieval errors such as cloud contamination or 3-D radiation effects [Loeb and Manalo-Smith, 2005; Zhang et al., 2005; Wen et al., 2007] can also explain this positive correlation between the cloud cover and aerosol optical depth.

[4] The obvious example for the cloud cover effect is ship tracks [Coakley et al., 1987; Durkee et al., 2000; Schreier et al., 2007]. The MSC in which ship tracks form exist under the subtropical highs and the midlatitudes migratory anticyclones. According to Rosenfeld et al. [2006], the structure of the marine boundary layer (MBL) in these areas supports three MSC regimes which are associated with changes in aerosol concentration. High concentration of aerosols supports the formation of closed Benard cells that form nearly full cloud cover. Under lower concentration of aerosols two kinds of broken MSC with significantly lower cloud cover can be defined. The first is large open Benard cells [Agee et al., 1973; Atkinson and Zhang, 1996] that are accompanied with significant drizzle [Comstock et al., 2007; Stevens et al., 2005; Wood et al., 2011]. The drizzle induces gust fronts that induce additional convergence and cloud formation on the colliding zones with neighboring precipitating cells [Feingold et al., 2010]. Therefore, as the open cells are deeper and precipitate heavier, the downdrafts from these clouds are stronger and the distances between the open cells are greater. This is in accordance with Wood and Hartmann [2006], who showed a positive relation between the vertical and horizontal dimension of MSC. They investigated the relationship between cell sizes and the MBL depth for open and closed cells, and showed that the Marine Stratocumulus aspect ratio is typically in the range from 30:1 to 40:1 and is quite insensitive to the regime type. Therefore, in the atmosphere, shallower cells are also less wide. The second type of broken MSC is small open cells, which occur under ultra clean conditions when the MBL is considered to be collapsed due to scarcity of Cloud Condensation Nuclei (CCN) that inhibits cloud formation [Ackerman et al., 1993; Rosenfeld et al., 2006; Wang et al., 2010]. Under these conditions the clouds are relatively thin and therefore the horizontal dimension of the open cells is also smaller.

[5] A progression between the three regimes, from closed to large open cells and further to the small open cells, can occur when the CCN in the MBL are lost by drizzle at a greater rate than their replenishment [Ackerman et al., 1993; Rosenfeld et al., 2006; Wang et al., 2010]. A reversed transition from large open cells to closed cells might be also possible due to the addition of aerosols, as suggested by Rosenfeld et al. [2006]. Simulations of aerosols closing open cells were calculated by Wang and Feingold [2009]who simulated ship tracks by injecting a moderate amount of CCN into the background open cell clouds. However, the ship tracks in their simulations did not self-sustain for many hours.Ackerman et al. [1993] suggested a mechanism for the creation of ship tracks within a collapsed MBL, when small open cells occur (they termed this regime as a “shallow fog layer”). They suggested that adding CCN to such ultra clean conditions by ship emissions can restore the collapsed MBL to form deeper clouds that later become closed cells.

[6] Studies have attempted to estimate the regional and global extent and distribution of ship tracks, mainly in order to estimate their potential radiative forcing. Durkee et al. [2000] used data obtained from polar orbiting satellites and described the composite ship tracks to be 296 ± 233 km long, 9 ± 5 km wide and to last 7.3 ± 6 h, with many ship tracks found to be older than 12 h. Schreier et al. [2010], who used observations from geostationary satellites, further estimate the mean observed length to be 458 ± 317 km, with maximum of 1500 km and mean life time of 18 ± 11 h. These values are larger than those of Durkee et al. [2000], perhaps because Schreier et al. [2010] used measurements with higher temporal resolution, or due to different definition of the measured parameters and the location of the measurements. These characteristics of the ship tracks suggest that the radiative forcing that is caused by ship tracks can be significant, at least over busy shipping regions [Schreier et al., 2006, 2007]. According to Schreier et al. [2007], who presented the first global distribution of ship tracks coverage and the resulting radiative forcing using polar-orbiting satellites, the global annual mean radiative forcing due to ship tracks is considered to be of negligible climate impact. They further added that mixing and dilution of ship emissions with the ambient air can affect the surrounding clouds beyond the recognized form of ship tracks and so increase the climate impact.Peters et al. [2011] attempted to find these wide scale effects of ship emissions and implemented a different method in which, instead of identifying specific ship tracks, they analyzed regions downwind from known busy shipping regions. However, they didn't find statistically significant microphysical and macrophysical impacts. It should be mentioned that their study does not apply to regions where persistent high pressure systems exist, which is where the MSC susceptibility with respect to aerosols is largest.

[7] Whether ship emissions can increase the cloud cover to create radiative forcing that is globally significant is not the aim of this study. In this study we demonstrate the potential impact of aerosols on the MSC by presenting a case study in which local increase of CCN from ship exhausts generated a long-lived regime change from open to closed cells. This single well documented case study is sufficient for demonstrating the possible large impact of aerosols on the MBL clouds. The data and the methods that were used for the 24 h documentation of the ship tracks and the surrounding clouds are described insection 2. Section 3 presents the well documented case study and the conclusions are discussed in section 4.

2. Data and Methodology

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[8] Observations from geostationary and polar-orbiting satellites are used in this study. The Spinning Enhanced Visible and Infrared Imager (SEVIRI) instrument on board the Meteosat Second Generation (MSG) covers the full disk every 15 min and allowed us to document ship track evolution. The SEVIRI is a 12-channel imager consisting of 11 channels with spatial resolution of 3 km at nadir and High Resolution Visible channel (HRV) with spatial resolution of 1 km at nadir. The 11 channels measure radiation in various wavelength bands, of which three are in the solar part of the spectrum, one in the boundary region between solar and terrestrial radiation (3.9μm) and the remaining seven channels measure the outgoing thermal radiation. In addition, cloud products from the MODerate-Resolution Imaging Spectroradiometer (MODIS) detector on board Terra and Aqua satellites were used to retrieve microphysical properties of the clouds at 1 km spatial resolution [Platnick et al., 2003]. Visible quick-looks images of 500 m spatial resolution from the MODIS rapid response system (http://rapidfire.sci.gsfc.nasa.gov/) were used for high quality visual description of the clouds.

[9] The satellite analysis was separated into day and nighttimes, with the solar channels used during the day and the thermal channels during the night. The analysis during the night was based on the night-microphysics RGB (Red-Green-Blue) composite. In this technique, the RGB Brightness Temperature Difference (BTD) of 12.0–10.8μm modulates the red as a measure for the clouds' opaqueness (stronger red means more opaque cloud), BTD of 12.0–3.9 μm modulates the green and is sensitive to particle size (stronger green means smaller droplets) and the 10.8 μm cloud top brightness temperature modulates the blue (stronger blue means warmer clouds) [Lensky and Rosenfeld, 2008]. Therefore, in the night microphysics RGB warm water clouds such as the MSC may appear purple when the cloud drops are relatively large and white when the cloud drops are relatively small.

[10] The cloud microstructure was depicted by the retrieved cloud drop effective radius (re) as obtained from MODIS level-2 cloud products [Platnick et al., 2003]. In order to find the cloud radiative effect (CRE) of a selected area, we must use all the cloudy pixels in that area. Analysis of the MODIS re uncertainties showed that within the area that is covered by broken clouds about 95% of the pixels have uncertainty that is less than 20% while within the area of full cloud cover about 95% of the pixels have uncertainty that is less than 6%. This implies that by using all of the available re measurement the error due to the re uncertainties cannot be too large. Therefore, we preferred to use all the measured re data instead of increasing the uncertainty of our results by using fewer measurements. In order to distinguish between heavily drizzling and other clouds, which relate to open and closed cells, respectively [Comstock et al., 2007; Stevens et al., 2005; Wood et al., 2011], an re of about 15 μm was used as the threshold between the two regimes. This assumption is based on remote sensing measurements and simulations which specify re of near 12–14 μm to be a threshold value above which coalescence creates drizzle and raindrops very quickly [Gerber, 1996; Freud and Rosenfeld, 2012; Rangno and Hobbs, 2005; Rosenfeld, 1999, 2000; Rosenfeld and Gutman, 1994; Rosenfeld et al., 2006, 2012; Suzuki et al., 2010].

[11] The synoptic background for the case study was taken from sets of the National Centers for Environmental Prediction-National Centers for Atmospheric Research (NCEP/NCAR) reanalysis data (http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.html) and includes sea level pressure and 500 hPa geopotential height with 2.5° × 2.5° spatial resolution. The history of the air masses was identified by deriving back trajectories using the National Oceanic and Atmospheric Administration Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) Model (http://ready.arl.noaa.gov/HYSPLIT_traj.php).

[12] In order to estimate the radiative impact of the clouds that relate to ship tracks and other unperturbed clouds we calculated the top of the atmosphere CRE, which is defined as the difference in solar reflected radiation between the cloudy and cloud free situations at the same time and place. The top of the atmosphere CRE was retrieved by using the radiative transfer model of Freidenreich and Ramaswamy [1999]. In this model the ocean albedo is parameterized as a function of solar zenith angle according to the formulation of Taylor et al. [1996]. The atmospheric temperature, water vapor, and ozone profiles are from the midlatitude summer profile of McClatchey et al. [1972]. In the cloud layers, the water vapor mixing ratio is increased to the saturated value. In this implementation of the algorithm, the Slingo parameterization [Slingo, 1989] was used to represent the dependence on cloud re, with cloud optical depth scaled appropriately to the observed optical depth. No background aerosols were included in the simulations. The model inputs were the geographic location, time of the year and the cloud optical depth and re over a selected area, as retrieved from MODIS level 2 cloud products [Platnick et al., 2003]. The radiative fluxes were calculated over the full spectrum every hour over a 24-h period, taking into account the geographic latitude and time in the year. The radiative fluxes were calculated twice, with clouds, based on the cloud optical depth and the re over the selected area, and without clouds. Then, the diurnal CRE was retrieved by the difference between the averages of the calculated net fluxes with and without clouds, and averaged for a 24 h basis.

3. Case Study

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[13] During the 17 to the 19 of January 2006 many ship tracks were observed over the busy shipping region west of the Iberian Peninsula. The analysis of this case study is based on night microphysics RGB and daytime HRV imageries and shows ship tracks that continually form, expand and merge to create an extensive cloud deck, which was found to have large negative radiative forcing. Accompanying MODIS re and HRV images provide the microphysical and dynamical characteristics of the clouds. The evolution of the ship track and the surrounding cloud field can be seen in the 15 min sequences available at the online supporting materials.

3.1. Synoptic Background

[14] The sea level pressure and the 500 hPa geopotential height are shown in Figure 1. The synoptic configuration during the ship tracks creation was characterized by an upper level ridge accompanied with a strong high pressure system at the surface. At the last day of the case study an upper level trough moved eastward accompanying a cyclone in the lower levels. At this time the clouds lost their familiar structure as open or closed MSC. Back trajectories for this event are shown in Figure 1 over the map for 19 January 2006, and indicate that the air masses passed only over the ocean for at least three days prior to the event.

image

Figure 1. Sea level pressure in color and 500 hPa in black contours for 17, 18, and 19 January 2006 at 12Z. The map for 19 January also includes a back trajectory for the previous three days (the triangles over the trajectory represent 6 h intervals). The rectangles represent the area where the ship tracks were observed.

Download figure to PowerPoint

3.2. The Night Between 17–18 January 2006

[15] Ship tracks started to appear crossing the east Atlantic Ocean on the night between the 17 and 18 of January 2006. Figures 2 and 3 show the night microphysics RGB images for 17 January at 20:57 GMT and 18 January at 6:57 GMT, respectively. The ship tracks in this RGB can be seen as bright bands due to the reduced re. Please refer to section 2 for description of the color combination and its physical meaning.

image

Figure 2. Ship tracks start forming during night over the northeast Atlantic. The shown scene is roughly located between 50°N to 32°N and 33°W to 8°W. The Iberian Peninsula and Morocco are seen in the east part of the image. In this “Night Microphysical” color scheme, water clouds with small drops appear white, and water clouds with large drops appear purple. The ocean is colored dark blue. The open cells look purple due to their large cloud drop size and the small areas of sea surface that is seen through their open centers. The white areas are closed cells with small cloud drops. Ship tracks within the closed cells appear brighter due to their smaller cloud drops (point “a”). The ship tracks within open cells appear more conspicuously as bright lines in the purple areas. The image is based on the geostationary satellite METEOSAT-8 on the 17 January 2006, 20:57 UTC. Triangles represent the heads of the ship tracks.

Download figure to PowerPoint

image

Figure 3. Same as Figure 2, but for 18 January 2006, 06:57 UTC. Note the west-east pair of tracks (“a”). Ship tracks that form to the north of them appear to deepen the small open cells (“b”), whereas those to the south appear to close large open cells (“c”).

Download figure to PowerPoint

[16] The visual difference between small and large open cells is marked by arrows that refer to each one in the satellite images. Large open cells are recognized in the night microphysics RGB by broken purple clouds that resemble the cells' structure and their relatively large re (Figure 3), while in the visible images they appear as organized polygons that have larger horizontal dimension that corresponds to their greater depth (Figure 4) [Wood and Hartmann, 2006]. Note that the polygon structure of the large open cells can be seen more clearly in the MODIS HRV image (Figure 5a) and that they are characterized by relatively large re (Figure 5b). The small open cells can be recognized in the night microphysics RGB by the granular texture of the clouds (Figure 3).The documentation of the transition from night to day allowed us to see that the small open cell areas as seen in the night microphysics RGB are seen in the day HRV images as faint granular clouds, that probably have extremely small cell structure that reflects the small vertical scale of the clouds [Wood and Hartmann, 2006] (Figures 4 and 5a). Following Wood and Hartmann [2006], who showed that the Marine Stratocumulus aspect ratio (cell center separation/MBL depth) is typically in the range from 30:1 to 40:1, we assumed that the smaller open cells are shallower. This led us to determine an arbitrary threshold in which the small open cells are less than half of the depth of the large open cells. It is important to mention that as the open cells get smaller, the MBL approaches the collapsed stage [Ackerman et al., 1993; Rosenfeld et al., 2006;], and we cannot determine whether the transition between the large and small open cells is sharp or rather gradual. Analysis of MODIS resupports the assumption that the small open cells exist under ultra clean conditions. This will be discussed in the next sub-section. Note that due to partial pixel filling in the small open cells, they are hardly noticeable in the 1 km resolution of the MODIS re cloud product, as shown in Figure 5b.

image

Figure 4. High resolution visible image taken by METEOSAT-8 on 18 January 2006, 10:57 UTC. The shown scene is located roughly between 43°N to 32°N and 23°W to 13°W. Note the cellular structure of the large open cells and the granular structure of the small open cells. The west-east pair of tracks in the middle of the image (“a”) maintained its linear structure for about 12 h. The ship tracks that form to the north of the west-east pair of tracks appear to deepen the small open cells (“b”) whereas those to the south appear to close large open cell (“c”). Note junctions where ship tracks crossed each other within the large open cells, creating larger areas of closed cells (area “d” for example). Point “e” represents the area in which rejuvenation of ship emissions was observed.

Download figure to PowerPoint

image

Figure 5a. MODIS Aqua satellite image with resolution of 500 × 500 m of ship tracks in marine stratocumulus in an area of about 900 × 650 km in the northeastern Atlantic Ocean on 18 January 2006 14:30 UTC. The shown scene is located roughly between 38°N to 32°N and 23°W to 14°W. The cloud radiative effect (in Wm−2) is given for the marked rectangles. The pair of the west-east tracks is marked by the cloud radiative effect of −82 Wm−2. Area “d” is where several ship tracks crossed each other and created larger closed cells area within a large open cells area. Area “f” is small open cells over which an analysis of the re spatial distribution was made for determining the possible biases of its retrievals.

Download figure to PowerPoint

image

Figure 5b. Same as Figure 5a but for the effective radius. Latitudes and longitudes are shown with 2 degrees intervals. The ship tracks appear as a marked decrease in cloud drop effective radius (re in μm).

Download figure to PowerPoint

[17] The attention in Figures 2 and 3is focused on tracking a pair of closely parallel ship tracks that appear in a west-east orientation within the area that undergoes a transition from closed to open cells. This area further develops with time into small open cells regime that is clearly identified as the aerosol starved cloud regime (point “a” inFigures 2 and 3). The regime change can be seen by the change of the smooth texture of the bright areas to a coarse texture due to the exposure of the sea surface. The fact that this area undergoes a cleansing process and then returns to a closed cells regime, as can be seen by the creation of these ship tracks, is highly important because under such clean conditions addition of aerosol can explain the formation of the closed cells regime [Ackerman et al., 1993, 2000; Rosenfeld et al., 2006; Taylor and Ackerman, 1999; Wang et al., 2010]. In Figure 3it can be seen that the west-east pair of ship tracks grows to be about 750 km long just after 12 h from their initiation, and that more ship tracks are being created overnight within the small open cells in the northern area (area “b” inFigure 3). The ship tracks that are being created far to the south seem to close large open cells (area “c” in Figure 3) as they create continuous bright cloud lines with reduced re. These three areas (a, b and c) compose the initial elements of the future created cloud field, as will be shown next. The sequence further shows that over the night new ship tracks cross older ship tracks while adding to them more aerosols. These aerosols reinforce the maintenance of the older ship tracks as closed cells.

3.3. Day of 18 January

[18] The transition from night to day is shown in Figure 3 and 4 for the time steps of 6:57 and 10:57 GMT, respectively. Figure 6 shows the HRV image of 15:57 GMT for that day. During that day the ship tracks in the northern part, where the small open cells exist, seem to become thicker as they form deeper clouds, while in the southern part the ship tracks close large open cells, as clearly supported by the image sequence. Within the large open cells in the southern part, several ship tracks crossed each other to create larger areas of closed cells, as can be seen in the sequence. One such junction is marked by the letter “d” in Figures 46that show the mid-day MODIS HRV and re imageries, respectively. The large extent of this area and the smaller re can be related to the fact that higher concentration of aerosols is present because of the higher amount of emitted aerosols in the same area. This implies that when the aerosol concentration is higher, the influence on the cloud microphysics and dynamics is greater.

image

Figure 6. Same as Figure 4 but for 18 January 2006, 15:57 UTC. Note the expansion of the pair of tracks in the middle of the image (“a”), the ship tracks that thicken the small open cells (“b”) and those that close large open cells (“c”). Also note junctions where ship tracks crossed each other within the large open cells, creating larger areas of closed cells (area “d” for example). Point “e” represents the area in which rejuvenation of ship emissions was observed.

Download figure to PowerPoint

[19] The pair of parallel ship tracks from the previous night can be seen in the middle of the area stretching west-east through different types of broken cloud cover (Figures 46). Most of these broken clouds are composed of small open cells in which their small horizontal dimension probably reflects their respective small vertical dimension [Wood and Hartmann, 2006]. Such clouds apparently exist in ultra-clean air with collapsed MBL that prevents cloud formation [Ackerman et al., 1993; Mauritsen et al., 2011; Rosenfeld et al., 2006]. It can be seen by the texture of the clouds, as well as by their horizontal extent, that the small open cells are more dominant north to the parallel pair of ship tracks in the middle of the area, while the large open cells are more dominant south of them. MODIS re imagery (Figure 5b) provides further insight to the cloud microphysics that shows smaller re over the ship tracks compared to the surrounding clouds. It is also possible to see that the re increases toward the tails of the ship tracks (i.e., their older parts), and when it reaches the size of about 20 μm the ship tracks break up, implying a rainout cleansing process as suggested by Ackerman et al. [1993] and Rosenfeld et al. [2006].

[20] The re of the open cells is found to be higher than the 15 μm threshold for significant drizzle [Gerber, 1996; Freud and Rosenfeld, 2012; Rangno and Hobbs, 2005; Rosenfeld, 1999, 2000; Rosenfeld and Gutman, 1994; Rosenfeld et al., 2006, 2012; Suzuki et al., 2010]. However, it has been argued that due to the broken nature of the cloud cover of open cells, partial pixel filling or 3-D effects may reduce the reliability of the re retrievals [Loeb and Manalo-Smith, 2005; Wen et al., 2007; Zhang et al., 2005]. Such a bias should be detectable when the re is shown as a function of the size of contiguous clouds in the broken cloud area. In the case of the area of small open cells that is marked by the letter (f) in Figures 5a and 5b, the average re over the whole area was 23 μm. The largest cluster with measured re in this area was a pixel which is surrounded by 6 pixels that also had retrieved re. The analysis shows that the value of the re does not depend on the number of contiguous cloudy pixels within the small open cells, and varies within a range of 2 μm. This might be caused by the very small cloud size of the small open cell elements, which are nearly uniformly distributed within a MODIS pixel. For comparison, we did the same analysis for a large open cell area in which the average re was 27 μm. The re in the isolated cloudy pixels of these open cells was found to be on the average larger by 3 μm than fully surrounded cloudy pixels. Therefore, if the partial filling of the pixels incurs a bias in the retrieved re, the variability of this bias is within 3 μm, indicating that the mean re of 23 μm for the small open cells might be corrected to 20 μm at most. An re of 20 μm is considered to be significantly larger than the threshold of 15 μm, which account to heavy drizzle. This implies that the area that was identified as small open cells is indeed an ultra-clean and collapsed MBL cloud regime and that the emitted aerosols from the ships restored its closed cell regime.

3.4. The Night of 18–19 January

[21] The cloud field at the end of the following night is shown in Figure 7 for 19 January at 6:57 GMT. In this Figure the upper rectangle marks closed cells that can be related to the ship tracks within the small open cells (relate to area “b” in Figures 3, 4 and 6). These ship tracks were extended and spread to cover a larger area. The middle rectangle marks the parallel pair of ship tracks that merged together and were crossed by other ships to form large closed cell area while losing its linear ship track structure (relate to area “a” in Figures 24 and 6). The lower rectangle marks the ship tracks that closed large open cells (relate to area “c” in Figures 3, 4 and 6). These were also expanded to cover large areas with closed cells. Please refer to the image sequence in the online supporting materials for better representation of the evolution of these areas.

image

Figure 7. Same as Figure 2, but for 19 January 2006, 06:57 UTC. The shown scene is located roughly between 42°N to 26°N and 32°W to 9°W. The Iberian Peninsula and Morocco are seen in the east part of the image. Northwestern Africa and the Canary Islands are seen in the east and middle part of the image. The northern rectangle is related to the ship track within the small open cells, the middle one to the east-west pair of ship tracks and the southern one to the ship tracks that closed large open cells, as can be tracked backward in the previous figures and in the image sequence that is available in the supporting online materials. The ship tracks rotated clockwise with the anticyclonic flow.

Download figure to PowerPoint

3.5. Day of 19 January

[22] In the following morning an extensive cloud cover with dimensions of up to 800 km wide and long was observed in a structure of the three nearly horizontal bands of closed cells (Figure 8). These bands can be attributed to the ship tracks that were created within the small open cells in the north, the parallel pair of ship tracks in the middle, and those that were created within the large open cells in the south (the rectangles in Figure 7). MODIS high resolution visible and reimageries for the mid-day (Figures 9a and 9b, respectively) show that the large closed cells MSC that are related to the ship emissions have re smaller than 15 μm, the threshold which below MSC are consider being low-precipitating closed cells. During the day the vast cloud cover was observed to be gradually reduced, as can be seen in the image sequence. Note that during the whole event the clouds moved with an anticyclonic rotation that indicates the favorable meteorological conditions for ship tracks.Figure 1 shows the synoptic conditions in which the area where the ship tracks formed is located in the middle of the anticyclone.

image

Figure 8. Same as Figure 4but for 19 January 2006, 10:57 UTC. The shown scene is located roughly between 39°N to 30°N and 24°W to 15°W. Note the expansion of the ship tracks and their loss of their linear structure to form amorphous clouds. The northern rectangle is related to the ship tracks within the small open cells, the middle one to the east-west pair of ship tracks and the southern one to the ship tracks that closed large open cells, as can be tracked backward in the previous figures and in the image sequence that is available in the supporting online materials.

Download figure to PowerPoint

image

Figure 9a. MODIS Terra satellite image of marine stratocumulus in an area of about 450 × 550 km in the northeastern Atlantic Ocean on 19 January 2006 12:00 UTC. The shown scene is located roughly between 36°N to 31°N and 24°W to 18°W. The cloud radiative effect (in Wm−2) is given for the marked rectangles. The larger rectangles are associated to former ship tracks that developed to form amorphous cloud structures.

Download figure to PowerPoint

image

Figure 9b. Same as Figure 9a but for the effective radius. Latitudes and longitudes are shown with 2 degrees intervals. Note the smaller re over the areas which are associated to the former ship tracks.

Download figure to PowerPoint

3.6. Rejuvenation of Ship Tracks

[23] Hill et al. [2008] found a diurnal cycle of the MSC due to competition between cloud top long wave cooling and day time solar heating that cause a peak in the cloud fraction at night and early morning. This diurnal cycle is described by Sharon et al. [2006] as cloud filling of broken cloud cover that favors to occur at night, due to the minimum of the diurnal solar radiation that allows more effective radiative cooling and mixing. A similar cloud filling is also observed in the case study here, however, in this case study the cloud filling is observed to occur also during the day time, evidence that the filling cannot be caused only by diurnal enhanced radiative cooling, but rather due to other factor, possibly such as photochemical production and subsequent growth of additional aerosols. Bretherton et al. [2004] found a correlation between short wave radiation and condensation nuclei (CN) in the MBL, apparently resulting from photochemical reactions. Kazil et al. [2011] simulated these reactions and showed that during the day SO2, originating from dimethyl sulfide (DMS), oxidizes to H2SO4 to form new aerosols.

[24] Sources for gaseous phase aerosols can be DMS, anthropogenic aerosols transported from land or ship emissions. In the present case study the cloud filling is observed to occur in air masses in which ship tracks are observed up to 12 h prior to the filling. The area that is marked by the letter “e” in Figures 4 and 6 (corresponding to 18 January at 10:57 and 15:57 GMT, respectively) symbolizes the area where the broken cloud cover appears to be closed spontaneously. Refer to the corresponding area in the image sequence starting at 18 January at 9:57 GMT for better representation of the cloud filling with time. Tracking back to the previous time steps with this specific air mass shows ship tracks that dissipated up to 12 h before, suggesting that the ship emissions experienced photochemical reactions to create CCN that nucleated to form additional cloud drops that caused the cloud filling.

[25] However, not all the aerosols must be newly formed. The high concentration of small aerosols in the original ship track may incur a competition for water vapor and a reduction of the supersaturation, so that activation of the smaller particles is deferred for hours later, after the particles undergo some growth, possibly facilitated by photochemical reactions. This mechanism is possible in the presented case study due to the large number of previous ship tracks that were observed, which probably emitted large amount of aerosols.

[26] In the following night (the night between the 18 and 19 of January) the cloud filling continues to be observed to close gaps between the ship tracks. These areas persist as closed cells into the next day. It is interesting to note that the cloud filling is especially evident within the area of the heavy ship traffic. This implies that at least in this case, the night cloud filling is being further enhanced with the renewed supply of aerosols that originated from ship emissions in areas of previously dissipated ship tracks.

3.7. Radiative Forcing

[27] The cloud radiative effect (CRE) is calculated over the vast cloud cover that is observed in the last day of this case study, which is shown here to be caused by the expansion of ship tracks and their rejuvenation. Figure 9a shows the CRE for the three large areas of full cloud cover that is observed about 40 h after the first ship tracks appeared. The cloud cover had CRE with values of −78, −77 and −85 Wm−2. This is a radiative forcing, which is the difference between the CRE in the closed and open cells, of about 40 Wm−2 in excess of the nearby clouds. For comparison, the radiative forcing of the ship tracks in the previous day was 60 Wm−2 in excess of the nearby open cells (Figure 5a), implying an even larger CRE. Note that the calculated CRE is made for this specific case, which occurred in winter time when the solar radiation is relatively low (latitudes 30°–40°N). In order to give a comparable standard CRE, the CRE in Figure 9a was calculated for the same clouds assuming solar radiation of an equinoctial day and of a summer solstice day. The results show radiative forcing of about 70–80 Wm−2 in excess of the nearby broken clouds for the equinoctial day, and of about 110–120 Wm−2 for the summer solstice day.

[28] Narrow cloud free strips were seen along some of the ship tracks (Figure 5a) and offset some of the enhancement of the CRE over the total area. However, as can be seen in this case study, these cloud free margins occur only in a small fraction of the ship tracks. Moreover, as the ship tracks expand, these cloud free margins become insignificant compare to the ship tracks width.

4. Discussion and Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[29] The presented case study shows the ability of aerosols from ship emissions to change broken cloud regimes of MSC into closed cells regime with full cloud cover over extensive areas.

4.1. The Observations of Ship Tracks Closing Open Cells

[30] We used observations obtained from the MSG geostationary satellite and applied a method for nighttime microphysical observations. This allowed us to document precisely the clouds evolution throughout the 24 h diurnal cycle and to demonstrate that ship emissions clearly close large open cells, and form closed cells within small open cells, which is an ultraclean and collapsed condition of the MBL. The importance of sufficient amount of aerosols to cause a regime change from large open cells to closed cells is demonstrated by the creation of large and sustained closed cell areas where ship tracks crossed each other. In these areas a larger amount of aerosols from numerous ship tracks closed deep open cells for longer time than the surrounding individual ship tracks, which dissipated much sooner. Regarding the small open cells, where scarcity of CCN incurs a collapse of the MBL and inhibition of cloud formation, the addition of aerosols appears to restore the MBL in the regime of closed cells. The inference that ultra clean conditions are manifested by the small open cells is supported by the ship tracks of closed cells that form in them. The formation of such ship tracks within a collapsed MBL was already observed in previous studies [Ackerman et al., 1993; Christensen and Stephens, 2011]. Further research is needed to clarify the distinction between large and small open cells of MSC, which are often just referred to as open cells.

[31] The aerosol effect on the cloud microstructure is expressed by the clear difference in the re between the ship tracks and the surrounding clouds, as the re is significantly smaller within the closed cells that constitute the ship tracks. This supports the hypothesis that the microphysical effects of the aerosols triggered internal mechanisms of feedbacks that lead to changes in the MSC cloud regime [Rosenfeld et al., 2006]. Interestingly, ship tracks in which the re increased with time to above 15 μm started to break up into open cells, and completely dissipated when the re reached 20 μm. Considering that closed cells are characterized by light or no drizzle and open cells by heavy drizzle [Comstock et al., 2007; Stevens et al., 2005; Wood et al., 2011], the breakup of the closed cells-ship tracks with re between 15 to 20 μm highlights the role of drizzle formation in the mechanisms of regime changes [Ackerman et al., 1993; Rosenfeld et al., 2006; Stevens et al., 2005; Xue et al., 2008], and a threshold re value for significant drizzle [Gerber, 1996; Freud and Rosenfeld, 2012; Rangno and Hobbs, 2005; Rosenfeld, 1999, 2000; Rosenfeld and Gutman, 1994; Rosenfeld et al., 2006, 2012; Suzuki et al., 2010].

[32] Large areas of open cells were observed to close in areas of previously dissipated ship tracks, possibly by aerosols that were too small for nucleating cloud drops in the original ship tracks. These aerosols, along with newly nucleated aerosols stemming from photochemical reactions of the gaseous ship emissions, could grow to CCN size and be responsible for the rejuvenation of the closed cloud regime. The rejuvenation of ship tracks was observed during both night and day. The rejuvenation during afternoon eliminates the possibility that this cloud filling could be caused by diurnal enhanced radiative cooling, as previously suggested by Hill et al. [2008] and Sharon et al. [2006]. Peters et al. [2011], who attempted to find similar effects on clouds downwind of main shipping corridors in the tropics didn't find statistically significant microphysical and macrophysical impacts. However, their study does not apply to regions where the MSC susceptibility with respect to aerosols is largest. Therefore, in our view the study of Peters et al. [2011] does not represent adequately the potential impacts of ship emissions.

4.2. The Cloud Radiative Forcing and Its Extent

[33] The cloud radiative effect over the closed cells that were related to form by ship emissions i.e., (i) by closing large open cells; (ii) by creating closed cells in small open cells, and (iii) by rejuvenation of ship tracks, was found to be much larger over the closed cells in relation to the background clouds. This is in fact an aerosol cloud-mediated forcing, which may even exceed 100 Wm−2, depending on the season and latitude.

[34] The areas subject to such large radiative forcing due to ship tracks can be quite large and may reach 105 km2. The ship tracks that were observed to last hours after their creation lost their linear appearance. They expanded far beyond their initial shapes while gaps between adjacent ship tracks became closed. As a result, large areas of closed cells were created. These areas, when viewed out of sequence and context, would not be attributable at this stage to aerosol perturbations. The same applies to clouds that are formed by rejuvenation of ship tracks.

4.3. Limitations in Simulating Aerosols Closing Open Cells

[35] Wang and Feingold [2009]simulated ship tracks by injecting CCN into the background of open cell clouds. They found that the open cells turned into solid deck but not manifested as real closed cells that can self-sustain for many hours. It appears that the “memory” of the open cells dynamics cannot be easily erased, unless the large open cells eventually dissipate or the boundary layer collapses into small cells. Then, adding aerosols may create clouds that no longer have the memory of the circulation of the open cells. In contrast, our observations show a long-lived regime change of ship tracks closing both small and large open cells. This discrepancy between observations and simulations indicates a major gap in our understanding of how to simulate the MSC regime changes. This, in turn, limits the ability to asses related climate impacts. Obviously, additional simulations that are intimately linked to observations are required. Furthermore, the atmospheric general circulation models treat processes of cloud-aerosol interactions in a simplified manner, limiting the confidence with which conclusions can be drawn. Therefore, high priority should be given to the challenge of more realistic representations of cloud-aerosol interactions in climate models.

4.4. Concluding Remarks

[36] The presented case study was chosen to demonstrate the potential of aerosols to convert marine stratocumulus regime of large and small open cells into closed cell regime, along with significant negative radiative forcing. This implies that major climate impacts that are yet not recognized may incur when large quantities of aerosols arrive to the MBL. Such aerosols can be transported from continents through the boundary layer or travel in the free troposphere and mix with the MBL from above. If anthropogenic aerosols do induce closing of MSC at large areas, it should incur large negative radiative forcing at a global scale. In such case, this effect may mask part of the warming due to greenhouse gases.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References

[37] The authors thank Guy Sapir for his support in analyzing the data, Carynelisa Erlick, who set up the radiative transfer algorithm for the forcing calculations, and Jochen Kerkmann from EUMETSAT for helpful access to the satellite data.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methodology
  5. 3. Case Study
  6. 4. Discussion and Conclusions
  7. Acknowledgments
  8. References