1. Using Ship Tracks to Characterize the Effects of Haze on Cloud Properties Matthew W. Christensen, James A. Coakley, Jr., Matthew S. Segrin, William R. Tahnk College of Oceanic and Atmospheric Sciences, Oregon State University Objective: Determine how clouds respond to haze pollution. In particular, how does haze pollution affect cloud liquid water and fractional cloud cover. 1. Data: Terra MODIS radiances for May – August (2001 – 2003). 20º - 45 º N, 110 º - 140 º W (Eastern Pacific). Properties derived using partly cloudy pixel retrievals (Coakley et. al., JAOTech., 22, 3, 2005): Visible optical depth,  c Droplet effective radius, R e Liquid water, Fractional cloud cover, A c 2. Location of ship tracks: Each track was located and logged by hand using 2.1 and 3.7-μm imagery data. An automated track finding scheme was then used to locate the tracks based on track location log. The scheme identifies the portions of the cloud polluted by the ship. This scheme also selects nearby unpolluted clouds so that they are close to the clouds identified as being polluted. 3. Example of ship track analysis Image was constructed from 2.1-μm radiances illustrating the analysis of a single ship track. Colored pixels indicate the cloud droplet effective radius and are selected using the automated track finding scheme. 0.64-  m reflectances2.1-  m reflectances Droplet sizes for the first 180 km of the ship track are noticeably smaller than both controls. At the tail of the track the droplet sizes tend to merge together and the ship track disappears. Optical depths in the ship track pixels are larger than in both sets of control pixels at the head of the track. In the middle and tail of the track the difference is smaller, but the optical depths for the ship track are generally larger. No apparent difference exists between the ship track and control pixels for cloud liquid water path. 5. Fractional cloud cover Fractional cloud cover in the ship track pixels is generally greater than in the control pixels. Acknowledgements: This work was supported through NASA Radiation Sciences Grant NNG04GF42G. 3.7-  m radiance Droplet radius 0.4 < A c < 0.80.8 < A c < 1.0 Overcast Droplet Radius Means and standard deviations for the distributions of droplet effective radii for the ship track pixels (red line) and the control pixels (dotted blue line). Terra droplet radii are nearly identical except for 0.4 < A c < 0.8, for which morning clouds have droplets with radii that are about 1 μm larger. Droplets in polluted clouds exhibit much narrower distributions  droplet growth inhibited in polluted clouds. Droplets larger in clouds that only partially cover 1-km pixels  dissipation is through drizzle. Differences in Optical Depth Means and standard errors for the distributions of optical depth for polluted − controls (red line) and control 2 − control 1 dotted (blue line). Optical depth changes for Terra about ~10% smaller  morning clouds are less sensitive to pollution than the thinner afternoon clouds. Increase in optical depths caused by pollution highly significant, the changes increasing as pixel-scale cloud cover fraction decreases  clouds filling larger areas have larger optical depths. 0.4 < A c < 0.80.8 < A c < 1.0 Overcast Liquid Water Means and standard deviations for the distributions of liquid water amounts for the ship track pixels (red line) and the averaged control pixels (dotted blue line). Terra liquid water amounts ~10% larger  morning clouds are thicker than afternoon clouds. For broken cloud conditions, polluted clouds have more liquid water than nearby unpolluted clouds and 20-pixel track segments with A c < 0.8 have clouds with larger liquid water amounts than those in segments with 0.8 < A c < 1.0  dissipation of clouds is through drizzle and suggests that regions with broken clouds have moisture support in the overlying troposphere. 0.4 < A c < 0.80.8 < A c < 1.0 Overcast Cloud cover fractions is nearly identical for Terra. For broken cloud conditions, polluted clouds have greater pixel-scale cloud cover than nearby unpolluted clouds, qualitatively consistent with LES model results. (Ackerman et al., Geophys.Res.Lett.,30,10,2003) 0.4 < A c < 0.80.8 < A c < 1.0 0.4 < A c < 0.80.8 < A c < 1.0 Overcast Differences in Liquid Water Means and standard errors for the distributions of liquid water amounts for polluted − controls (red line) and control 2 − control 1 dotted (blue line). For overcast conditions, polluted clouds have less liquid water than nearby unpolluted clouds  overlying free troposphere sufficiently dry that the increased entrainment in clouds with smaller droplets leads to the drying of polluted clouds as suggested by the results of a large eddy simulation (LES) model (Ackerman et al., Nature,432,1014,2004). Liquid water amount changes for Terra are about 20 − 40% larger so that loss of liquid water is greater for overcast clouds and gain of liquid water is greater for clouds that only partially cover the 1-km pixels. Overcast pixels lose liquid water while partially cloudy pixels gain liquid water when polluted  dissipation is through drizzle and is supported by a relatively moist overlying troposphere. 6. Summary Droplet growth is inhibited in polluted clouds. For overcast conditions, polluted clouds lose liquid water consistent with LES model results for sufficiently dry overlying troposphere air. Marine stratus appear to dissipate through drizzle which may be supported by relatively moist overlying troposphere air. Under broken cloud conditions, haze pollution causes increases in cloud cover, cloud optical depths, and liquid water amounts. 4. Influence of ship plumes on fractional cloud coverage Cloud properties averaged over 20-pixel long segments aligned along the ship tracks for three years of summertime observations from both Terra and Aqua. Segments had to have at least 20 pixels identified as polluted and as unpolluted on both sides of the track. The number of control pixels on both sides of the track had to be at least 70% of the number of polluted pixels. Segments were selected in which fractional cloud cover for the control pixels ranged between 40-80%, greater than 80% but not overcast, and overcast. Aqua-MODIS (afternoon clouds) results are shown. Terra-MODIS (morning clouds) results are similar and not shown. 0.4 < A c < 0.80.8 < A c < 1.0 Overcast 7. Future work Investigate the sensitivity of polluted clouds to additional particle loading – as occurs when ship tracks cross. Optical Depth Means and standard deviations for the distributions of optical depth for the ship track pixels (red line) and control pixels (dotted blue line). Terra optical depths are larger by ~2 for A c > 0.8  morning clouds are thicker than afternoon clouds. Difference between optical depths for polluted and nearby unpolluted clouds increases as the fractional cloud cover of the unpolluted clouds decreases  clouds filling larger areas have larger optical depths.