The appearance and disappearance of ship tracks on large spatial scales

The 1-km advanced very high resolution radiometer observations from the morning, NOAA-12, and afternoon, NOAA-11, satellite passes over the coast of California during June 1994 are used to determine the altitudes, visible optical depths, and cloud droplet effective radii for low-level clouds. Comparisons are made between the properties of clouds within 50 km of ship tracks and those farther than 200 km from the tracks in order to deduce the conditions that are conducive to the appearance of ship tracks in satellite images. The results indicate that the low-level clouds must be sufficiently close to the surface for ship tracks to form. Ship tracks rarely appear in low-level clouds having altitudes greater than 1 km. The distributions of visible optical depths and cloud droplet effective radii for ambient clouds in which ship tracks are embedded are the same as those for clouds without ship tracks. Cloud droplet sizes and liquid water paths for low-level clouds do not constrain the appearance of ship tracks in the imagery. The sensitivity of ship tracks to cloud altitude appears to explain why the majority of ship tracks observed from satellites off the coast of California are found south of 35°N. A small rise in the height of low-level clouds appears to explain why numerous ship tracks appeared on one day in a particular region but disappeared on the next, even though the altitudes of the low-level clouds were generally less than 1 km and the cloud cover was the same for both days. In addition, ship tracks are frequent when low-level clouds at altitudes below 1 km are extensive and completely cover large areas. The frequency of imagery pixels overcast by clouds with altitudes below 1 km is greater in the morning than in the afternoon and explains why more ship tracks are observed in the morning than in the afternoon. If the occurrence of ship tracks in satellite imagery data depends on the coupling of the clouds to the underlying boundary layer, then cloud-top altitude and the area of complete cloud cover by low-level clouds may be useful indices for this coupling.