Articles | Volume 3, issue 3
https://doi.org/10.5194/amt-3-655-2010
https://doi.org/10.5194/amt-3-655-2010
07 Jun 2010
 | 07 Jun 2010

Ground-based observations for the validation of contrails and cirrus detection in satellite imagery

H. Mannstein, A. Brömser, and L. Bugliaro

Abstract. Contrails and additional cirrus clouds caused by air traffic have a potential warming effect due to their optical properties and their location in the upper troposphere. The effect of contrails is directly related to their coverage and optical properties, which both can be derived from satellite observations. However, considerable local and global uncertainties remain, as detection limits and efficiency are still unknown. A six months time series of the occurrence of high-level clouds and contrails was analysed visually using an all-sky camera situated at Oberpfaffenhofen (Southern Germany). It shows a contrail occurrence of 21% (fraction of time with visible contrails during one hour) which is nearly constant over daytime and a cirrus occurrence that increases from 27% in the morning to 48% in the evening, suggesting a possible influence of air traffic or, more probably, convective cloud formation. Furthermore, we compared selected all-sky camera images with data of the satellite instruments NOAA/AVHRR and MSG/SEVIRI. As expected, the fraction of contrails visible and detectable in satellite images depends strongly on their width. Of the contrails observed with the all-sky camera of 1–5 km width 60–65% are visually detectable in AVHRR data while only 17% are identified by an automated contrail detection algorithm (CDA). This means that the automated CDA detects approx. 28% of the contrails which are identified by visual inspection in AVHRR data alone. As far as SEVIRI is concerned, visual inspection yields 48% of the contrails of 1–5 km width, the CDA 19%. That means 40% of all contrails visually identifiable in SEVIRI data are found by the automated algorithm. As far as cirrus detection using SEVIRI data is concerned, an automated algorithm tends to overestimate cirrus occurrence but correctly measures cirrus changes during the day compared to visual inspection.

Download