Recognition and inventory of oceanic clouds from satellite data using an artificial neural network technique
Satellite remote sensing can be used to monitor the genesis and evolution of oceanic clouds. In order to link cloud genesis and dimethylsulphide production it is necessary to classify and extract the properties of cloud fields accurately in imagery. An experiment is reported in which an artificial neural network has been used to identify the main types of clouds found in NOAA/AVHRR imagery of the northern Atlantic Ocean. The method is based on the use of an unsupervised pattern recognition approach: the topological (or self-organising) map neural network. Such a network has been trained with 362 examples of 40 x 40 cloud fields extracted from imagery and has been used to provide 25 different cloud classes.
Bibliographic Reference: Paper presented: International Symposium on Dimethylsulphide, Oceans, Atmosphere and Climate, Belgirate (IT), October 13-15, 1992
Availability: Available from (1) as Paper EN 37207 ORA
Record Number: 199211461 / Last updated on: 1994-11-29
Original language: en
Available languages: en