Date: Fri, August 11, 15:30-17:30
Place: Room An301, An Block, IIS, The University of Tokyo
Invited Speaker: Prof. Ranjana Abhang (Abasaheb Garware College)

Title: Weather Prediction using Artificial Neural Networks

Abstract:
Weather satellites have brought a revolution in Meteorology
(Atmospheric Sciences) too in addition to many other fields.
Depending upon the type of electromagnetic waves used for
transmission, satellite imagery is divided into 4 types-Visible (Vis),
Infrared (IR), Water Vapor (WV) and Microwave (μwave). 
An objective analysis of meteorological satellite images is now
possible with modern fast and powerful computers using automatic
pattern recognition techniques, the foremost of them being the
technique of Artificial Neural Networks (ANN). ANNs have been used
extensively in Satellite Meteorology due to their ability to handle
multi-spectral, multi-temporal, multi-source spatial data more
efficiently and other desirable properties.
A survey of research on ANNs in Satellite Meteorology reveals that
different types of ANNs are used to perform various functions e.g.
analysis of climatological systems, Time Series data, satellite
imagery etc; estimation and prediction of various weather parameters,
classification of clouds etc. These studies are mainly carried out
using Vis, IR and WV images.
The properties of Microwaves are very much suited for cloud
classification. However very rare studies using microwave data have
been carried out until now. In our recent studies it is found to be
feasible to employ ANN to carry out the analysis of weather patterns
based on the International Cloud Classification using images from
microwave satellites also. The special feature of using ANN technique
is that it can classify even noisy, fuzzy or incomplete images due to
the adaptive property of ANN.