Indonesia, by the location of its geographic and
geologic, it have more potential encounters for natural disasters.
This nation is traversed by three tectonic plates, namely: IndoAustralian, the Eurasian and the Pacific plates.
One of the tools employed to detect danger and send an early disaster warning is sensor device for ocean waves, but it has drawbacks related to the very
limited
time
gap
between
information/warnings

obtained
and
the
real
disaster
event,
which
is only
less
than
30

minutes.
Natural
disaster
early
detection
information
system
is

essential
to
prevent
potential
danger.
The system can make use of the pattern recognition of satellite imagery sequences that take place before and during the natural disaster.
This study is conducted to determine the right wavelet to compress the satellite image sequences and to perform the pattern recognition process of a natural disaster employing an artificial neural network.
This study makes use of satellite imagery sequences of tornadoes and hurricanes.