Tropical
cyclone and associated storm surges result a large damage around the world.
Improved cyclone forecasting techniques need to be developed to reduce the
losses both of biological and physical resources. This writing is one my
research study and it introduces a real-time automated cyclone forecasting
system that is different from existing forecasting techniques. It is an integrated system based-on remote
sensing techniques, machine learning and wireless communication system. This
system has not been tested and implemented.
All the forecasts are not error free and forecasting of tropical cyclone
is not free from error. The accuracy of subsequent forecasts is affected by the
error in registering the initial position and motion of the tropical cyclone. There
are some limitations of the forecasting techniques and lack of proper
understanding of the mechanisms behind the formation and growth of tropical
cyclone. And these matters create the error during cyclone forecasts.
Automated
cyclone forecasting system has been developed but real-time automated cyclone
forecasting system has not been developed. This article aims to introduce an
integrated real-time automated cyclone forecasting system based on remote
sensing techniques, GNSS CORS, machine learning and wireless communication
system.
To form a cyclone there
needs some parameters which play influencing factors to form cyclone and these
factors are used to predict the cyclone event. These are air temperature,
relative humidity (RHUM), wind speed (WS), sea surface temperature (SST) and cloud
pattern. Here it is stated that the data
are collected by the remote sensing sensors and GNSS CORS in this system.
The data acquiring have
to be a continuous process by the sensors and GNSS CORS. Continuously Operating
Reference Stations (CORS) are very effective to provide Global Navigation
Satellite System (GNSS) data consisting of carrier phase and code range
measurements (CORS). CORS helps to
locate the position of the cloud in the sky as well as other atmospheric
parameters which are responsible to form cyclone. The next step of this system
is the Machine Learning process. To learn general regularities hidden in the
data, Machine Learning have been applied to a variety of large datasets
(Mitchell, 1997). In this process
computer can identify the pattern from a large datasets and discover knowledge.
Artificial Neural Networks (ANN) technique can be used in machine learning
process for pattern identification from large data sets.
Figure: Real-Time Automated Cyclone Forecasting System (RACFS)
Source:
Developed by the Author
Kovordányi
and Roy (2009) stated that, the shape and relative position of cumulonimbus
clouds indicate the cyclone tracks and ANN can detect and categorize these features
based on satellite images. In this way ANN would provide valuable input to
automated cyclone forecasting.
Machine
Learning process is the major phase of this system. But it is very complicated
process. Because to fit logic and build algorithm are not so easy at all.
Actually all the data are processed in this stage. After pattern identification
and discovering of knowledge a simulation is created. Identified patterns and
discovered knowledge are stored to a server situated in the weather station for
further analysis by the forecaster. Simulated results are also stored in the
server and broadcasted via Television, Radio, Internet, and SMS for early
warning.
Here
are some problems to information dissemination for early warning of cyclone
event because during a disaster event the supply of electricity is stopped. In
this regard wireless telecommunication system such as mobile phone SMS and web
can be used to information dissemination among the vulnerable area.
This
system is not too easy as like as the diagram to build it. There need huge
amount of economic budget and also need to have the strong technical and
technological resources as like as NASA and NOAA of USA, ESA of Europe and ISRO
of India. It is need efficiency of strong economy, strong technical and
technological resources and good coordination to build this system.
I
cannot say strongly that this system is hundred percent errors free. Because
this system have not run or tested in a weather forecasting center till now. It
has been developed to introduce of my generated idea. Now it has to be tested and implemented its
function and efficiency for real-time cyclone forecasting in a weather
forecasting center.
Author
Mithun Kumar
Remote Sensing Specialist & GIS Developer
Scientific Officer & Head, Aeronautics & Space Applications division
and
Founder & President, Space and Environment Research Center (SERC), Bangladesh.
Founder & Director, Project Origin
SEE ALSO:
Comments
Post a Comment