MU developing early flood warning system

Konsam Samita
IMPHAL, Apr 1 : A highly efficient AI model for predicting flood in advance is being developed at the Computer Science Department, Manipur University, Canchipur.
With an efficiency ratio of 0.97 (0 being the lowest and 1 being the highest), the theoretical work has been completed, and a model based on it is being developed, says Leishangthem Sashikumar, a research scholar at the Computer Science Department.
Sashikumar is developing the AI (Artificial Intelligence) application/model under Dr Khumukcham Robindro, Head of Department of Computer Science.
In theory, this AI model could predict floods with high accuracy. This could help in installing a reliable early warning system for floods in Manipur, says Sashikumar.
Sashikumar has completed his thesis on the same subject of developing an early flood warning system.
Explaining the system in simple terms, he said, the AI model would be fed with data, for example water levels of rivers, precipitation rate etc, and then it would process it to predict water levels of the rivers for the next 3 days.
Elaborating how the system would work in theory, he says, the AI model will incorporate deep learning to make the predictions.
He explained, deep learning is a specialized machine learning that uses artificial neural networks to learn from data and make predictions.
Sashikumar said the preparation of the thesis took him 6 years and the case study was carried out using data on water level of rivers at six gauge sites (locations where water levels are observed and measured).
For his study, Sashikumar used gauge sites at Minuthong bridge, Thong Nambonbi, Lilong bridge, Irilbung bridge, Thoubal bridge and Ithai Barrage.
In theory, the AI model would process various data to make predictions. In addition to analysing the water levels at the gauge sites, it would process parameters such as precipitation (rainfall), humidity and temperature.
"After analysing the data and all the parameters in the last 5 days, the AI would predict water levels of the rivers for the next 3 days," Sashikumar said.
Further, using the predictions, an algorithm for an early flood warning system is developed, Sashikumar added.
In theory, this AI model has an efficiency of 0.97 [0 being the lowest and 1 being the highest]--which is highly reliable, he said.
Developing a working application would take time. It would need a dedicated website. When the application comes alive, Manipur would have an early warning system for flood, Sashikumar informed.
Manipur has seen frequent floods in recent years which affected thousands of people. Floods also cause damage to properties and crops.
An early warning system would give the Government authorities and people ample time to prepare for floods and save properties worth lakhs. It would control floods and greatly help minimise the damage and risk, Sashikumar added.
Leishangthem Sashikumar hails from Kakching Wairi Sabal Leikai.
(This article was written under CAU Imphal's 5th Scientific Journalism Programme on  "Artificial Intelligence and it's uses")