FLOOD PREDECTION USING MACHINE LEARNING

MACHINE LEARNING

One of the undesirable natural catastrophes are floods. Significant property and life losses can result from flooding.Additionally, it could harm crops and trees that have been planted in agricultural areas. Flooding may be brought on by prolonged rainfall, inadequate precipitation drainage, surface runoff from melting snow, or dam failure. Today, the natural water storage places, such as rivers and lakes, have been destroyed and turned into construction sites. Flash floods can happen quickly within a few hours, as opposed to a typical flood. Research on flood prediction has evolved in order to reduce the amount of fatalities, property losses, and other concerns caused by flooding.Machine learning techniques are commonly used to build efficient prediction models for weather forecasting. Due to advancements in forecasting technology, more advanced and efficient solutions are now available. Utilizing the precipitation data from this research, a prediction model has been developed to anticipate the occurrence of floods triggered by rainfall. The model generates a forecast on the likelihood of a "flood occurrence" based on the expected range of precipitation for specific locations. Data on rainfall from various districts in India has been employed to construct the forecast model. The dataset has been utilized to train different algorithms, including Multilayer Perceptron, Support Vector Machine, and Linear Regression. Among these algorithms, the Multilayer Perceptron (MLP) classifier demonstrated satisfactory performance, achieving the highest accuracy rate of 97.40%. Consequently, a climate scientist can reliably predict the occurrence of a flood during heavy rainfall events using the Machine Learning Predection flash flood prediction model.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.