Madhusudhan M V, A Y Harshitha, Avani M "Projection of the Extent of Inundation Corresponding to the Forecasts of Flood Levels in a River"

Paper_id: 36_4

Abstract

The analyzed project is Flood Prediction Using Satellite Imagery and Machine Learning, mainly the creation of the machine
learning back-end. Floods should be predicted, among other things, to preserve infrastructures, human lives, and the environment from flood disasters. This project applies machine algorithms to the satellite image data to make the correct flood prediction. The back-end system is utilizing K-Means, XGBoost, and Decision Trees machine learning models, which have been trained on environmental data derived from a mixture of satellite information. The parameters that were considered to enable better prediction of the future are those that were observed, e.g. rainfall intensity, sizes of water bodies, soil moisture, etc. These models were elaborately put to the test by trial and error, and a great number of experiments were made in order to find out the improved performance of the learning based on the accuracy rate. The reader can get an idea of the trend and patterns from the graphs over parameter results. The correctness of the models is also compared by the measures that have been set giving an idea of what models would be fitting the data. The project will come up with a solution that is both environmentally friendly and can be easily deployed to solve the problem of flood prediction so as to avoid the difficulties of setting up an end-to-end and scalable backend infrastructure.