Weather Monitoring and Cloudburst Prediction Based on Machine Learning Algorithms: An Initiative Towards Disaster Management

With the advent of machine learning techniques, it has now become possible to anticipate natural disasters and take precautionary measures to mitigate the impact of disasters. In this paper, cloudburst has been targeted as one of the disasters and with the proposed system comprising of a web interface predicts the occurrence of cloudburst in the susceptible regions. Proposed system works by integrating real-time data obtained from weatherstack API with machine learning and deep learning models and predictions are based by comparing training the model based on the historical meterological data obtained from authentic sources. Likelihood of the occurrence of cloudburst is based on the meterological variables such as wind direction, environment humidity, temperature, wind direction, air pressure and precipitation. Number of machine learning and deep learning models have been used and their performance is evaluated based using metrics precision, recall, test accuracy, cross-validation accuracy and F1-score. Deep learning models such as the Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) have been found to perform well as compared to other models, since these models results in high accuracy for the cloudburst prediction and hence taking proactive measures for disaster management. This is an indeed an important initiative towards the providing safety to the lives and resilience of the population residing in the communities which suffer because of extreme weather conditions.