Impact of climatic anomalies and reservoir induced seismicity on earthquake generation using Federated Learning

Abstract

Abstract In this article, earthquake forecasting model using federated learning (FL) technique has been proposed. Federated Learning is the most advanced technique of machine learning (ML) that guarantees data privacy, ensures data availability, promises data security, and handles network latency trials inherent in prediction of earthquakes by prohibiting data to be transferred over the network for model training. The main objective of this study is to determine the impact of artificial stresses and climatic anomalies on regional seismicity. Experimental verification has been carried out within 100 km radial area from 34.708o N, 72.5478o E in Western Himalayan region. Regional data of atmospheric temperature, air pressure, rainfall, water level of reservoir and seismicity has been collected on hourly bases from 1985 till 2022. In this research, four client stations at different points within the selected area have been established to train local models by calculating time lag correlation between multiple data parameters. These local models are transmitted to central server where global model is trained for generating earthquake alert with ten days lead time, towards a specific client where high correlation among all parameters have been reported.