Influence of Space Weather on Precipitation-Induced Floods – Applying of Solar Activity Time Series in the Prediction of Precipitation-Induced Floods by Using the Machine Learning

Abstract

This paper investigates hidden dependencies between the flow of particles coming from the Sun and 20 flood events in the United Kingdom (UK). The dataset analyzed in the study contains historical data covered on the daily level for the period October 2001 – December 2019. Solar activity parameters were used as model input, while rainfall data 10 days before and during each flood event were used as model output. To determine the degree of randomness for the time series of input and output parameters the correlation analysis has been performed. Machine Learning Classification Predictive Modelling is then applied to try to establish an eventual link between input and output data. Specifically, the decision tree, as the machine learning approach is used. In addition, it is analyzed the accuracy of classification models forecast. It is found that the most important factors for flood forecasting are proton density, differential proton flux in the range of 310-580 keV, and ion temperature. Research in this paper has shown that the classification model is accurate and adequate to predict the appearance of precipitation-induced floods.