Hurricane genesis modeling based on the relationship between solar activity and hurricanes II

Abstract

This research presents improved results on modelling relationship between the flow of charged particles that are coming from the Sun and hurricanes. For establishing eventual link, the methods of Big Data, such as Adaptive Neuro Fuzzy Inference System (ANFIS), Parallel Calculations, Fractal analysis etc., are applied. The parameters of solar activity were used as model input data, while data on hurricane phenomenon were used as model output, and both of these on daily level for May–October in period 1999–2013. The nonlinear R/S analysis was conducted to determine the degree of randomness for time series of input and output parameters. The time lag of 0–10 days was taken into account in the research. It led to growing input parameters up to 99. The problem of finding hidden dependencies in large databases refers to the problems of Data Mining. The ANFIS with Sugeno function of zero order was selected as a method of output fuzzy system. The “brute-force attack” method was used to find the most significant factors from all data. To do this, more than 3 million ANFIS models were tested on Computer Cluster using Parallel Calculation. Within the experiments, eight input factors were calculated as a base for building the final ANFIS models. These models can predict up to 39% of the hurricanes. This means, if causal link exists, approximately every third penetration of charged particles from coronary hole(s) or/and from the energetic region(s) toward the Earth precede the hurricanes.