Calculation of Classical Polar-Activity Indices Using Their Modern Variants

At present, the auroral-electrojet AE, AU, and AL indices are calculated using a smaller number of stations than earlier. This influences the accuracy of estimates of the magnetospheric activity components, which are performed using the relationships with modern indices AE₈ (obtained from the data of 8 stations) substituted instead of the classical indices AE₁₂ (obtained from the data of 12 stations). In this paper, we develop an artificial neural-network (ANN) algoritm for recalculating the classical polar-activity indices from modern ones. It is demonstrated that the developed ANN is capable of retrieving the AE₁₂-index from data of 8 stations (AE₈) with efficiency reaching 90% in some cases. The results obtained allow one to estimate the error in the representation of the auroral-electrojet activity by modern AE₈ indices as compared with the ones introduced into geophysical research by their creators. We also develop a retrieval technique for polar indices, which employs data on solar-wind parameters.