### Papers from SWICo members

F. Siciliano G. Consolini, R. Tozzi, M. Gentili, F. Giannattasio, and P. De Michelis.

Geomagnetic indices are proxies of the geomagnetic disturbances observed on the ground during geomagnetic storms and substorms. So, their forecasting represents a key point to develop warning systems for the mitigation of possible effects of severe geomagnetic storms on critical ground infrastructures. Here, we forecast SYM‐H index using two artificial neural network models based on two conceptually different networks: the Long Short‐Term Memory (LSTM) and the Convolutional Neural Network (CNN). Both networks are trained with two different sets of data: 1) interplanetary magnetic field (IMF) components and magnitude, and 2) interplanetary magnetic field components and magnitude and previous SYM‐H values. Specifically, we selected 42 geomagnetic storms among the most intense occurred between 1998 and 2018.

The performance of the two models has been compared thus pointing out the peculiarity of each model. In summary we have found that: 1) both networks are able to well forecast SYM‐H index 1 hour in advance, with values of the coefficient of determination R^{2} larger than 95%; 2) when using the data set that includes SYM-H index the model based on LSTM is slightly more accurate than that based on CNN; 3) differently, when using the data set consisting of IMF values only the model based on CNN displays a higher accuracy than that based on LSTM.

**Publication**: F. Siciliano G. Consolini R. Tozzi M. Gentili F. Giannattasio P. De Michelis, Forecasting SYM‐H Index: A Comparison Between Long Short‐Term Memory and Convolutional Neural Networks, *Space Weather*, 19 (2), 2021. https://doi.org/10.1029/2020SW002589