Using Data Assimilation Methods for Physics-Based Capabilities to Predict Solar Activity Cycles

Difficulties of building reliable forecasts of the strength and duration of solar activity cycles are associated with numerous problems from both observations and dynamo models. Utilization of the mathematical data assimilation approach, in which a theoretical model is ‘trained’ by observational data, allows us to improve the model solution according to available observations in an optimal way by taking into uncertainties in both observations and model. The data assimilation approach covers a large number of different methods, as well as their parameters that may affect predictive capabilities. In this presentation I will compare application of four data assimilation methodologies: Ensemble Kalman Filter method, Extended Kalman Filter, Ensemble Kalman Filter Smoother and Ensemble Adjustment Kalman Filter for predicting the sunspot cycles using a low-order solar dynamo that takes into account effects of the magnetic helicity balance, and discuss the prediction results for the next solar cycle.