IN41B-1703
Predicting Ground Based Magnetometer Measurements Using the Ensemble Transform Kalman Filter

Thursday, 17 December 2015
Poster Hall (Moscone South)
Erin Michelle Lynch, University of Maryland College Park, College Park, MD, United States
Abstract:
Ensemble data assimilation techniques, including the Ensemble Transform Kalman Filter (ETKF), have been successfully used to improve prediction skill in cases where a numerical model for forecasting has been developed. These techniques for systems for which no model exists are developed using the reconstruction of phase space from time series data. For many natural systems, the complete set of equations governing their evolution are not known and observational data of only a limited number of physical variables are available However, for a dissipative system in which the variables are coupled nonlinearly, the dimensionality of the phase space is greatly reduced, and it is possible to reconstruct the details of the phase space from a single scalar time series of observations. A combination of the phase phase reconstruction with ETKF yields a new technique of forecasting using only time series data. This technique is used to forecast magnetic field variations in th magnetosphere, which exhibits low dimensional behavior on the substorm time scale. The time series data of the magnetic field variations monitored by the network of groundbased magnetometers in the auroral region are used for forecasting at two stages.. In the first stage, the auroral electrojet indices computed from the data from the magnetometers are used for forecasting and yields forecasts that are better than persistence. In the second stage, the multivariate time series from several auroral region magnetometers is used to reconstruct the phase space of the magnetosphere-solar wind system using Multi-channel Singular Spectrum Analysis. The ETKF is applied to ensemble forecasts made using model data constructed from long time series of the data from each magnetometer and observations of the magnetometer measurements. The improved prediction skill, e.g., with respect to persistence, is achieved from the use of the dynamical behavior of nearby trajectories. The near-real time forecasts of space weather based on this approach will use the correlated data of the solar wind at L1 and data of the magnetosphere from the networks of magnetometer stations.