Predicting Space Weather Impacts on the North American Power Grid Using Perturbed-Input Ensemble Modeling

Errors in space weather predictions typically arise through inaccurate specification of initial conditions or boundary conditions, as well as model uncertainties that arise because the model only approximates the system it simulates. For physical modeling of the solar wind-driven magnetosphere, the specification of the solar wind driver applied at the upstream boundary of the simulation domain is particularly important. As part of a large, interdisciplinary project studying the impacts of extreme space weather events on power grid infrastructure we recently developed a method to assess the effect of uncertainties in solar wind driving on the predictions from a modeling framework. We used the Space Weather Modeling Framework, coupling a global magnetohydrodynamic code with an ionospheric electrodynamics solver and an inner magnetosphere code, to simulate the geospace response to solar wind driving. Applying our perturbed input ensemble method also allows us to examine the uncertainty in predicted field quantities such as magnetic perturbations, as well as combining the ensemble members to provide better predictions. To examine the impacts on the North American power transmission system we combine spatiotemporal predictions of magnetic perturbations with a ground conductivity model to estimate geoelectric field. We then use the multiple realizations of geoelectric field to drive a model of the North American power transmission system, developing an ensemble prediction of geomagnetically induced currents and transformer heating. Using multiple realizations, we examine the variability in system response to different realizations of the same simulated event to begin to quantify the uncertainty.