Quantifying the Performance of Geomagnetically Induced Current Models

Abstract

We describe a metric that has been repeatedly applied to assess the performance of models aimed at predicting geomagnetically induced currents from Space Weather events. The used parameterization, based on the well-known root-mean-square error between model and observations, is simple and intuitive. Its use is exemplified, and its advantages and disadvantages are discussed, as well as its relationship with the widely extended correlation coefficient, r. Although the use of r alone is inappropriate for purposes of evaluating the agreement between model and observations, its use is recommended to complement the described performance parameter.