Evaluating Empirical Lightning Parameterizations in Global Atmospheric Models