Through methodology unique for tropical cyclones in peer-reviewed literature and through examination of two tropical cyclone events, this study uses ensemble forecasts, sensitivity analysis, and ensemble data assimilation to explore the dynamics and predictability of tropical cyclone formation with state-of-the-art cloud-resolving mesoscale models. In both of these cases, high convective available potential energy and mid-level moisture were found to benefit genesis. The strong sensitivity to initial condition differences in both cases exemplifies the inherent uncertainties in hurricane intensity prediction where moist convection is the key that limits predictability. The success of using an ensemble Kalman filter in assimilating Doppler radar radial velocity observations of a hurricane demonstrates that, even in cases of extreme uncertainty in a tropical cyclone, advanced data assimilation and subsequent event-dependent probabilistic forecasts can offer significant benefits.