Extratropical cyclones drive midlatitude weather, including extreme events, and determine midlatitude climate. Their dynamics and predictability are strongly shaped by cloud diabatic processes. While the cloud impact due to latent heating is much studied, little is known about the impact of cloud radiative heating (CRH) on the dynamics and predictability of extratropical cyclones. Here, we address this question by means of baroclinic life cycle simulations performed at a convection-permitting resolution of 2.5 km with the ICON model. The simulations use a newly implemented channel setup with periodic boundary conditions in the zonal direction. Moreover, the simulations apply a new modeling technique for which only CRH interacts with the cyclone, which circumvents changes in the mean state due to clear-sky radiative cooling that has complicated the interpretation of previous work. We find that CRH increases the kinetic energy of the cyclone system. The impact is most prominent at upper levels. To understand the CRH impact on the upper-tropospheric circulation, we diagnose the evolution of differences in potential vorticity between a simulation with and without CRH, and we quantify through which processes these differences grow over the course of the cyclone's life cycle. According to this diagnostic, CRH affects the cyclone mostly via the intensification of latent heating from cloud microphysical processes. Near the tropopause, direct diabatic modification of potential vorticity by intensified latent heat release precedes further changes in the tropopause by the upper-tropospheric divergent flow, which represents an indirect impact of latent heat release. Subsequently, differences in the tropopause structure amplify with the rotational flow during the highly nonlinear stage of the baroclinic wave. Our results show that although CRH is comparably small in magnitude, it can affect extratropical cyclones by changing cloud microphysical heating and subsequently the large-scale flow. The CRH impact follows a previously identified mechanism of multi-stage upscale error growth. At the same time, simulations in which CRH is disabled after certain days show that the CRH impact operates throughout the entire intensification phase of the cyclone. This means that CRH does not merely provide an arbitrary initial perturbation to the cyclone, from which differences grow in a generic way. Instead, our results suggest that uncertainties associated with the representation of CRH in numerical models have a more systematic impact and may more fundamentally influence model predictions of extratropical cyclones.
