Secondary ice production (SIP) commonly occurs in coastal Antarctic stratocumulus, affecting their ice number concentrations (N_ice) and radiative properties. However, SIP is poorly understood and crudely parametrized in models. By evaluating how well SIP is captured in a cloud‐resolving model, with a high resolution nest within a parent domain, we test how an improved comparison with aircraft observations affects the modelled cloud radiative properties. Under the assumption that primary ice is suitably represented by the model, we must enhance SIP by up to an order of magnitude to simulate observed N_ice. Over the nest, a surface warming trend accompanied the SIP increase; however, this trend was not captured by the parent domain over the same region. Our results suggest that the radiative properties of microphysical features resolved in high resolution nested domains may not be captured by coarser domains, with implications for large‐scale radiative balance studies over the Antarctic continent.