Exploring the intrinsic predictability of an Arctic polar low

Abstract Accurately forecasting polar lows is a major challenge, due to their small scale and short lifetime, but the limits on their predictability are not well understood. In this study we investigate the intrinsic predictability (i.e., forecast limits arising from a perfect model with nearly perfect initial conditions) of an Arctic polar low from March 2024. The growth of small‐amplitude initial‐condition perturbations is driven by moist processes, with faster error‐growth rates at smaller scales and in convectively unstable regions in the vicinity of the polar low. Differences in latent heating produce differences in the divergent‐wind component that spread outward to the Rossby radius of deformation in a manner resembling gravity waves. These divergent‐wind differences undergo geostrophic adjustment, eventually producing differences at larger scales in the rotational wind component. This general picture of upscale error growth is consistent with previous studies on midlatitude‐cyclone predictability, but, unlike in those cases, there is no evidence of a distinct stage of error growth on the polar‐low scale. As a result, the differences are primarily localized to the displacement of individual convective cells, and the intensity and location of the polar low remain mostly unchanged by the upscale error growth over the course of its 36‐hr life cycle. These results suggest that, at lead times of the order of a polar‐low life cycle, the convective‐scale predictability can be intrinsically limited by the growth of unobservably small‐amplitude initial‐condition errors, but forecasts of the polar low at large can be improved via initial‐condition error reduction at larger amplitudes and scales.