Anticipating climate impacts and risks in present or future climates requires predicting the statistics of high-impact weather events at fine-scales. Direct numerical simulations of fine-scale weather are computationally too expensive for many applications. While deterministic-based (deep-learning or statistical) downscaling of low-resolution climate simulations are several orders of magnitude faster than direct numerical simulations, it suffers from several limitations. These limitations include the tendency to regress to the mean, which produces excessively smooth predictions and underestimates the magnitude of extreme events. They also fail to preserve statistical measures that are key for climate research. We use a conditional GAN (cGAN) architecture to downscale daily precipitation as a Regional Climate Model (RCM) emulator. The cGAN generates plausible residuals on top of the predictable expectation state produced by a deterministic deep learning algorithm. The skill of cGANs is highly sensitive to a hyperparameter known as the weight of the adversarial loss (\(\lambda_{adv}\)), where the value of  \(\lambda_{adv}\) required for accurate results varies with season and performance metric, casting doubt on the reliability of cGANs as usually implemented. However, by applying a simple intensity constraint to the loss function, it is possible to obtain reliable performance results across \(\lambda_{adv}\) spanning two orders of magnitude. CGANs are considerably more skillful in capturing climatological statistics, including the distribution and spatial characteristics of extreme events. With this modification, we expect cGANs to be readily transferable to other applications and time periods, making them a useful weather generator for representing extreme event statistics in present and future climates.

Atmospheric rivers (ARs) play a critical role in moisture transport across the Southern Hemisphere mid-latitudes and often produce extreme rainfall events across New Zealand. Here we examine ARs in a new set of high-resolution (12km) dynamically downscaled simulations from select CMIP6 models. We begin with a historical evaluation of AR properties from this model ensemble followed by future projections. We demonstrate that by end-of-century, the maximum integrated vapour transport associated with ARs will robustly increase by as much as 20% in some regions of the South Pacific, and ARs contribute a larger proportion (up to 20% more) of the annual precipitation climatology in certain regions of New Zealand. The spatial structure and seasonality of these changes indicate the role of a poleward shift and intensification of the westerly jet. This is further quantified through a decomposition into dynamic and thermodynamic components. While the thermodynamic change dominates the increase in AR frequency, the additional positive dynamical change is notable over the mid-latitudes and southern New Zealand. Separating events using an AR categorisation scale shows that higher category (longer duration, more intense) events more than double in frequency, underscoring the increasing role that ARs will play in extreme weather events in the future in this region.