Challenges of tracking extratropical cyclones in regional climate models

Extratropical cyclones (ETCs) play a crucial role in determining the climate in a large part of the world. For this reason, climate models’ ability to reproduce ETCs’ characteristics (such as track location and development and decay regions) is of great importance. ETC tracking algorithms are sophisticated diagnostic tools that have been used extensively to evaluate models’ ability to simulate ETCs. In general, these algorithms have been developed and applied to coarse resolution global datasets such as global climate models and reanalyses. As regional climate models (RCMs) mature, the potential for tracking ETCs at considerably higher resolutions has naturally become attractive. Application of ETC tracking algorithms to RCM-produced datasets, however, introduces additional challenges. In order to investigate these challenges, we compare ETC tracking results of a simulation produced by the Canadian Regional Climate Model (CRCM) over North America at a 45-km resolution driven by ECMWF ERA-Interim, with the results obtained directly from the driving reanalyses. The ERA-Interim data is treated in three different ways applying spatial smoothing, the tracking algorithm, and the extraction of the regional grid in a different order, so that individual sources of error can be identified. It is shown that the mere existence of boundaries in the regional domain of the CRCM affects tracking results not only near the boundaries but also well within the domain. To a lesser extent, the use of different spatial smoothing techniques also affects the number of tracked ETCs. However, even after accounting for these artifacts, the CRCM produces a lower ETC count than the driving dataset. This underestimation of the tracking statistics might be related to known CRCM biases, or to some problems with the surface representation at 45-km resolution. Nevertheless, it is concluded that RCMs are a suitable choice for ETC trajectory studies, but only when handled carefully and used over large domains targeting the cyclone track of interest.