Identifying the drivers of the observed springtime cooling trend in northern North America with large ensemble simulations

Over the last four decades, observations reveal widespread warming in the mean surface temperature for most land areas of the Northern Hemisphere. However, one exception from this trend is found extending across the northern United States and southern Canada during late boreal winter and early spring. In fact, reanalysis and station-based observational datasets reveal surface cooling within this same area, which is largest in April. It remains unclear whether this regional temperature trend is due to external forcing, internal variability, or a combination of factors.

To revisit this question, we leverage large initial-condition ensembles of coupled climate models and a collection of factual and counterfactual AMIP-style experiments with variations in prescribed external forcing, sea surface temperatures (SST), and sea ice boundary conditions. We find that this regional temperature trend is near or outside the ensemble spread of historical trends within the fully-coupled large ensembles. While the ensemble mean of the factual AMIP experiments with observed SSTs also do not reveal this regional cooling pattern, we find several individual ensemble members which can reproduce a similar spatial pattern of temperature trends. This suggests that this warming hole may be due to an unlikely realization of internal variability. By comparing the different large ensemble simulations, we further diagnose the dynamical mechanisms, atmospheric teleconnections, and land surface radiative feedbacks from anomalous snow cover, that may correspond to the important drivers of this distinct temperature trend pattern. The results of this case study further emphasize the importance of leveraging multiple climate model large ensembles for capturing the entire range of possible regional temperature trends due to internal variability within the observational record.