A robust method for using climate models to forecast extreme precipitaiton

What is it about?

General circulation models (GCMs) are commonly used to model future climate and learn about how specific types of events may change in the future. However, the representation of rainfall processes in GCMs struggles from spatial and temporal biases. In this paper we propose a methodology for modeling rainfall from GCMs which seeks to (1) identify the large-scale climate variables that cause heavy rainfall; (2) check that the GCMs simulate these large-scale indices well; and (3) use a statistical model for heavy rainfall based on the large-scale climate variable. We find that this model improves representation of heavy rainfall in the Ohio River Basin for a single GCM. We also find that this approach predicts an increasing trend in future heavy rainfall, but that the trend is lower than the trend estimated using the rainfall field directly from the GCM.

Featured Image

Why is it important?

Heavy rainfall causes important economic damage, and predicting how it may change in the future is key for successful climate adaptation. However, making these predictions is complicated by the limitations of existing models. Our approach is an alternative to statistical down-scaling and bias correction techniques that are often applied and is directly informed by physical principles.