Twenty-first century snowfall projections within the eastern Great Lakes region: detecting the presence of a lake-induced snowfall signal in GCMs

In this study, an analysis of model-derived snowfall from 2006 to 2100 is conducted, investigating the presence of a lake-induced (LI) snowfall signal to the lee of Lakes Erie and Ontario. Output from seven fully coupled global climate models is used from phase 5 of the Coupled Model Intercomparison Project. Snowfall trends over the entire study area, and in the defined LI snowbelt, are examined seasonally for two representative concentration pathway (RCP) scenarios. The term lake-induced is used to encompass both lake effect processes, along with lake-enhanced snow. In addition to snowfall, physical mechanisms associated with these trends, such as surface temperatures and snow to total precipitation ratios, are examined. The models perform well when compared to snowfall observations despite model difficulties in resolving small-scale orographic and lake effect processes. Modelled snowfall declines by 20-45% over the entire region by 2100, with snowfall increasing in the LI belt during the first quarter century before declining rapidly in RCP 8.5. An additional statistically significant 1.8-1.97 cm signal of LI snow water equivalent is detected within the ensemble mean of the defined snowbelt to the lee of Lakes Erie and Ontario that is not seen in the grid cells outside of the lake belt. This LI signal is also projected to decline more quickly than the non-lake-induced signal. Surface 2-m temperature is found to have a significant impact on snowfall changes in the region, with impacts not only on total precipitation but also on the percentage of precipitation falling as snow.