The lagged relationship between Arctic sea ice extent (SIE) and United States snowfall is explored in this study. Monthly snowfall amounts are computed from 440 stations in the contiguous United States from 1979 to 2009 and compared with total and regionalized Arctic SIE over the same time period. Principal Components Analysis along with multiple linear regression is used to determine the sea ice regions and lags that explain the most variation in snowfall amount. Although previous work has identified total summer sea ice as a predictor for the following winter's US snowfall, this study shows that the Kara Sea SIE is more influential for fall snowfall, and total SIE explains snowfall variability more in the spring. In the fall, longer lags indicate that the previous spring SIE is a good predictor of snowfall. In the late winter and spring there are generally shorter lags, where early winter SIE explains the most variability in snowfall. When just the first two principal components are considered, Kara Sea SIE and atmospheric teleconnections become the most consistent predictors of subsequent US snowfall. Results highlight the skill of regionalized Arctic SIE in explaining snowfall variability and how attention to Arctic regions could potentially improve seasonal snowfall forecasts.
- multiple linear regression
- principal components analysis
- sea ice