Climate change has brought about rapid change in the Arctic including reduced sea ice cover. This has, in turn, increased the need for accurate forecasts of where sea ice is located. Of particular concern is determining the location of the contour marking the region with greater than 15% ice coverage. Current sea ice forecasts are developed from deterministic prediction systems, such as numerical climate models. While these predictions are informative, they do exhibit substantial biases. However, these biases tend to follow particular spatial and temporal patterns, so it is feasible to model and correct them. In this talk, I will introduce a statistical post-processing method for contours. This technique, called contour-shifting, corrects the predictions outputted from deterministic forecasting systems, using information about spatial location, long-term trends, and seasonality. On a test data set using observations and predictions from 2001-2013, contour-shifting reduced the area predicted incorrectly by an average of 21.3%.