Validation of ENVISAT ASAR based lake ice maps on Lake Päijänne

In this paper, we propose a general classification method for determining ice cover on lakes using ENVISAT ASAR data. The method assumes a Gaussian mixture model for feature vectors extracted from a local neighborhood around each pixel. We focus on Lake Paijanne in this work, and the classification results are validated against in situ observations of freeze up dates and a snow cover product from the optical MODIS sensor. The method is fully automatic, but it requires a manual post-processing step to remove poor results. The classification of the selected scenes compared fairly well to both sources of validation data. However, the manual selection step is time-consuming and subjective.