Remotely sensed monitoring of snow cover based on AMSR-E passive microwave brightness temperature

Snow cover is one of the most widely used parameters in snow monitoring. In this paper, AMSR-E brightness temperature, land cover production and in-situ meteorological observations in Xinjiang, China were used, and the decision-tree method was introduced to monitor the snow cover. In its first step, a new snow index, NDPMSI, was proposed to classify the study area into snow cover, non-snow cover and misclassified land cover. The misclassified area was classified into two types, bared area and non-bared area. And then the snow cover in the vegetation cover area was extracted from misclassified land cover with the aid of land cover production. After that, the MPDI23.8 was calculated to separate snow cover and non-snow cover in the bared area. A comparison of decision-tree with the traditional methods using K_hat coefficient showed that the classification accuracy of decision-tree has an obvious improvement.