An improved understanding of the temporal variability and the spatial distribution of snowmelt on Antarctic sea ice is crucial to better quantify atmosphere-ice-ocean interactions, in particular sea-ice mass and energy budgets. It is therefore important to understand the mechanisms that drive snowmelt, both at different times of the year and in different regions around Antarctica. In this study, we combine diurnal brightness temperature differences (dTB(37GHz)) and ratios (TB(19GHz)/TB(37GHz)) to detect and classify snowmelt processes. We distinguish temporary snowmelt from continuous snowmelt to characterize dominant melt patterns for different Antarctic sea ice regions from 1988/89 to 2014/15. Our results indicate four characteristic melt types. On average, 38.9±6.0% of all detected melt events are diurnal freeze-thaw cycles in the surface snow layer, characteristic of temporary melt (Type A). Less than 2% reveal immediate continuous snowmelt throughout the snowpack, i.e. strong melt over a period of several days (Type B). In 11.7±4.0%, Type A and B take place consecutively (Type C), and for 47.8±6.8% no surface melt is observed at all (Type D). Continuous snowmelt is primarily observed in the outflow of the Weddell Gyre and in the northern Ross Sea, usually 17 days after the onset of temporary melt. Comparisons with Snow Buoy data suggest that also the onset of continuous snowmelt does not translate into changes in snow depth for a longer period but might rather affect the internal stratigraphy and density structure of the snowpack. Considering the entire data set, the timing of snowmelt processes does not show significant temporal trends.