Temporal and spatial variability of the beginning and end of daily spring freeze/thaw cycles derived from scatterometer data

The variability of snowmelt dates is important for the terrestrial carbon balance in boreal and subarctic environments. Scatterometers such as the K u-Band QuikScat have been proven applicable for the detection of surface thaw. We present an improved method for the capture of thaw events based on significance of diurnal differences with respect to long term noise. For each 10 km × 10 km grid point two products are derived for the major thaw period: 1) the onset of thaw and 2) the end of daily freeze/thaw cycles at the surface. Both dates may be related to biogeochemical processes, especially carbon fluxes in boreal forests. The onset of the spring thaw period coincides with the first days of increased CO 2 fluxes above the late winter baseline. The end of daily freeze/thaw cycles corresponds to the switch from source to sink in evergreen boreal forest environments as illustrated by comparison with eddy-flux tower data and xylem sap flow records from other investigators. The approach is suitable for detecting freeze/thaw cycle periods in boreal forest and tundra biomes. The mean absolute difference in end of freeze/thaw cycling date within the central Siberian study area (3 Mio km 2 comprising tundra, boreal forest and steppe grassland) was 9 days for 2000 to 2004. Largest mean differences occurred in the southern taiga and all tundra regions, which were highest for spring 2000. The improved extraction method delivers more precise products from the viewpoint of carbon accounting in evergreen boreal forest environments. This widens the application potential of scatterometer data beyond the current status.