Automatic ice thickness estimate
The key to estimating sea ice thickness is the conductive heat flux, between the water and the ice surface. The temperature of the water is known (freezing temperature) and the surface temperature may be estimated from thermal satellite data. Assuming a linear temperature profile, a model for the conductive heat flux, as inversely proportional to the ice thickness may be established.
We describe the heat balance on the ice surface by modelling the individual contributing heat fluxes, closely following the approach by Yu \& Rothrock (1996). This leads to an equation which may be solved to yield an estimate of the ice thickness.
We also model the heat flux from solar radiation, allowing the estimate of ice thickness from daytime as well as nighttime images.
The input data for the algorithm are air temperature and surface temperature. Re-analyzed two meter air temperature data is is obtained from the ERA project, while the surface temperature is estimated from the thermal bands of the MODIS sensor on board the Aqua satellite.
The algorithm is incorporated into a fully automatic framework, capable of processing large datasets.
Thick ice mask
Passive microwave data are used to narrow down areas of potential thin ice. New ice is distinguished from multi-year ice and thicker first year ice by its spectral profile. We mask out thick ice using a simple empirical relation, thresholding the ratio of the vertically polarized 19 GHz and 89 GHz channels.
For this purpose we use the AMSR-E passive microwave sensor which is also on board the Aqua satellite. The algorithm is integrated into the processing chain together with the ice thickness estimate.