Lake Ice Cover product
The lake ice cover algorithm is based on spectral mixing model, where the reflectance spectrum in each pixel is assumed to be a linear combination of a some endmembers (snow, open water, etc.). The contribution from each endmember spectrum is further assumed to represent the abundance of the relevant endmember within the pixel. The individual endmember spectra are obtained by training the algorithm on a population of pure endmembers. The abundances of each endmember within a given pixel may then be retrieved by linear decomposition of the spectrum. This is then used to estimate the fractional ice coverage within each pixel.
The validation of the lake ice cover product is performed using observed ice-off dates for Finnish lakes provided by SYKE.
Lake Surface Temperature product
For estimating the lake surface temperature we use the algorithm developed by Key et al. (1997) which utilizes split-window (two frequencies) and is extended to take advantage of the dual-view properties of AATSR. The algorithm makes use of the 11 µm and 12 µm channels of AATSR.
The lake surface temperature product is validated using temperature measurements of Finnish lakes, provided by SYKE.