Assimilation of MODIS and AMSR-E Snow Parameter Observations into a Physical Snow Model

The assimilation of binary (yes/no) type observations of non-state variables, as well as lower resolution observations of state variables, into any data assimilation scheme poses a significant problem to the scheme. However, these data may still be of use to the assimilation scheme. In the problem of modeling the evolution of snow there are high spatial resolution snow cover observations from the Moderate Resolution Imaging Spectroradiometer (MODIS), which is not a state variable of the numerical model but is a binary function of the state variable, Snow Water Equivalent (SWE). An addition problem with these data is that this binary function is non-unique, as a positive snow cover signal will correspond to any value of SWE potentially. Direct observations of SWE are available from the Advanced Microwave Scanning Radiometer - for the Earth Observing System (AMSR-E) but are at a coarse spatial and temporal resolution, introducing another non-unique function between SWE and the a set of observations. In this paper we present some data assimilation schemes to assimilate MODIS snow cover and AMSR-E SWE observations; firstly on their own into a high resolution physical snow model and then through different combinations of the two. The results from these experiments help demonstrate that we can assimilate binary type observations if we can constrain the non-uniquessness and that they have a positive impact on the modeled snow evolution and to have better agreement with future MODIS snow cover observations.