Passive remote sensing of snow with hydrological model and constraints on grain size and snow density

Predicting the snow parameters, such as snow depth or snow water equivalent is an important task in the geoscience study. In the past, parameter retrieval algorithms mainly use the relationship between a subset of the snow parameters and passive remote sensing measurements. However, the brightness temperatures depend not only on snow depth but also on other snow parameters. Thus, it is desirable to develop a multi-parametric inversion algorithm using multi-frequency and dual polarization measurements. The authors´ current approach is using a constrained neural network iterative inversion algorithm incorporating a priori estimates provided by the snow hydrology model to retrieve the snow parameters. For the real-time application, the weather forecast model precipitation input instead of station precipitation data is used to provide the precipitation input for the snow hydrology model. The results of the constrained inversion will be illustrated for the stations in the Northern Hemisphere