An evaluation of DMRT-ML for AMSR2 estimates of snow depth

Abstract

Modeling the physical state of a snowpack is widely recognized as a challenging aspect in the snowpack physical properties retrieval using passive microwave remote sensing. The Advanced Microwave Scanning Radiometer 2 (AMSR2) launched on JAXAs Global Change Observation Mission Water in 2012 with 10-15 years mission, continues observation record of Earth from space. The SWE product for AMSR2 is being developed as a satellite-based retrieval system that relies on static ancillary datasets to parameterize land surface properties that initialize retrievals. In this research, Dense Media Radiative Transfer Theory for Multi Layered (DMRT-ML) snowpack, a physically based numerical model, is employed [1]. The model is based on the Dense Media Radiative Transfer (DMRT) theory for snow scattering and extinction coefficients computation and uses Discrete Ordinate Method to numerically solve the radiative transfer equation. Using DMRT-ML assumptions, the application of the DMRT-ML model to the February 2013 snowstorm in southern Ontario to the Eastern seaboard of USA as well as Canada wide stations in December 2013 and January 2014 are explored. To supply DMRT input variables, Canadian Meteorological Center (CMC) daily snow depth, analysis snow depth product, and AMSR2 brightness temperature have been used. AMSR2 data has been utilized for surface physical temperature estimation. Using forward DMRT simulation for one layer snowpack, model sensitivity to snowpack grain size via AMSR2 observations is studied. This provides insight into the inversion look-up table matrix that is being developed using DMRT-ML for AMSR2 SWE retrievals.