Comparison among physical process based snow models in estimating SWE and upwelling microwave emission

Snowpack serves as a critical water resource and an important climate indicator. Accurately estimating snow water equivalent (SWE) and melt timing has both civil and scientific merits. Physical process based multi-layer land surface models (LSM) characterize snowpack by tracking the energy balance and mass balance in each layer. However, in terms of the number of layers used to model the snowpack stratigraphy, as well as the complexity of the simulated mass/energy exchanges in each single layer, significant variances exist among different LSMs. Previous work has largely focused on assessing the impact of layering and stratigraphy representation on mass and energy balance, with little attention paid to the implications of these factors on predicted microwave brightness temperature (Tb). In this paper, three LSMs with varying snow layer schemes: SSiB (3-layer), CoLM (5-layer), and SNOWPACK (N-layer), are coupled to the Microwave Emission from Multi-Layer Snowpacks (MEMLS) radiative transfer model (RTM) to simulate the snowpack mass/energy budgets and microwave signature over a full season. The simulations are performed at five in-situ gage locations in the Kern River Basin, Sierra Nevada, CA where it is known that large snow events occur that can be problematic to represent using a small number of snow layers. A particular emphasis is placed on assessment of the impact of layering scheme on the results. Preliminary results show that even for SSiB which has a relative simple empirical layering scheme, the modeled annual SWE could be highly correlated with the in-situ SWE (r¬2=0.91) if the precipitation bias is corrected, also, the comparison between the Tb simulated by SSiB+MEMLS and the downscaled AMSR-E Tb measurements shows a correlation coefficient of 0.94 during the snow accumulation season (Oct to Apr) if the grain growth parameters and the soil snow reflectivity is properly calibrated. Future work includes comparing SWE and Tb from all threemodels and quantitatively determining how the more complex models (SNOWPACK) could possibly further improve the Tb estimates, and how they will increase the computational loads, which are highly relevant to the ultimate goal of estimation of SWE via assimilation of multi-frequency passive microwave observations.