Skip to main content
Climate & Radiation
Geodesy and Geophysics
Wallops Field Support
Code 614.3 Brown Bag Seminar: Ally M. Toure
Centre for Research and Applications in Remote Sensing (CARTEL), University of Sherbrooke, Quebec, Canada
Thursday, May 14, 2009 - 08:00
Seasonal snow cover has a strong impact on climate, hydrological processes and on human activities (Kukla, 1981). Snow is the frozen storage term in water balance. And snow is also a valuable resource at a variety of spatial scales. Monitoring of the snow's physical parameters (extent, water equivalent and melting condition) is essential for weather and hydrological forecasts in many regions, as it provides basic data for meteorological forecasts, for hydroelectric power generation, fresh water supply, river traffic, irrigation and flood control (Mätzler, 1987). So far, the passive microwave-based retrievals of terrestrial snow parameters such as snow water equivalent (SWE) from satellite observations have been generated primarily by regression-based empirical “inversion” methods based on snapshots in time. Snow parameters retrieved by such methods are sometimes used in product-based data assimilation (Pulliainen et al., 2006; Dong et al., 2007). With this type of approach, the retrieval and the forward data assimilation do not necessarily have consistent physics or assumptions. Consistency is much easier to achieve when radiances are directly assimilated. This radiance assimilation (RA) approach has been used for years to retrieve atmospheric parameters by the operational weather forecasting community with great success. This approach can also be a interesting way to retrieve snow physical parameters (Durand and Margulis, 2006). The performance of the ensemble Kalman Filter (EnKF) for SWE estimation is assessed by assimilating microwave observations at Ground Based Passive Microwave Radiometer (GBMR-7) frequencies into a snow physics model, CROCUS (Brun et al. 1989, 1992). CROCUS has a realistic stratigraphic and ice layer modeling scheme. This work builds on previous methods that used a snow physics model with a limited number of layers. RA assimilation methods require accurate predictions of the brightness temperature emitted by the snowpack. It has been shown that the accuracy of radiative transfer models is sensitive to the stratigraphic representation of the snowpack. However, as the stratigraphic fidelity increases, the number of layers increases, as does the number of state variables estimated in the assimilation. One goal of the study was to investigate whether radiance observations could be used in a Tb assimilation scheme to characterize a more realistic stratigraphy, in a snow model with an unlimited number of layers. We predicted snow radiance by coupling CROCUS to MEMLS. Despite the errors in the prediction of the observations, the EnKF was able to retrieve SWE with a bias of 1.4 mm and an rmse of 14.92 mm, which represents 78.7% improvement over the nominal run of CROCUS.