H21H-1494
SMAP/SMOS Soil moisture brightness temperature virtual observations to study data assimilation scheme

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Pablo Saavedra, University of Bonn, Bonn, Germany
Abstract:
A multidisciplinary research unit has been established in order to develop a data assimilation framework for coupled subsurface-land surface-atmosphere systems (SLASs), i.e. the coupled suit of models TerrSysMP comprised by ParFlow (subsurface), Community Land Model (CLM, surface), and COSMO (atmosphere). It aims to test how different kinds of observations may improve system state estimations with a focus on inter-compartmental fluxes of matter and heat energy. To that goal a simulated virtual reality (VR) catchment is being generated as a tool to test data assimilation schemes for SLAS. The virtual reality overcomes the problem of data scarcity for the different components as subsurface, soil and atmosphere in the real world and provides the full system state as a basis for the evaluation of the effectiveness of data assimilation strategies.

The first version of the VR uses TerrSysMP - reduced to COSMO and CLM - to generate virtual observations such as satellite measurements, radar observations and meteorological station data. Currently VR simulations are available for a region encompassing the Neckar catchment located in south-west Germany with 1.1km horizontal resolution for the time period from 2007 to 2013.

This contribution focuses on the evaluation of satellite observations of the microwave emission at L-band as observed by the current satellite missions SMAP and SMOS.

By adjusting the Community Microwave Emission Model (CMEM) as a forward operator for the VR framework, a first set of virtual passive microwave observations is generated. SMAP and SMOS observations are simulated taking into account orbit characteristics, revisit times, and angular viewing geometries. The virtual observations will be statistically compared with available real observations to evaluate the degree of reality in terms of mean values and dynamic ranges. These comparisons will hint to systematic differences between TerrSysMP and reality, which need to be addressed by appropriate bias correction schemes or by adjusting parameters in the virtual reality itself. The presentation will discuss results from the statistical comparison oriented towards identifying potential deficiencies in the forward operator over specific areas i.e. frozen soil and dense vegetation to support the further development of the VR and the CMEM model.