P2.6
Improving the assimilation of land-surface sensitive satellite channels in NCEP's Global Data Assimilation System (GDAS) by improving land-surface emission in the Community Radiative Transfer Model (CRTM)
Jesse Meng, University of Maryland Baltimore County, NOAA/NWS/NCEP, and NASA/GSFC, Camp Springs, MD; and K. Mitchell, H. Wei, J. C. Derber, G. Gayno, and Q. Liu
The operational NCEP Global Data Assimilation System (GDAS) assimilates satellite brightness temperatures (Tb), as well traditional satellite retrieval products, in both cases applying a 3D-Variational assimilation method that utilizes the adjoint model of the assimilating background model (the NCEP's Global Forecast System). To assimilate Tb, a forward radiative transfer model is applied to transform model simulated states of atmospheric, ocean and land into model simulated top-of-atmosphere (TOA) satellite Tb. The radiative transfer model used is the Community Radiative Transfer Model (CRTM) of the NOAA-NASA-DOD Joint Center for Satellite Data Assimilation (JCSDA). For satellite channels that are sensitive to the land surface, such AMSU A/B microwave channels 1, 2, 3, and 15, and TOVS/HIRS infrared channel 8, the accuracy of the model simulated surface emission is critical to the effective assimilation of such channels. Presently, GDAS assimilates very few Tb observations from land-surface sensitive channels. Two reasons for this are rather large errors in model simulated land surface skin temperature and model's treatment of land surface emissivity for various land surface types and hence rather large differences in model simulated and observed TOA Tb in such channels. These large differences cause the analysis to reject the satellite Tb observations. This presentation will present recent research and testing aimed at reducing the error of the land-surface simulated skin temperature in the warm season in the land-component of GFS. In particular, we will show the sensitivity of the model simulated LST to the treatment of A) the surface turbulent exchange coefficient (i.e. surface layer treatment at bottom of PBL), B) source of initial land states, and C) treatment of surface emissivity. Secondly, we will demonstrate how the improved model simulations of LST and improvement of specification of surface emissivity in the land surface emission module of CRTM result in substantial increase in GDAS of the number of assimilated land surface sensitive infrared Tb observations, e.g., HIRS channel 8 and eventually AIRS surface channels. Future follow-on research will focus on further improvement to the background LST simulations by improving the land states like soil moisture and snowpack by assimilating satellite retrievals of soil moisture and snowpack via land data assimilation methods such as ensemble Kalman filtering being developed by PIs funded by the JCSDA.
Poster Session 2, Observations
Wednesday, 17 January 2007, 2:30 PM-4:00 PM, 212B
Previous paper Next paper