Godelieve Deblonde
A one-dimensional variational (1D-Var) scheme to assimilate passive
microwave brightness temperatures (e.g. from the SSM/I, TMI and AMSR-E
radiometers) in rainy atmospheres and satellite derived rain rates
(PATER algorithm) has been developed at the Meteorological Service of
Canada (MSC). The 1D-Var scheme is very similar to that developed at
ECMWF (Moreau et al., 2004). The observation operator includes a
radiative transfer model and moist physical parametrization schemes.
Moist processes are described by the Kain-Fritsch scheme for deep
convective clouds and by the Tompkins-Janiskova scheme for stratiform
clouds. The Jacobians of the physical processes are computed via finite
differences. Atmospheric profiles are taken from a research version of
the GEM global forecast model named "mesoglobal" whose configuration
includes a horizontal resolution of 45 km and 58 vertical levels with a
model top of 10 hPa. Both 1D-Var techniques allow the retrieval of
temperature and humidity profiles in rainy areas. A preliminary
evaluation of the two methods has been done for several tropical
cyclones. Both methods are successful in improving significantly the
model surface rain rate. However, temperature increments (~ 0.4 K) are
larger than those obtained at ECMWF. As a consequence, assimilating
1D-Var analyzed Integrated Water Vapour in the MSC 3D-Var or 4D-Var
systems does not seem to be an interesting option.