Development of a Two-way Nested LETKF System for Cloud-resolving Model - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 490

(mm)
(m /s)
Fig. 20.8 Horizontal distributions of 1 h rainfall ( shaded regions ) and horizontal wind at the height
of 20 m ( vectors ) at 15 JST 5th September 2008 obtained by deploying 4 Inner LETKFs side by
side
produced by the spatial interpolation of the forecast of the Outer LETKF. Namely,
the boundary conditions of the Inner LETKF in CNTL did not have the information
of the Outer LETKF's assimilation data, even if the assimilation data existed
just outside of the Inner LETKF. To reflect the assimilation data of the Outer
LETKF into the Inner LETKF, the analyzed fields obtained by the no-cost smoother
( Kalnay et al. 2007 ; Yang et al. 2009 ), which is equivalent to the Ensemble Kalman
Smoother ( Evensen 2003 ), were used in producing the initial seeds of the Inner
LETKF (indicated by thick open arrows in the Outer LETKF in Fig. 20.4 a). The
boundary conditions of the Inner LETKF were also obtained from the forecast
from the analyzed fields of the no-cost smoother. Figure 20.9 shows the ensemble
mean distributions of rainfalls at 17 JST that were obtained with the initial seeds
and boundary conditions from the Outer LETKF's forecast (CNTL) and from the
analyzed fields of the no-cost smoother. When the analyzed fields of the no-cost
smoother were used, the rainfall regions, which were not generated in CNTL, were
generated at the northwestern part of the Inner LETKF's domain (indicated by an
arrow in Fig. 20.9 ). In the no-cost smoother experiment, the conventional data that
was used in Outer LETKF was assimilated again in the Inner LETKF. However, it
is considered that the conventional data can be used in the Inner LETKF, because
this data provides the small-scale information through the small scale localization
in the Inner LETKF. Since these rainfall regions that were generated in the no-cost