Development of a Two-way Nested LETKF System for Cloud-resolving Model - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 495

Osaka international airports (indicated by two circles in Fig. 20.2 a) ( Tanaka and
Suzuki 2000 ). Although horizontal wind obtained by dual analyses of radial wind can
be assimilated directly as assimilation data, the regions where the horizontal wind is
estimated are smaller than those of radial wind, because the horizontal wind can be
obtained in the overlapping regions of two Doppler radars. The second kind of data
is radial wind of Doppler radar. This wind data are expected to be more effective to
improve the rainfall forecasts because it provides the information from a wider area.
Figures 20.12 a, c are the rainfall distributions at 17 JST that were obtained by the
assimilation of the horizontal wind and the radial wind. The horizontal and radial
winds from 1400 to 1500 JST were assimilated, because the rainfall regions were
fewer and smaller before 1400 JST. When the horizontal winds were added to assim-
ilation data, the number of ensemble members in which the intense rainfall regions
extended northwestward was increased from 4 to 7 (#001-004, #007-#009). When
the radial winds were assimilated, the intense rainfall regions that extended north-
westward were reproduced in 9 ensemble members (#001-#008 and #011), though
the rainfall intensity remained relatively weak. These results indicate that the wind
data, especially radial wind, can improve rainfall forecasts. This impact of wind data
was the same as that of variational data assimilation methods ( Kawabata et al. 2007 ).
To show the synergistic effects of water vapor data and wind data, simultaneous
assimilation of the horizontal wind and PWV was performed. Figure 20.12 bshows
the rainfall distributions of 17 JST. When both data were added to assimilation data,
the rainfall forecasts in the ensemble members #007 and #011, in which the intense
rainfall region was not reproduced by the individual assimilation of the PWV and
horizontal wind, were improved (Figs. 20.11 aand 20.12 a, b). The improvements
of rainfall forecasts in the ensemble members #002 and #003 became obscure,
compared with those in which the PWV or horizontal wind was assimilated sep-
arately (Figs. 20.11 aand 20.12 a, b). However, the rainfall distributions of #002 and
#003 remained better than those obtained from the assimilation of conventional data
(Figs. 20.6 aand 20.12 b). These results indicate that the simultaneous assimilation
is useful for increasing the number of members in which local heavy rainfalls are
reproduced. This result is consistent with that of Seko et al. ( 2004 ). In this study,
the synergistic effects of water vapor data and wind data were investigated by the
combination of the PWV and horizontal wind. Further improvements are expected
when the SWV and radial wind data are assimilated simultaneously, because the
impacts of the SWV and radial wind were more significant than those of the PWV
and horizontal wind. The combinations of water vapor data and wind data except
PWV and horizontal wind is to be investigated in the future study.

20.6 Summary and Future Plan

The nested LETKF system was developed to reproduce local heavy rainfalls. In
the experiments in this study, the convection cells of local heavy rainfalls were
well reproduced in one third of ensemble members of the Inner LETKF, when the