Geoscience Reference
In-Depth Information
However, some AWSs were destroyed during the 24 h period due to intensive
rainfall. Therefore, we interpolated the irregular (also with different amount of
stations) hourly precipitation to the corresponding model grid spacing (i.e., 25 km),
and then added the above gridded precipitation in 24 h to a total precipitation
observation. We only consider the model grids falling into the territory of South
Korea, thus the total number of points
N
to be verified is 129 for a 25 km grid
spacing.
27.4
Results
The micro-GA is performed for optimal estimation of two parameters in the KF
scheme - the convective time scale (
), for the
case of Typhoon Rusa (2002) using the WRF. The optimized values are obtained in
100 generations (with population size of 5) using the micro-GA. For the parameter
T
c
, the optimized value is about 1922 s which locates in the default range. However,
for the parameter
T
c
) and the auto-conversion rate (
c
s
1
that is two orders smaller
c
, the optimized value is
0:0004
s
1
/
than the default value
in the KF scheme implies that the
condensed cloud water is more detrained to the grid-resolved environment rather
than converted to convective precipitation falling down (
Correia et al. 2008
).
Figure
27.1
compares the ETSs of three groups of experiments (NOCP, KFEX
and OPTM). It turns out that the ETSs with OPTM are much higher than those
with NOCP and KFEX in all thresholds, indicating a significant improvement in
quantitative precipitation forecasts. The sum of ETS with OPTM, KFEX and NOCP
is 7.84, 2.78 and 2.33, respectively. Here, both KFEX and NOCP have almost
no forecast skill for light precipitations (
.
c D
0:03
. A smaller
c
mm d
1
), while NOCP generally
performs better than KFEX at light to moderate precipitation rate (
<60
mm d
1
).
< 200
mm d
1
), NOCP loses forecast skill abruptly but
KFEX has considerable forecast skill - higher than that for moderate precipitation
rate and even closer to OPTM. This indicates that the KF scheme is useful for
forecasting high precipitation rate at the grid resolution of 25 km, relevant to the
forecast with optimized parameters.
Figure
27.2
a depicts horizontal distribution of the observed 24-h precipitation
amount from 1200 UTC 30 to 1200 UTC 31 August 2002. Heavy precipitations
(
For heavy precipitate rate (
> 200
mm d
1
) are observed at the northeastern to southwestern parts of South
Korea with three regions of local maximum (
> 100
mm d
1
) located at mountainous
areas - one at the north-eastern coast of South Korea near the eastern side of the
Taebaek Mountain Range and two at the southern side of the Sobaek Mountain
Range. Correspondingly, light to moderate precipitation (
> 400
mm d
1
) occurs at
the northwestern part of South Korea. More details of the precipitation processes in
this typhoon have been analyzed by
Park and Lee
(
2007
)and
Lee and Choi
(
2010
).
Figure
27.2
b represents the 24-h accumulated precipitation from 12 UTC
30 August to 12 UTC 31 August 2002 from three experiments (NOCP, KFEX and
OPTM) with a 25 km horizontal resolution. OPTM shows the best agreement with
5
-
100
Search WWH ::
Custom Search