Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
Low (web) Quality - BALTEX
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
88<br />
of the simulations due to the difference of the initial<br />
conditions. The dark bars in Figure 2 show the results from<br />
the simple vertical interpolation (P2S), whereas the white<br />
bars indicate the incremental interpolation (INC). The<br />
incremental interpolation significantly improves regional<br />
simulation for nearly all ranges of pressure levels with the<br />
exception of precipitation in 17L, 9L, and 3L. The<br />
performance of the 7L results became very similar to that of<br />
17L without the incremental interpolation (P2S-17L), and<br />
even 3L produced a reasonably good regional simulation<br />
compared to P2S-17L. Therefore, from a practical point of<br />
view, approximately 5 pressure levels will be sufficient to<br />
obtain reasonably accurate regional simulations. We should<br />
note that the improvement is more apparent for 2-meter<br />
temperature and 10-meter winds. Reasonable improvement<br />
is also seen in precipitation.<br />
resolution. Integration was done for one year (year of 2047<br />
in A1B scenario, according to MIROC experiment) with<br />
identical initial conditions.<br />
Figure 3 shows monthly precipitation distributions<br />
and their difference from CTL. Due to less vertical<br />
information given by the forcing data, more precipitation<br />
is simulated over the Central Valleys in COA, whereas the<br />
wet condition is fixed in INC, even though exactly same<br />
amount of information has been used. As shown in Figure<br />
4, this is mainly due to more surface convergence<br />
simulated in COA.<br />
Total Precipitation<br />
*Precipitation intensified<br />
over central valley.<br />
improved<br />
Figure 3. Monthly precipitation from 10-km<br />
regional dynamical downscaling of MIROC’s global<br />
future projection results. Upper panels show<br />
monthly distribution and lower panels show<br />
difference from the control (CTL).<br />
Surface Wind field<br />
Figure 2. Ensemble means of area averaged RMS<br />
between CTL and experiments with different numbers<br />
of vertical levels used as forcings are shown for 2-<br />
meter air temperature (a), 10-meter wind speed (b) and<br />
precipitation (c).<br />
4. Application to Regional Future Projection<br />
Though highly expected, few of IPCC/WCRP’s CMIP3<br />
simulation results can be directly used as lateral boundary<br />
condition for dynamical downscaling study because these<br />
data are archived with small number of vertical levels and<br />
scarce temporal interval (at most daily). We therefore<br />
investigated the impact of small number of vertical levels<br />
and longer intervals of forcing data, particularly focused on<br />
a purpose of the regional future projection. Japanese T106<br />
MIROC simulation results are used, and all 23-level data are<br />
used as lateral boundary for the control regional integration<br />
(CTL). Similarly to the previous section, we applied the<br />
simple vertical interpolation from coarse vertical data<br />
(COA) and the incremental interpolation (INC) from the<br />
lowest 9 levels (up to 200 hPa). In this experiment, we used<br />
the ECPC RSM, but for a domain covering western U.S.A.<br />
and Mexico and vicinity oceans in 10 km horizontal<br />
*Virtual convergence<br />
leads more precipitation<br />
improved<br />
and cooling.<br />
Figure 4. Same as Figure 3, but for surface wind<br />
fields.<br />
References<br />
Yoshimura, K. and M. Kanamitsu, Specification of<br />
external forcing for regional model integrations, Mon.<br />
Wea. Rev., 2009. (in print)<br />
Kanamaru, H. and M. Kanamitsu, Scale-selective bias<br />
correction in a downscaling of global analysis using a<br />
regional model, Mon. Wea. Rev., 135, 334–350, 2006.