Data Assimilation ofWeather Radar and LIDAR for Convection Forecasting and Windshear Alerting in Aviation Applications - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 526

Guangdong. As a result, the areas of convective developments were “pushed” to the
south over the coastal waters of Guangdong, moving away from Hong Kong. As a
result, the rain over the territory weakened gradually.

22.3.3 Data Assimilation and Forecast Experiments Using

RAPIDS-NHM

Numerical experiments on the data assimilation and forecast using radar retrieval
winds in RAPIDS-NHM are conducted in 0300 UTC run. As the vertical velocity is
not a control variable in the 3DVAR system of RAPIDS-NHM, only the horizontal
components of radar retrieval winds are ingested. Data thinning is applied to the
retrieved radar wind data in horizontal direction using a grid separation of about
4 km. Quality control procedure is then used to remove those suspicious radar
retrieval wind data that are opposite in direction to the wind vectors in model first
guess of 3DVAR. To study the impact of retrieved wind data, control experiments
(CNTL) are used in which all available observations except radar retrieval winds are
ingested in 3DVAR.
Figure 22.6 shows the analyzed wind and relative humidity (WIND C RH) on
500 hPa and 700 hPa levels using the radar retrieval wind (upper panel) and CNTL
(lower). Radar retrieval winds and reflectivity at 5 and 3 km of altitude are shown
in same figure. It can be seen that the retrieval winds are effectively assimilated in
the 3DVAR, resulting in enhancement in southwesterly flow over the coastal waters
of Guangdong and generate a short-wave westerly disturbance on 500 hPa level.
In CNTL, only moist westerly flow is found in the analysis field. Convergence in
lower troposphere is also enhanced as the southwesterly winds are analyzed over the
coastal waters using the radar retrieval winds data, thus resi;tomg in improvements
in the forecast cyclonic shear on 850 hPa is shifted to the vicinity of HK and
Pearl River Estuary where it is located more to the north in CNTL (Fig. 22.7 ). The
main cyclonic shear is located over the northern part of HK corresponding to the
intense convection development areas labelled as A and B in Fig. 22.4 . In CNTL,
the low-level cyclonic vorticity is forecast over the northeastern part of HK only, in
accordance with actual active development area (Label A in Fig. 22.4 ).
Figure 22.8 depicts the 5 h forecast equivalent reflectivity (derived from
RAPIDS-NHM forecasts of specific humidity of cloud hydrometeors) and the
actual radar imagery. The east-west oriented reflectivity band is better forecast
with the assimilation of radar wind retrieval data, whereas the intensity of simulated
reflectivity is appreciably weaker in CNTL. However, the model forecast reflectivity
field shows a time lag by about 1 h as compared to the actual radar image. Another
difference from the actual radar image is found for the convection over the coastal
sea areas, in which both experiments do not show any sign of the storm development
over the region. To investigate whether this discrepancy could be alleviated through
the use of hourly-update cycle of RAPIDS-NHM, the numerical experiment is
repeated for analysis and forecast at 0400 UTC. The first guesses of 3DVAR in