Xh. Wu et al., SHORT-RANGE PRECIPITATION FORECASTS USING ASSIMILATION OF SIMULATED SATELLITE WATER-VAPOR PROFILES AND COLUMN CLOUD LIQUID WATER AMOUNTS, Monthly weather review, 123(2), 1995, pp. 347-365
These observing system simulation experiments investigate the assimila
tion of satellite-observed water vapor and cloud liquid water data in
the initialization of a limited-area primitive equations model with th
e goal of improving short-range precipitation forecasts. The assimilat
ion procedure presented includes two aspects: specification of an init
ial cloud liquid water vertical distribution and diabatic initializati
on. The satellite data is simulated for the next generation of polar-o
rbiting satellite instruments, the Advanced Microwave Sounding Unit (A
MSU) and the High-Resolution Infrared Sounder (HIRS), which are schedu
led to be launched on the NOAA-K satellite in the mid-1990s. Based on
cloud-top height and total column cloud liquid water amounts simulated
for satellite data, a diagnostic method is used to specify an initial
cloud water vertical distribution and to modify the initial moisture
distribution in cloudy areas. Using a diabatic initialization procedur
e, the associated latent heating profiles are directly assimilated int
o the numerical model. The initial heating is estimated by time averag
ing the latent heat release from convective and large-scale condensati
on during the early forecast stage after insertion of satellite-observ
ed temperature, water vapor, and cloud water information. The assimila
tion of satellite-observed moisture and cloud water, together with thr
ee-mode diabatic initialization, significantly alleviates the model pr
ecipitation spinup problem, especially in the first 3 h of the forecas
t. Experimental forecasts indicate that the impact of satellite-observ
ed temperature and water vapor profiles and cloud water alone in the i
nitialization procedure shortens the spinup time for precipitation rat
es by 1-2 h and for regeneration of the areal coverage by 3 h. The dia
batic initialization further reduces the precipitation spinup time (co
mpared to adiabatic initialization) by 1 h.