Impact of Variational Data Assimilation for Simulating Tropical Cyclones over Bay of Bengal Using WRF-ARW - Monitoring and Prediction of Tropical Cyclones in the Indian Ocean and Climate Change - page 238

To analyse the model simulated rainfall intensity in both experiments, WRF
Model 24 h accumulated rainfall for the fourth day forecast in control and
3DVAR experiments are compared with TRMM estimates (Fig. 5) for Jal (top
panels) and Nisha cyclone (bottom panels) cases. The model simulated rainfall
in control experiment showed peak rainfall of 12-18 cm from Cuddalore to
Vedaranyam and extended longitudinally up to Mangalore. The 3DVAR
simulated rainfall pattern for the fourth day (07 th Nov 2010) well matches with
TRMM estimates. For Nisha cyclonic storm, TRMM rainfall estimates on 27 th
November 2008 indicate the maximum rainfall of upto 15 cm/day along the
south-east coast of India. CTL experiment over predicted the rainfall with heavy
rainfall peak 30 cm/day distributed over south central India. It failed to simulate
observed rainfall pattern over Tamilnadu. The 3DVAR experiment better
predicted the rainfall distribution as in TRMM observations both in intensity
and distribution. There is significant reduction in the over rainfall amount in
3DVAR than the control. To study the performance of the model-simulated
rainfall pattern, simulated 24 h rainfall pattern for four days in the both
experiments for 10 tropical cyclone cases are compared with TRMM spatial
rainfall data (Table 2). This shows that the domain average rainfall in 3DVAR
experiment is significantly less than Control experiment from the first day
onwards. The bias between the TRMM and model-simulated rainfall in both
experiments started increasing from 1.5 mm/day to 4 mm/day, though bias in
3DVAR experiment reduced significantly from 0.2 mm for the first day to
Fig. 5: Model simulated rainfall for 4 th day for two tropical cyclones Jal
(top panels) and Nisha (bottom panels).