During South Asian summer monsoon season, depressions formed over the Bay of Bengal (BoB) bring great amount of precipitation along their paths. These depressions could be destructive and fatal if they traverse over populated urban areas in India, Bangladesh, or even Burma. With the Advanced Weather Research and Forecast (ARW) model simulations and various observations including satellite imagery, radiosonde observations, and global reanalyses, the depressions formed in August 2006 were studied extensively to further quantify the dynamical predictability of monsoon depressions. According to the NCEP/NCAR reanalysis and the precipitation observed in the Tropical Rainfall Measuring Mission (TRMM 3B42), four depressions formed over the northern tip of BoB and evolved in a similar pattern. With life spans from 3 to 5 days, the depressions propagated westward or west-northwestward after their genesis while producing heavy rainfall on their flanks and gradually weakening as they moved further inland. Hindcast experiments with the ARW model are performed for the event during August 2-5, 2006.

In the preliminary experiment, a two-way two-nested domain is set up in ARW. The outer domain covers from 7°S to 40°N and from 40°E to 125°E which includes the entire India, the Bay of Bengal, parts of the Arabian Sea, parts of the South East Asia, and the Tibetan Plateau with 30-km horizontal resolution. The inner domain is set up with 10-km resolution to better resolve the mesoscale organizations in the depression. Boundary conditions and initial conditions are generated from the NCEP Final (FNL) global analysis and the NCEP Real-time Global Sea Surface Temperature (RTG SST) analyses.

Model sensitivity to cumulus parameterizations has been examined using Kain-Fritsch, Betts-Miller-Janjic, and Grell-Devenyi ensemble schemes. In all experiments, the initial location of the depression at the northern tip of BoB is correctly simulated; its movement towards the seasonal low over Pakistan is captured. However, the model still has difficulty realizing characteristics which would have great societal and economic impacts during the evolution of the depressions, such as the detailed trajectory and the extreme precipitation. The simulated depression moves either too slowly that consequently brings much greater accumulated rainfall for central India, or too rapidly and dissipates too soon. In contrast to the observation which shows asymmetric precipitation concentrated on the southwestern flank of the observed depression, the model rainfall tends to be symmetrically distributed. Simulations with higher resolution aimed at resolving cloud systems are to be conducted. The influence of the steady-state monsoon circulation on the formation and propagation of the depressions will be assessed. Ensemble experiments will be conducted to further quantify the dynamical predictability of monsoon depressions.