*****************************************************************
Preliminary reports indicate that between 22 and 31 people were killed in eastern Bangladesh by a tornado Sunday evening(May 4). Village Noabadi near the Indian border  was hardest hit. Many of the injured lost hands or feet
from flying debris. This area was outlooked with a HIGH risk of severe
storms in the day 5 convective outlook. A day 5 outlook was given since
I was headed out of town on the April 30. Unfortunately I downgraded to
a MODERATE risk in the day 1 outlook on May 3 since storm initiation
was in question. More information will be posted as details are available.
The forecast discussion from April 30 is as follows:

Day 3-6 discussion  updated 4-30-03 0445 UTC

An upper level ridge over the northern part of the Tibetan plateau will gradually shift east over the next 72 hours. A deep upper level trough will progress eastward from the middle east over the next 72 hours. By May 2, this trough will be centered over far northwest India and Pakistan.  Strong westerly flow at the base of this trough will tend to be suppresed just south of the Tibetan Plateau, leading to an increase in the mid-upper level wind fields immediately south of the Himalayas. The 500mb winds may increase to 45-60kts as early as May 3. If everything pans out, tornadic storms will be likely from May 2 to May 5. Thus I will go ahead and issue a moderate risk for much of central and northern Bangladesh and adjacent Assam and West Bengal, India for May 2-4, and a high risk of severe storms for May 4 and May 5 for much of the Bengal region. I may not be able to update these forecasts for a few days since I will be away from computing resources.
Days 3-4 outlook  May 2 and May 3  moderate risk of severe storms
Days 5-6 outlook May 4 and May 5    high risk of severe storms

Click here for a surface chart for 06 UTC May 4, here for the 12 UTC 500mb chart and here for an approximate sounding
for Agartala, India. Neither Agartala nor Dhaka made balloon launches at 00 and 12 UTC which is disappointing.
The 12 UTC 200mb reanalysis can be found here.

*****************************************************************