On April 14, 2004, a tornado moved southeast across northern Bangladesh, killing 75 people. Daily news reports
        attest to the violence of the tornado. A journalist from Sylhet visited the affected area and reported the tornado damge
        path was over a mile wide in spots. My convective outlook issued at 0100 UTC(700 am BST) discussed the possibility
        of tornadic storms near the outflow boundary in northern Bangladesh.  I updated the discussion at 08 UTC and upgraded
        from moderate to high risk for parts of central and northern Bangladesh.  It was not clear from the sparse surface data
        just how close the outflow boundary was to the Khasi Hills. In retrospect, the outlook should have been a hair further
        north. The tornadoes occurred in the moderate risk area. It sure would be nice if we had a surface station at Mymensingh.
        I issued a tornado watch at 0820 UTC(220 pm BST). At the time, I only had access to global view--5km res meteosat-5
        data. After the fact, I was given a glance of much higher resolution, half-hrly, meteosat-5 data which is proprietary. At
        6 pm, no more than a large towering cu was visible on satellite near the affected area. The tornadoes occurred in the
        extreme northern part of the tornado watch area between 1245 UTC and 14 UTC(645-8 pm BST)  In fact, part of the
        tornado paths were a few miles outside the watch area. But as the saying goes in the United States, tornadoes may occur
        in and close to the watch area. It is still not clear whether there were one or two tornadoes. I will be getting maps of each
        district of Bangladesh soon, with all the villages shown. I will then trace the path of the tornadoes, which may shed some
        light on this issue.
 
    Here are surface charts, upper air charts and model initializations for April 14:
 
                surface maps: 06 UTC    09UTC  12 UTC
                500mb: 00 UTC

                  500mband  300mbUKMET initialization for 12 UTC April 14(obtained from vortex.plymouth.edu)
                  300mb GFS initialization for 18 UTC April 14(obtained from vortex.plymouth.edu)
                  UKMET maps with terrain for 00 UTC April 15
                          500mb wind barbs, height,
                          700mb wind barbs, height, temp
                          850mb wind barbs, temp
                    GFS maps with terrain for 12 UTC April 14
                          300mb  wind barbs, temp

       New  Morning soundings from April 14 and  April 15 were modified for conditions 12 hour later and 12 hours earlier.
                The sounding were only modified below 900mb to reflect boundary layer changes. Also, the April 15 sounding
                was missing some upper level data and this was approximated.  The capping inversion was apparently weaker
                across far northern Bangladesh near Meghalaya, India.