That's an interesting topic to take a look at -- I haven't heard of that one before.

Tornadoes tend to form in relatively warm environments in the presence of abundant moisture and heating at the surface, with relatively cool mid-levels of the atmosphere. This provides the thermodynamic energy for the formation of the tornadoes; many other dynamical processes serve to initiate convection, form the preceding horizontal vortex, tilt it into the vertical, and lead to tornado formation, notably wind shear.

But, the thermodynamic parameters are the important ones here -- not only as they can lead to tornado/storm formation but are less likely to change on a day-to-day basis as are the dynamical parameters. Thus, a similar regime is likely to hold through, oh, a week or two -- whereas the dynamical parameters are tied to many other transient features (e.g. surface cyclone development, upper-level conditions, and more).

What do we know about hurricanes? We know that they need warm water to form, particularly not just at the surface but also below the surface. This warm water pools as a result of increased heating over a region, but takes take as waters warm/cool much slower than does land (due in part to the specific heat capacity of water vs. land). Of course, there are many other factors -- low vertical wind shear, sufficient low-level spin to result in vortex/convective development, and so on -- but again, these are more dynamical than thermodynamical.

Most tornadoes/convective systems across the southeast U.S. (mainly central Texas & points east) receive their moisture from the Gulf of Mexico, with heating serving to continually warm these waters through time. So, without doing any study, it seems as though early-season tornado activity in the S.E. U.S. could be a good indicator of tropical activity across the western Atlantic -- particularly the Gulf -- at least early in the season. Of course, like in 2004 as opposed to a lot of other years, this may not be terribly well-correlated to hurricane activity in that basin. I don't think that *all* tornadoes are going to be a good indicator of activity, just those in the southeast U.S.; I also think that while tornadoes are a good parameter (and simple to use), observed severe thunderstorms may be a better indicator, since the basic conditions for formation are the same between the two and using the latter does not require as much dynamical response, making it likely more representative of what we are looking for here.

Hope this wasn't above everyone's heads...comments anyone?