Maybe you can't? The suggestions above are all interesting but only point to the occurrence of a high impact, high energy, atmospheric event. That is certainly one definition of a tornado, but strong straight line winds (derechos) or just big old storms can produce similar results in the amount of run-off and debris. The impact of a tornado is so localized and relatively short lived that its direct preservation potential in the sedimentary record is small and the indirect evidence will always be inconclusive. I think the only way I could be convinced is if a clearly defined corridor or "destruction" were exposed on a large bedding surface, uprooted trees with standing in-situ roots on either side of a sharply defined linear track of destruction. I love the question! -and am quite happy to be rebuffed :-)

Hi Mark. I think that while shredded plant material might be an indication of tornado occurence, you might have to effectively rule out other potential causes of shredded vegetation - such as severe flood events. What sort of scale are you working on (landscape level)???

The quantity of shredded plant material would have to be very consentrated, along with chaotic sedimentation. What age and scale are we dealing with?

I am considering this as a possible explanation for particulate carbonized plant fragments in the Mesozoic rift valley lacustrine strata in the Connecticut Valley region. What criteria, if any, might we use to distinguish flood from tornado fragmentation of fossil plant material? Most accumulations are scattered on a single bedding plane.

Is there any evidence that your plant material fragments are oriented in the same direction? This might indicate a flooding event rather than a tornado (although my knowledge of tornado-related material deposition is not good). The fact that these fragments are in lacustrine strata may make it a little harder to distinguish flood from tornado. I was going to suggest looking at diatom content to show flooding as a possibility, but clearly that won't work in this case.

Prof Jon Nott works on cyclones and their impact on landforms (e.g. http://dx.doi.org/10.2112/05A-0005.1 ) - you might get some ideas from this paper and others he has written about associated tsunamis and deposition of coral above normal tidelines, extensive sand deposition and so on.

Yearly drops of Br content mark extremal floods during typhoon events 1-3 times per decade. Source data - recent bottom sediments in Amur Bay near Vladivistok (Japan Sea), which were analysed with step ~0,5 mm corresponding ~3 month of sediment accumulation time

Considering 1) tornadoes are largely indiscriminate in the size of materials they pick up and 2) you are looking at lacustrine strata, I would think you would potentially observe a pulse of sediments (poorly sorted, no directionality indicative of flow) including larger clasts amongst the fine lacustrine layers (i.e. dropstones), as well as possibly large fragments of "shredded" vegetal material, as suggested.

Very difficult indeed, specially if the rocks are of Precambrian age. Still we should look for strong storm effect, thick massflow deposits with poorly sorted material with outsized clasts.

Unsorted sedimnet with mixture of hetrogeinity in all respect is understood as material belonging to a natual catastrophic. One needs to relate such rock with possible catastrophic s occuring in a givenregion. Look at the source material and correlate with directions of progress of such events

HCS, Hummocky Cross-stratification (or "Crazy Bedding", sensu Howard) could be one way, although it is generally believed to form below fair-weather wave base.

If there is no size discrimination among the plant debris, as well as a highly disruptive bedding plane, with an erosive base, it could be tornado related. You would want to correlate facies tracts along the same bedding plane to see how far laterally the plant debris continues. The best way to distinguish from a flood event is directionality and 'imbrication' of the plant debris that could give you a flow direction would rule out tornado. You should also find very porrly sorted conglomerate around the area that you find the 'tornado' deposit.

Among the other things I said above, hummocks and swales can possibly indicate a storm even like a tornado as Michael Risk stated.

Maybe you can't? The suggestions above are all interesting but only point to the occurrence of a high impact, high energy, atmospheric event. That is certainly one definition of a tornado, but strong straight line winds (derechos) or just big old storms can produce similar results in the amount of run-off and debris. The impact of a tornado is so localized and relatively short lived that its direct preservation potential in the sedimentary record is small and the indirect evidence will always be inconclusive. I think the only way I could be convinced is if a clearly defined corridor or "destruction" were exposed on a large bedding surface, uprooted trees with standing in-situ roots on either side of a sharply defined linear track of destruction. I love the question! -and am quite happy to be rebuffed :-)

I would think you would need something more peculiar to Tornadoes like trees

twisted in a spiral, and possibly tree branches passing through tree trunks.

I would think the possibility of preservation of this would be far more un-

likely than the preservation of flooding where new deposition of materials

rapidly covered and protected, and preserved the detritus carried and

deposited by flood waters.

The key to most preservation is rapid burial. This includes both plant

and animal remains. Rapid Burial can occur through flooding,

wind ( dunes ), or debris flow, or mud flow, or collapse of an embankment,

or volcanic material like ash fall.

I do not see rapid burial as even a remote possibility associated with

tornadoes.

I think the shredded plant material can used for one part of an indication, but it is not enough. May be you need observe the scratch of the rock, whether is have some scratch in the same place for the plant or not . If there have some scratch near the plant. May be it is offected by the tornado occurrence. And may be need fine some plant fragment in the scratch, and test by AMS 14C, then we can get the information about when the tornado occurrenced.

I have observed the effects of storms in an endoreic lake in Spain where stromatoliths were growing. After the storms, large pieces ( up 50 cm or 1 m ) of the upper part part of bacterial mats had been 'peeled off' and were floating at the surface of the lake. May be there exists somewhere a fossil record of such a situation, showing large piece of mats fossilized in a chaotic way.

The absolute, strongest tornado's - can suck up, (completely removing) all the asphalt from a housing subdivision's road(s).  I saw a photograph once where such a tornado moved over a green pasture.  It sucked up all the sod, removing 2' of top soil - some 50 yards wide by 200 yds long.  It was some impressive scar.

Look for this geologic signature that should remain.  Also, there will not be creek bed rocks withing the bottom of this channel.