Many of the public in Goderich were surprised by the occurrence of a Strong Tornado (i.e. F3 damage), stating they thought it was just another thunderstorm (as opposed to the reflection often given by the public in high-dewpoint events: “It felt like there was going to be a bad storm”).  Many chasers and meteorologists were also surprised with the evolution of events.

Environment Canada did a good job – a warning was out with about ten minutes lead time, which is sufficient for members of the public to take action and responsibility for their safety, if they received the warning.  EC also had a Severe Weather Watch out that did mention a tornado was possible – but did not have a Tornado Watch in effect.

This was a day where the possibility of a strong tornado was evident, if you looked in the right places.  If you looked in the wrong places you could be easily have missed it coming.

I, unfortunately had other things on my plate and did not look farther than SPC Tornado Probability guidance:


and assessed that, since the front had passed overnight, it was a reasonable solution to not expect a significant weather event, and went on with other plans (SPC categorical guidance was suggesting the possibility of general thunderstorms for the area surrounding SW Ontario).

Finding The Needle in The Haystack

Lets looks at this event in detail – starting with where NOT to look first:  Derived Indices!

Derived indices do have some guidance values in a balanced forecast.  Combined with a understanding of ‘what is happening’ –  garnered through a proper diagnosis of the current state of the atmosphere, and understanding of why it is happening, tempered use derived indices (i.e. within their known limitations) can help one arrive at a reasoned forecast for ‘what will happen’.

Lets start with something really blatant – like the Significant Tornado Index and hour before the Goderich tornado – both Fixed Layer:


… and Effective Layer.  Nothing happening here, feel free to move along…


Or Storm Relative Helicity at 3 km around the time of the tornado:


And at 1km – some higher values are evident (and potentially suspect, possibly skewed by a single ob), but not near Goderich:


1 km EHI plot shows no reason to suspect anything supercellular or tornadic (3 km was similar):

Craven-Brooks Significant Severe Index does show some very low end values for Lake Huron and the Goderich region, but suggests nothing ‘significant’.

And the Supercell Composite Index suggests, well, no Supercells:


Old indices like Lifted Index, also suggest only limited instability:



Surface-based CAPE is a little more generous, with roughly between 500 and 1000 J/kg along the path of the storm (which tracked SW from N Lower MI across Lake Huron):

And ML CAPE suggests around 500 J/kg (or less!):

If one were to do a proper surface analysis (or even just peruse the obs) in the hours before the tornado things start to raise flags for the existence of some necessary ingredients for a tornado – like a 64F Dewpoint at Goderich:

But that drops to 63F at 19z – and look at the veering of the surface flow to NNW and NW at Goderich:
And even down to 61F at around the time of the tornado.  Sufficient yes, but doesn’t really scream a strong, long-tracked tornado was imminent!

Looking beyond indices, we see both low and mid level lapse rates are only marginally favorable for Deep Moist Convection:


But things start to look a lot more favorable when we look at upper level flow:

At 12z, an unseasonably strong mid and upper Jetmax is seen rotating through a sharp upper and mid level trough seen here at 250:

By 00z Monday, Goderich is under the Left Exit region of the cyclonicly-curved jet max – the associated vertical motions would likely contribute to destabilization and an erosion of convective inhibition:
Water vapour shows much the same, with a dark channel associated with the jetmax over Lake Michigan earlier in the day:
Advection of dry air from the NW implies the presence of an EML, though the weak lapse rates suggest that it has eroded.

00z 500 mb charts also suggest a favorable flow over Goderich, possibly greater than 50 knots:


No real weakness in the 700 mb and 850 mb flow:


An examination of Storm-Relative flow paints a much more favorable scenario – even suggesting Classic Supercell modality  – as SR flow was sufficient at all levels for supercells.





and mid-level SR flow should displace precipitation away from the updraft:


0 – 2 kn SR flow is sufficient and directionally opposite from the projected and observed storm flow:


This storm was seeing storm-relative inflow VERY different from the NW sfc flow at Goderich!

The 0 – 6 km shear vector is certainly favorable for rotating convection:


And even stronger from 0 – 8 km:


CAPE Density favorable for tornadogenesis (in the first three kilometers above ground level) in relativity close proximity to the storm at 19z and 20 z may have also played a role:

It was apparent that a Classic Supercell convective mode was in progress for much of the storm’s translation over Lake Huron, as evidenced by the Exeter, Ontario’s Reflectivity image (note prominent hook echo):
And just before affecting Goderich:

Photographic evidence shows a large tornado was in progress over the lake (tornado wrapped in rain/spray – corresponding withe the ‘ball’ at the end of the hook).  Image courtesy of Ruth Montgomery.  It would be reasonable to suggest that it was in progress for a substantial period of time before this image was taken by Ruth Montgomery as it approached Goderich:


The large tornado also continued after passing trough Goderich – image courtesy Joe Gowanlock:


And retained impressive structure as a large tornado – image courtesy of Danica Harvey:



Necessary and sufficient conditions for a significant tornado were in place as the afternoon evolved on the afternoon of 21 August, 2011.  Some of these ingredients may have been barely sufficient, and thus skewed various derived parameters that are are often used to suggest tornado potential.  The passage of a surface cold front would suggest, if one was simply relying on pattern recognition, that the probability of a strong, long-tracked tornado, was very low.  This events underscores how reliance on one or both of these approaches is inadequate when attempting to forecast Deep Moist Convection.

If one looks directly at the ingredients needed for significant Deep Moist Convection (Conditional Instability, Moisture, and Lift) all were present and sufficient before this event.  Additionally a shear and storm-relative flow pattern (along with low-level CAPE and low Lifting Condensation Level) was very favorable for classic supercells and tornadoes.  A focus on a complete top-down diagnosis of the atmosphere would have made this apparent, and a dedicated monitoring of the evolution of the atmosphere should have highlighted the potential for a significant event.

 Next I’ll contrast this event with the one that occurred just three days later…