Came across a paper from 2004 by Jim Kossin and Chris Velden at the Univ. of Wisconsin, available at http://www.ssec.wisc.edu/~kossin/articles/kossin_velden_MWR.pdf, detailing errors in the Dvorak technique. Without having to read the article, here's the quick summary...

* There is a prononunced latitude-dependent bias in sea-level pressure estimates using the Dvorak technique, with MSLP values generally being too high (lowering intensity estimates) at higher latitudes. The bias percentage is roughly 10% at higher latitudes (before extratropical transition) and is near-zero around 23N/Tropic of Cancer.
* However, a paper by Chris Landsea (and others) in 2003 suggested that for a given pressure obtained via the Dvorak technique, the wind estimate is generally too strong, a feature that grows worse with increasing latitude.
* These two "biases" contrast and, as Kossin & Velden discovered, nearly cancel each other (all but about 2% of the variance), which makes sense. Lower pressures correspond to higher wind speeds, all other factors being equal, so a bias towards higher pressures that is corrected is going to tend to lead to higher wind speeds...a factor already (mistakenly) accounted for by the Dvorak estimate of wind speed.

What does it all mean? Given that there has yet to be an update of the Dvorak technique with these two studies in mind to account for the latitude bias, it means that our pressure estimates for storms at high latitudes like Irene are generally too high. For instance, the objective Dvorak pressure estimate from the UWisconsin website (http://cimss.ssec.wisc.edu/tropic/tropic.html) for Irene corresponds to a pressure of 964mb -- or 11mb lower than the 5p NHC estimate. This meshes well with the ~10% bias discovered in the paper. Their wind speed estimate of 82kt matches well with their Dvorak estimate of T4.7 and is just below that expected for a T-number of 5.0 (as given by SSD and used in the initial intensity estimate for Irene by the NHC at 5p), suggesting little modification needed (or made) to the wind speed estimate.