> "What happens if you remove the periods of these singular, episodes from your QBO-TSI-CW fitting or is that what you’re about to do?"

Jim, I was satisfied with the results of my model and so finished writing the paper. However, I then happened across the ENSO BEST time series which is like the SOI but less noisy and so evaluated that WRT the model. That's when the volcanic events started to show up more obviously.

I am thinking about adding in these events as transient cooling disturbances, but of course this adds more degrees of freedom to the model.

For the CSALT model, these are the kinds of residuals one sees if the volcanic events are not included:

![csalt_no_volcanos](http://imagizer.imageshack.us/a/img36/7901/4wb.gif)

After including these events the residual is reduced.

![csalt_with_volcanos](http://imagizer.imageshack.us/a/img842/8661/ayw.gif)

But the CSALT model also includes the SOI as a compensating factor, so these volcanic disturbances are potentially doubly counted if they have an impact on ENSO as well. That's what makes it very intriguing in the bigger picture of impacts.