> I don’t think the IOD will help to compensate the ENSO model too much. Certain characteristics, such as the strong peak at 1941, do not make a strong contribution. Many of the peaks do show up, but it is the odd ones here or there which reduce the correlation. IOD is also a noisier data set from appearances, so that doesn’t help either.

Many thanks for the analysis. I'll see if I can see this. What is it you think is worthwhile investigating next?