This is an Excel Solver fit to ENSO without the wave equation transform applied. Only the angular momentum variations of 14 and 6.48 year periods along with the 3 aliased tidal periods were included.

![fit](http://imageshack.com/a/img923/383/C0aeGQ.png)

The training was divided into a 70 year early interval and a 63 year later interval. They essentially capture the same underlying time-series profile, suggesting that the ENSO process is ergodic on a relatively short time scale.

Again, this is not easily detected unless we assume that a biennial Mathieu modulation is applied to the wave equation and that a biennial phase inversion between even-and-odd years applies between the years 1980 to 1996.