One way to prevent overfitting of a model is to use parameters that are fixed by an independent measurement. In the case of forcing for ENSO, and according to the LTE model, this is all ultimately related to angular momentum changes acting on the fluid. The angular momentum changes are reflected in measurements of the dLOD of the earth's length of day anomaly. The same lunisolar gravitational force that slows and speeds up the earth's rotation rate also impacts the sloshing of the ocean.

This is a model of ENSO with the top panel fitted according to the dLOD in the scans below

![](https://imagizer.imageshack.com/img921/1227/yTjzaJ.png)

To extrapolate the LOD data before 1962 (which is when the high-resolution measurements started), the tidal pattern was decoded as a Fourier series.

![ft](https://imagizer.imageshack.com/img924/3396/oT68Vm.png)

This allowed the LOD data to be extended to the years before 1962.