Previously, I have talked about isolating the long term signal in climate data using a technique I thought up. Today, I’ve used it to create an index of ENSO with no long term variation and a sable level of variance throughout the series, which should have not the sort of thing that doesn’t raise questions about “contamination” of the index with the “forced signal”. Using the HADISST1 data for the NINO 3.4 index at KNMI Climate Explorer, (because the newer HADSST3 data are sadly incomplete, but HADISST1 is temporally complete) I first removed the long term component, then took absolute values of the deviations from it. I then computed the long term component of that, for the variations in the variance. That was divided by it’s long term average, and then the NINO 3.4 series that had the long term component removed is divided by that. The result is an ENSO data set which has a stable mean and variance by construction. Why would I do this? Heck if I know, mostly because I just like playing with data. I also wanted to have an ENSO index I could make reference to without people saying it was “contaminated” by AGW. This series cannot contain any AGW component, again, by construction, although I have probably also removed natural inter-decadal variability. So this is what the index looks like, with the original data in red along with it:
August 25, 2011 at 1:44 am |
[…] was curious what would happen if I used my “Invariant ENSO Index” to look at ENSO impacts in the US. So here is El Nino versus La Nina temperature […]
January 17, 2014 at 12:52 am |
[…] previously tried to create an index for ENSO which would have a stable long term mean and variance. Now, using […]