>NINO3.4 Data Comparison – HADSST and ERSST.v2


In a previous post “Standardized versus Raw (Not Standardized) NINO3.4 SST Anomaly Data” (11/13/08), which I have removed, I mistakenly attributed the differences between the two data sets to standardization. The standardized NINO3.4 SST anomaly data that I used in that post was prepared by Trenberth and Stepaniak. Trenberth and Stepaniak discuss their NINO3.4 index and its recipe here:
Their updated 1871 to 2007 NINO3.4 data is here:
Investigating the source of the Trenberth and Stepaniak NINO3.4 SST data further, it’s based on Hadley Centre HADSST data, which is available through the NCDC:
The reference to HADSST is at the bottom of that webpage. The HADSST source is confirmed by the NCDC discussion page:

With the background of this post complete, there are underlying differences between the HADSST and ERSST.v2 NINO3.4 SST data, with the most significant differences occurring before the mid-20th Century.

There are also small and subtle differences depending on whether the NINO3.4 anomaly data has been standardized or left in its raw state. For this reason, I’ve illustrated it both ways in the following. But first, there’s the…


Figure 1 compares the HADSST and ERSST.v2 versions of NINO3.4 SST (not anomaly) data. The two track well from the mid-1950s to present. Prior to that, they diverge and the differences can be substantial.

Figure 1

Figure 2 illustrates the difference between the ERSST.v2 and HADSST versions of NINO3.4 SST, where the ERSST.v2 data was subtracted from the HADSST data. The difference indicates there is an underlying trend in the SST data of the ERSST.v2 version that doesn’t appear in the HADSST data.
Figure 2


Figures 3 and 4 provide comparisons that are similar to the above, but in these graphs, SST anomaly data are compared. The same base years were used for both data sets, 1950 to 1979, which are those used in the referenced Trenberth and Stepaniak data. The comparisons of SSTs and SST anomalies are very similar.
Figure 3

Figure 4


In Figures 5 and 6, I took the process one step farther and compared standardized versions of the ERSST.v2 and HADSST NINO3.4 data. There are subtle differences between the raw and standardized data sets, though you’d have to flip between the images to see them. The most obvious is that the variations have been exaggerated by the standardization process. This can be seen by comparing the temperature scales and the peaks of the major El Nino events.

A note about the standardization used: I followed the recipe given by Trenberth and Stepaniak in the first link above, with the following exception: I did NOT smooth the data with a 5-month filter prior to normalization.
Figure 5

Figure 6


In the earlier post that I’ve just deleted, I had included graphs of the Trenberth and Stepaniak NINO3.4 data for use as references. The following are those graphs. The data in Figure 7 has not been filtered by me. Figures 8 and 9 have been smoothed, with 12- and 25-month filters.
Figure 7

Figure 8

Figure 9


The source of the HADSST data is discussed in the introduction, above.

The Smith and Reynolds Extended Reconstructed SST (ERSST.v2) data is available through the NOAA National Operational Model Archive & Distribution System (NOMADS).http://nomads.ncdc.noaa.gov/#climatencdc

The reasons for smoothing ENSO-related data with a 25-month filter are discussed here:

About Bob Tisdale

Research interest: the long-term aftereffects of El Niño and La Nina events on global sea surface temperature and ocean heat content. Author of the ebook Who Turned on the Heat? and regular contributor at WattsUpWithThat.
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s