The Southern Oscillation Index (SOI) presented below is computed using monthly mean sea level pressure anomalies at Tahiti (T) and Darwin (D).
The SOI [T-D] is an optimal index that combines the Southern Oscillation into one series
The SOI noise [T+D] series is a measure of small scale and/or transient phenomena that are not part of the large scale Southern Oscillation
These SOI values are similar to those calculated by the Climate Prediction Center in that they have been derived using normalization factors derived from monthly values.
The SOI values prior to 1935 should be used with caution. There are questions regarding the consistency and quality of the Tahiti pressure values prior to 1935. Ropelewski and Jones (1987) describe the data sources and methods used to extend the data. Further comments are provided by Allan and Ansell (2006).
The smoothed curves below were created using a a filter which effectively removes fluctuations with periods of less than 8 months but includes all others. At 24 months 80% of the variance is retained. The smooth curve denoted by a thick black line is that produced using a decadal filter over thge original monthly values.
As noted above, the SOI presented here are derived using monthly values as was done in Trenberth (MWR, 1984). However, Trenberth notes that better signal-to-noise ratios may be obtained by using normalization factors based upon annual means. To view figures and to download data derived using this approach
The data used are available in both netCDF and ASCII formats. The original sea level pressure datasets for Darwin and Tahiti from the Climate Prediction Center are contained on four files. The following datasets have been merged to facilitate use. (1) Click on the high lighted text and (2) use "Save As..." from the "File" dropdown menu to save the file.
The data contained on the ASCII files may read via a fortran format statement read (*,"(i5,(12f6.1))") iyear, (x(m),m=1,12)
Note: a -99.9 indicates a missing value.
Finally, the sea level pressure data spanning the period 1951-1980 were used to derive the monthly means used to derive the anomalies and standard deviations used in the calculations.