The data are here.
First, an anomaly field is computed on the array grid by subtracting climatological averages from the quarterly means after bilinearly interpolating the climatology field in latitude and longitude to the array grid. Second, the anomalies are linearly interpolated, first in latitude, and then in longitude, to a 1 deg by 1 deg grid. Third, the interpolated anomaly field is smoothed by passing a gappy running mean filter over the data twice, which approximates a Hanning filter by a triangle, first in the zonal direction, and then in the meridional direction. Triangle filter widths are 21 deg in the zonal direction, and 5 deg in the meridional direction.
Finally, this gridded and smoothed anomaly field is added to the climatology to produce the gridded quarterly mean field, after bilinearly interpolating the original climatology to the 1 deg by 1 deg grid.
This gridding method was developed by Dr. William S. Kessler, Dr. Michael J. McPhaden, and Dai McClurg, all at PMEL/NOAA.
Michael J. McPhaden