The Problem with PAOBS

PAOBS are "data" used by the current NCEP and ECMWF analyses/forecasts. Unlike conventional data, they are the product of human analysts who estimate sea-level-pressure (in our case) based on satellite data, conventional data, and time continuity. The Australians generated the PAOBS for the data-poor Southern ocean and we obtained the 1979-1992 digitized values from the ECMWF.

There was a problem with the reading of the PAOBS data. Details are given in a letter by E. Kalnay (head of EMC) to A. Oort (head of the Reanalysis panel).

The figures for the letter are given below.

Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9a , 9b
Figure 10

Our preliminary assessment is that this error only affects the results of the Reanalysis South of about 40-60S, and there it only doubles the intrinsic uncertainty of the analysis.

The Reanalysis Advisory Committee has been asked by the chair, Abraham Oort, to come with a recommendation with respect to options a and b in the letter (redo the 14 years or proceed with 40 years reanalysis and correct this and make other improvements in the Second Reanalysis)

We understand that people will need to satisfy their curiosity whether the mislocation of the PAOBS affects their results. We plan to perform a one year reanalysis with the correct PAOBS (like the "NEW" experiment discussed in the letter), and a "NULL" experiment to assess the intrinsic uncertainty of the Reanalysis. This data will be put online to allow researchers to make their own assessment, if desired. Of course, the NULL experiment is useful for other purposes as well. (EK)

The Climate Prediction Center has performed some additional diagnostic studies of the experiments discussed above.
Comments? Please send them to E. Kalnay (

0 back

Privacy Policy