The annual mean vorticity of the ERA and NCEP match very closely. The annual cycle is likewise in good agreement except in the eastern equatorial Pacific and Indian Ocean. Compared to the reanalyses, the models have adequate annual means but suffer in the depiction of the annual cycle in the regions of the jet maxima and in some regions of the Tropics. The CSM appears to inherit errors from the CCM3.
The annual mean divergence evinces a much larger difference between the reanalyses. This is most pronounced in the Tropics, especially over the African and South American land masses. The model simulations also show large differences, with the CSM being an outlier in the tropical Pacific. For many tropical and extratropical locations the annual cycle is not well defined between the NCEP and ERA reanalysis.
The NCEP, ERA, CCM3 and CSM agree with respect to the variance of the monthly mean vorticity. The variance for low pass filtered data is too large in the ENSO regions for the CCM3, but too small for the CSM. Both models tend to underestimate the low frequency variance in mid latitudes.
The ERA has substantially more monthly variance in the divergence than the NCEP data, especially over the tropical South America and Africa and the Dateline. Both models have variance more in line with that of the ERA, and have an anomalous maximum in the eastern Indian Ocean, the CSM much more so. The CSM shifts the maxima in the equatorial Pacific from Dateline seen in the reanalyses to 150E.
The CCM3 appears to be too sensitive to the Equatorial SST anomalies, which may contribute to or exacerbate the poor ocean simulation in the tropical Pacific by the CSM. There are errors in the CCM3 integration which foreshadow, deficiencies in the CSM integration, so the ocean is not solely at fault.
The amount of disagreement between the ERA and NCEP divergence fields on the time scales of the annual cycle and low frequency variations indicates that this field is poorly defined, and in some regions unknown except for sign. (pdf file)