CO2
                                  ---

The model used a constant CO2 concentration of 330 ppm for all the 
assimilations.  Of course observed CO2 concentration has an annual
cycle as well as an anthropogenic trend.  In addition, there is also 
spatial variability.  However, one cannot use doubled CO2 simulations 
to estimate a potential error in the reanalysis.

    Reasons why a fixed CO2 will have a minor impact on the analyses

(1) Reanalysis uses observed snow cover, sea ice and SST.  Simulations
    with doubled CO2 with fixed SST and snow cover would show a much
    smaller temperature change.  

(2) Observed temperatures are used in making the analyses.

(3) Observed humidities are assimilated.  (Another factor in CO2's effect 
    is the modification of the humidity.)

Beside CO2, Reanalysis used climatological values for ozone, vegatative
parameters, solar insolation and orography.