What is the best way to calculate anomaly of meteorological variables with the (GFS/GEFS) forecast

To calculate the anomaly of a variable (temperature, for example) I need the climatology (at least 30 years) of that variable. Which climatology data source should I use? Should I use the "climatology" of the weather forecasts? In this case, does the fact that the versions of the models have changed, and each model has a bias, interfere with the quality of the result? I have already read that it is not correct to use CFSR data to calculate the anomaly since 2011. I would like someone with experience in the area to explain the best way to calculate the anomaly for (GFS/GFS) forecasts. And where it is possible to obtain the climatological data for this calculation.
Thank you very much in advance.