I processed a lot of modeled climate data from CMIP and I want to check that I have not made any errors. I think the best way to do this is to somehow evaluate the data I produced geographically against other observational datasets.
In my processing I create a climatology of tasmax for the modeled historical data (1990-2010) and do some other 'tinkering'.
How can I check that my final tasmax 'product' is still realistic (i.e. how well does it match up to observations?)? Should I directly difference it from the observations to see how far my climatology differs? Are there any other methods that will highlight spatial anomalies in my new data set, if they exist?