While I am training my dumb model to not be so dumb, I was going to compare some temperatures to verify parts. I have major issues with the temperature data sets thanks to my computer issues, so I don't do that very often.
One of the neat things about the model is that it takes advantage of all the temperature data available and there is lots of that. Knowing the approximate relationships you can fine tune base or reference values to determine flux anomalies based on the temperature anomalies. Pretty straight forward.
One of the things on my Christmas wish list would be areal temperatures for specific locations in regions. You can only go so far with global and regional relationships. Urban Heat Island (UHI) is one of the best ways to fine tune flux changes because of the much great nocturnal temperature ranges. Since the CO2 induced radiation is a factor of 3 or more time the temperature change, large cities are a perfect test stand for flux calibration. Also since these larger areas have plenty of weather stations, the is much more data for normally a much longer time period. Oddly, I have not seen any studies published that compared UHI surface, lower troposphere, mid-troposphere and lower stratosphere data in any productive manner. Only the constant bickering about how the data sets should be adjusted.
That is the problem with adjustments to data. The raw data is full of information ready to be teased out. So if any of you time series geeks pop by, think about doing something productive for a change. Do me a Bucky surface in layers above a big humid city and a big dry city so we can calibrate the model for those area and see what happens.