The generation of each TMY file has a unique random process in it to select a series of 4 day chunks (see What is the “TMY” Data Type and how is it made?). Therefore, the actual days used to make a “typical year” will be different each time a TMY file is created. In our experience this manifests itself most when a customer orders an hourly TMY file from us, and then also soon after orders a TMY file at a different temporal resolution. The customer expectation would be that the values in the hourly file would be the hourly averages of the values in the 10 minutely file. The randomization process prevents that from happening. If consistency is required – then the 10 minutely TMY file should be used as the reference and each of columns averaged to create an hourly TMY file