I'm trying to assess and remove seasonality from yearly climate data. I don't have a stats background, just fyi. My understanding is that I can use an autocorrelation plot to determine the proper lag to use with the differencing method of removing seasonality. Is this true? And how do you interpret the autocorrelation plots to determine the lag to use with differencing? Bellow are some example outputs of the raw data (with lowess in red and difference between actual and lowess in green) and the autocorrelation plots.
Also, on all of my plots (I've got 8 different points/locations, 4 different GCMs (global climate models) and 8 different climate variables) the confidence intervals seem fixed.. but this should be dependent on the data... and I can't find any python documentation (I'm using the autocorrelation_plot() for Pandas in Python). Is the command normalizing the data in some way? Thank you in advance.