[ferret_users] regresst and variance

[David Wang <climater@xxxxxxxxx>]

Dear Ferreters,

I'd like to detrend two time series before calculating the correlation between them (actually two two-dimensional time series, say var1 (time,lat,lon) and var2(time,lat,lon)).

I thought of 
regresst.jnl and variance.jnl. But both go files require two user-defined variables of the same names, P and Q. As I understand it, FERRET won't evaluate an _expression_ until a plotting or a saving command is issued, and it does not have the concept of internal variables. So P and Q defined theretofore would be redefined thereafter (as well as all the expressions that contain P or Q). To save detrended variables as temporary netcdf files would be a trick, but not something that I particularly like. If there are any other solutions, I'd like to know them.

Thanks,
David