Variogram calculation should be more interactive

[ET-NCMP]

The variogram step is the step during which things are most likely to go wrong. At the moment, the calculation of the empirical variogram and the fitting of the variogram are contained in one script.

The calculation of the empirical variogram can be slow for moderately large numbers of stations, which is annoying for the user if they have to run the script multiple times to get a good set of variograms. Ideally, the calculation of the empirical variogam would be a stand alone script that gets run once. The output would then get passed to a script that fits different functional forms to the empirical variogram.

This second script would read in the empirical variogram and ideally:

• allow the user to specify the maximum station separation for the fit and to do this separately for each month.
• allow the user to fix certain parameters if necessary or choose default parameters if the empirical variogram lacks obvious structure.
• run the program repeatedly, improving the estimates with each iteration.
• choose to eliminate outliers in the empirical variogram to improve the fit or employ a robust fitting algorithm that uses something other than least-squares estimation.
• once the user is happy with the fits, the script should terminate and output the desired parameters.

[ET-NCMP]

I've added a basic interactive script for the variogram calculation - P3_Variogram_Interactive.R. This allows the user to specify a new maximum range and recompute the best fit parameters for the variogram as often as they like.

[ET-NCMP]

This has now been iterated a few times and run by team member. It is part of the system and included in the user manual.