Idea: Remove indexing from the extract routine into a separate function. Test this against an equivalent using interpolate2d that holds a list of functions ... this may remove the ad-hoc nature of tuning the decorrelation distance of the gaussian smoothing.
The indexing and gaussian weighted smoothing was implemented for speed more than anything else - but now in python it looks a little easier to implement the in-house functions efficiently.
Improvements can also be made in working on individual boundaries (e.g. if there's a northern and southern boundary that are disconnected) rather than reading in the data that covers both (also could add an option to output individual BDY files (e.g. BDY_north and BDY_south)
Idea: Remove indexing from the extract routine into a separate function. Test this against an equivalent using interpolate2d that holds a list of functions ... this may remove the ad-hoc nature of tuning the decorrelation distance of the gaussian smoothing. The indexing and gaussian weighted smoothing was implemented for speed more than anything else - but now in python it looks a little easier to implement the in-house functions efficiently. Improvements can also be made in working on individual boundaries (e.g. if there's a northern and southern boundary that are disconnected) rather than reading in the data that covers both (also could add an option to output individual BDY files (e.g. BDY_north and BDY_south)