If the “grd” part is not ideal I think you can use the -E option as a workaround to just input start/stop coordinates. Which would, I guess, fit your purpose

pwessel: sample1d looks like the thing, but I’m immediately bitten by the requirement to have a monotonically increasing or decreasing variable. So the yak shaving starts. Any golden tips on how to resample points that are arbitrarily sampled so I can use it with sample1d?

I think those terms only refers to time series. You are resampling an (lon,lat) line and the (guaranteed) monotonically increasing variable is the distance.

No, you need to use -I to give a sampling interval with a map distance unit, e.g. -I10k, to turn on spatial 2-D series resampling. Otherwise it is just a 1-D time series resampling.

I got really confused since sample1d docs does not mention -I as an option.

Is there an easy way of ‘snapping’ (rounding down/up) the resampled coordinates to the nearest specified increment? E.g. snap to the nearest 15 arcseconds.

Sorry, I get confused with our own deprecation rate at times. The -I was replaced with -T since we use -T for any 1-D array creation (gmt math, makecpt etc).

As for snapping: Probably not, but please give a short example of what you get and what you would like to get.