Original User: SaultDon
You could buffer your coastline so that you make a study area like in the image shown.
Then create lines that intersect the coastline where points would lay if they were exactly on the coast line. So your finished result will be something that looks like a railroad track. (the intersecting lines should be at 90deg angles where they meet the coastline centreline)
Attach the point values to their corresponding intersecting line.
Now what you could do is run these lines to create either a TIN or a more gradient resulted option would be to produce a DEM with the coastline buffer as the extent using the z-values from the intersecting lines.
Or you could just use eucl. dist. on each point individually, then sum the cells. If using this method, attention will need to be used as you will have a No Data impact when summing the cells if you do not use the con function to set the null values to an integer >= 0
I would suggest building a model for this and exporting to python for easier execution.