2

I am attempting to interpolate the elevations of data sets of cross-shore survey transects. The data set size range from 50k to 300k points with ~100m between the transects and 1-5m between the points within a transect. Find an image here. Apologies, I'm still working on quickly drafting figures in GRASS.

I interpolated the data to a 2m resolution raster with v.surf.rst. I used the default parameters and set a mask based on the data's concave hull. During the computations I received the following messages:

Ignoring xxx points (too dense)

and

WARNING: Taking too long to find points for interpolation - please change the region to area where your points are. Continuing calculations...

The RMSD which GRASS output with the rasters averaged about 0.35m absolute and 3% of the range. Can anyone tell me if these errors are typical?

I also tried to set scalex and theta (other params were default), but during the computation I received this warning:

WARNING: Overshoot - increase in tension suggested. Overshoot occurs at (1340,1023) cell. Z-value -779923520635822222878965760.000000, zmin -8.433820, zmax 3.060190.

And one of the cells had that large negative value.

I don't really know how to tune the parameters. I'll probably script an optimization using the -c flag, if it seems warranted after a quick lit review. Any other suggestions on changing the default params?

2

I suggest to watch this lecture for better understanding of spline interpolation and its implementation in v.surf.rst: https://www.youtube.com/watch?v=eLYsx-ADOGI

Generally, overshoots happen when you have very close points with very different z values. You can increase tension and increase smoothing to avoid them, but they can also come from errors in the data.

Using scalex and theta is probably a good idea with your point configuration.

Ignoring xxx points means that xxx points were skipped as being to close to each other, you can use dmin option to change default behavior, which is I believe half of current region's cell size.

Outputting either slope or curvature (pcurvature) raster gives you good idea about any discontinuities in the interpolated raster.

1
  • That lecture was pretty good. I think that I need to deal with the anisotropy in the data before any tunning. The data presented at ~16:50 is basically what I'm dealing with and she even calls it "beach" data. However, its unclear how to estimate scalex and theta based on that lecture. I think I want to increase tension in the cross-shore direction (along the survey transects, because these points are naturally related) and decrease tension in the alongshore direction. So, following the GRASS docs for v.surf.rst, I should consider the survey transect the feature, not the beach. Correct? – CoastalBoy Jun 27 '17 at 21:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.