Tension parameter in surface

[JackieC-A]

I’m wondering if someone can help me understand the tension parameter in the surface function. I’ve read the documentation but am struggling to understand what value is best suited to my data. For context, I’m trying to plot a cross section of seismic tomography data, pre-processed with blockmedian. I’m surprised by how strong the effect of the surface function is on the output, though–the velocities are significantly different depending on how much tension I use, and it isn’t clear that they anchor well to the existing nodes.

Does anyone have suggestions around best practices here? Obviously I want a figure that best represents the actual data values, and it seems like I’m not really understanding how surface works with the data. Has anyone done something similar?

[Esteban82]

@JackieC-A

isn’t clear that they anchor well to the existing nodes.

I am not an expert in gmt surface but I think surface is anchored exactly to the existing nodes. If not, it may be because the original data was modified by preprocessing with block-median.

[timhume]

The original paper on surface is here:
https://www.researchgate.net/publication/215755455_Gridding_with_Continuous_Curvature_Splines_in_Tension

[mkononets]

In addition to the original paper:

• GMT tutorial Session 3 especially explanations in Gridding with Splines in Tension (as well as everything before and after).
• Example 14 Gridding of data and trend surfaces (personally I like very much a previous version with grid cell boundaries plotted: GMT 5.4 Example 14, Gridding of data and trend surfaces)
• Example 16 Gridding of data, continued especially the explanatory text, but also the two examples of surface output with tension levels of 0 and 0.5 even if the color scale is quite far from ideal IMO.