# What is the best method to smooth TIN Surfaces?

I have a 3D TIN which shows an avocado sizes in Z-values. The 3D visualization is perfect, however, is not our needs because it requires too much interpretation from our end users perspective.

How can we create continuous surfaces with the same output, while we want to avoid using the interpolation toolbox?

The image below is the TIN surface clipped to a farm boundary. The lighter green is indicative of size 48 fruits, which are the largest. Ideally, I would like to create a continuous raster with similar output, such that interpolating the surface will not create an equal distribution across the farm.

• By definition, having a continuous surface is going to require interpolation. Right now, you have points at specific locations. In order to create a surface that is continuous, you either have to add more known points to fill in the gaps, which would still not make it continuous, or calculate those values in between your known locations. This is done through interpolation. Having said that, we really need some more detail about the project and visualization you are trying to accomplish. Please be as specific as possible. – Get Spatial Nov 5 '13 at 21:11

## 2 Answers

Why not try Interpolation rather than a TIN?

If you use a Inverse Distance Weighted algorithm, you can play with the parameters to manage smoothness.

If you have the license then run Kriging. Again, play with parameters until it looks nice.

To break your question down a bit, do you need to interpolate? Do avocado sizes vary in a way that displays spatial auto-correlation? Should you be plotting them as circles of varying sizes and showing other attributes underneath, like min/max/average temperature, slope, aspect, average insolation. If you're just visualising avocado sizes, maybe plotting avocado pictures of different sizes would work better (though it does sound a little absurd!)

• We used the krig function initially, however, the semi-variogram indicates an equal distribution across space. And this is not the case. The goal is to create a thematic map which shows larger fruits, medium fruits, small fruits, no fruits, etc. The TIN captures the actual density of fruit. – Geoffrey West Nov 6 '13 at 0:20
• Perhaps you could add an image, @GeoffreyWest? – Alex Leith Nov 6 '13 at 21:12

If you don't want tu use interpolation, you can convert your TIN to a raster file (TIN to Raster), then use a mean filter (focal statitistics) or a low pass filter (SA> Neighborhood > Filter with option "LOW")