I had a point file of an area around 70 km^2 collected through field survey using Auto Level. The survey was carried out in the lines 25 m apart. I converted the readings from auto level into point shapefile and then converted the point shapefile into DTM using Grid- IDW (nearest neighbour selection) tool with search radius of 25 m in QGIS. However, in the DTM prepared through GIS has lot of discontinuity in form of strips at the distance of 25 m. I wish to smoothen these strips. Can you suggest the methods.
If you are using a search radius of only 25m, you have not smoothing at all, as no information from the neighboring cells are used. It even looks that you have not only discontinuities but gaps in the border from one stripe too the other. You should use a large search distance, that points from the neighboring stripes are used in the interpolation
I agree with Dietrich Schroeder, your search radius is too small therefore the software will always default to points along the same lines and not include points collected on adjacent lines. My suggestion increase you search radius to at least 36m and you should see an improvement in your results then smooth if necessary.
The GIS is just a machine, it will let you do anything. Conventionally, a Digital Elevation Model has some expectation of nearly equidistant spacing in the orthogonal directions, but your input data is highly anistropic - so the only somewhat rigorous resulting grid can at best be at half the distance between your strips. There is all sorts of trickery you could do, like expanding the rows of data outward one cell at a time using some neighborhood until they meet in the middle, or use a TIN for interpolation, then convert to a coarser raster. But your interpolation isn't going to be anywhere near the resolution of the points along the data strips. the result is probably better described as a Digital Surface Model of some type, but not a proper DEM.
The actual question is 'fitness for use' - what is your intent for the result? If all you need is some sort of very general relative internal to the area 'there is ridge-ness in the upper corner, a rise in the middle, and the low point is in the other corner' impression, just down sample the points using a radius slightly larger than half the gaps.
Karan Gupta - you can try to generate a TIN model from the point shape file data. TIN model is always better than raster based models of DEM. I hope you will not have the stripping effect.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India