I would like to take this archaeological drawing and use QGIS to create a something similar to a DEM or TIN, so that depth/elevation values can be quickly and clearly understood.

I do not know what would be an appropriate workflow to accomplish this. Although lines do represent different depths, they are not regularly spaced like a topographic map. Additionally, there are some areas of the drawing which indicate different depths within the same feature (in the southern wall trench depths of 34, 29, 20, and 23cm are shown together).

At this point, georeferencing is not required, but the ability to do so in the future is one of the reasons I would prefer to use QGIS over other software alternatives. What recommendations do you have for converting this drawing into a DEM or TIN?

Here's an example of the type of drawing I'm working from.

archaeological drawing

  • If I understand correctly you would like to make a dtm out of vectors for the tips on how to do it see gis.stackexchange.com/questions/18116/… To add the elevation data to the attribute table just add a columne name i.e. Z and this is where you will manually add it in while digitizing the plan. – m_kos Nov 16 '19 at 10:31
  • There seems to be a lot of different takes on the linked post. I take it "Contour lines to DEM" is what you are suggesting? Any thoughts on what to do with wall trenches with multiple depths listed within the same feature? Seems points would provide flexibility, but lines would provide clear definition for the different features. – Corey Nov 16 '19 at 10:42
  • Any time you have an "and" in your question title you're at risk of a too-broad closure. In the GIS model (and unlike CAD), symbolization and slope are not germane to digitizing. In many respects, even attributes like elevation can be captured in a second pass during geometry QC. Then once you have all your data, you can symbolize and perform analysis. Note that what could be your hardest task, georeferencing the image (and potentially rubber-sheeting the geometries) is unmentioned. – Vince Nov 16 '19 at 11:52
  • 1
    It is an interesting task. It has too many challenges for a single question and I think you should divide it into specific questions, one at a time. Probably the least trivial is to digitize a LineString with different elevations for each vertex, for which I think it is convenient to digitize it at constant elevation, try to create an interpolation between the points of known elevation, and assign elevation to the vertices of the LineString based on that interpolation. However, the question should be reformulated. – Gabriel De Luca Nov 17 '19 at 17:43
  • 1
    Are you sure the lines are isohypses (ie lines of equal elevation)? The elevation values seem to contradict that idea. As far as I can tell then only elevation information in this image is the numbers, which you could digitize as points and interpolate into a DEM. – csk Nov 18 '19 at 19:36

According to my interpretation of the drawing, the lines are what are called structural lines, in the terminology of an interpolation.

  • Structural lines are vertical jumps in the terrain.

They are not necessarily constant level lines.
In manual interpolation, the structural lines should be considered in pairs, a line that represents the upper levels of the structure and a line that represents the lower levels of the structure.

The contour lines adhere to the lower structural line as they reach the levels it represents. They are grouped on top of each other while the structure is vertical, and they begin to separate from the upper line as they reach the levels that the upper line represents.

  • I think you should start digitizing these lines as simple closed LineString features. It is not necessary that you immediately worry about its level, since it seems more appropriate to define it later.

  • And digitize the known level points as point features, probably with their Z value included in the geometry, or in a separate field of the attribute table.

I am particularly interested in understanding in the best possible way how the restriction of structural lines is applied in the interpolation algorithms of QGIS.

If you could digitize some representative lines and points of known level, for at least some sectors, and share those layers in a geopackage file, avoiding us having to interpret the meaning of the different numbers and nomenclatures that are currently present in the drawing, I would be delighted to strive to investigate the application of structural lines to QGIS interpolation algorithms, try to build a DEM that represents the topographic condition of that terrain in the best way I could, and share the procedure used.

  • I started to digitize a few lines & I realized what you meant in your comment about different LineString elevations. Using P1 in the upper portion of the drawing as an example, there are several overlapping holes present. While the inner rings should have constant elevations, 82cm, 97cm, & 49cm, the outer rings would have variable elevations depending on where they contacted the other holes. Looking at the 97cm hole, the outer ring would be floor level on the south but where the it touches the 49cm hole, that section would have to be set to an elevation of 49cm. Seems a bit nightmarish... – Corey Nov 19 '19 at 8:03
  • Don't worry about the elevation of the LineStrings. Just digitize them in a topological way: two intersected LineString must share a vertice. We will deal with their elevations later. What are the numbers? P1, P17, P19 are sectors? 49, 82, 97 are depths in centimeters? 90.45, 90.54 are elevations in meters? – Gabriel De Luca Nov 19 '19 at 15:48

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.