I am working with ArcGIS. I have several vectorial data of a territory such as Geology, Land Use, Landslide and flood susceptibility map (polygon data), InSAR data, Earthquakes (Point data), river and road network (line data) and raster data such as DTM, Slope, Aspect.
In order to detect and then predict critical road sections on my studied road network I have to assign all these feature to the road network itself. The aim is to build a database that I can use to develop some predictive models (such as Support Vector Machines or other Machine Learning Algorithm).
Dividing the road network in several fixed-length sections, and then assign the features, does not work, because I can not take into account the spatial heterogeneity of the feature inside the homogeneous sections (that are not "so homogeneous").
I am thinking on sectioning the road network each time a certain feature changes its value (dynamic segmentation), or to build a cell-based database (so convert all the data into raster, and then overlay these resulting raster file).
Is these a good ideas? Have you got other ideas or suggestion?
Thanks,
Nicholas.
I wonder if you have tried an Intersect operation of the road file with other polynomial files one by one. The advantage of an Intersect operation is that the output will inherit all the attributes from both files. See https://pro.arcgis.com/en/pro-app/tool-reference/analysis/intersect.htm
Dear Srini,
Thank you for your kind answer.
The "Intersect" operation was one of the first I tried. And it was good enough. However, I think that the "Identity" operation works better since the output maintains not only the intersecting features but also the input features that don't have any attribute in common with the overlay layer.
You should find something here: http://desktop.arcgis.com/en/arcmap/10.3/analyze/commonly-used-tools/overlay-analysis.htm
Anyway, all of these is good enough for categorial polygon data. For raster such as Slope or Aspect in which we have continuous values, the "Identity" operation (or the "Intersect") doesn't work. I should convert continuous raster features into vectorial categorical ones, but it would require several assumptions in order to split the values in some interval.
In order to preserve the spatial heterogeneity, what would you do to assign these continuous features?
Thanks,
Nicholas.
Dear Taha Sadeghi Chorsi ,
Thank you for your useful answer.
How did you choose the cell size (considering that different data have different accuracy)?
Best,
Nicholas.
hello nicols im so sorry for the delay but i was so busy, i didn't saw your email.
About your question, tge way to solve is relative to the goal from it as you know each method help in a way to make a different analysis for the data, so yes your idea is good but i will send you more method, just i need time to make a pdf in that and send it to you
- Badges
- Science topic
- Similar topics
- Social Science
- Sociology