Rasterize centroids by default #84
Conversation
|
Not sure if this is a good idea, it might cause confusion and unexpected results: if several centroids fall into the same cell, the last one that is rasterized wins which is arbitrary. v.to.rast type=centroid,area is working as expected. Rather adjust scripts like v.rast.stats to also rasterize centroids? |
I guess you are right. BTW, does the order (type=centroid,area vs. type=area,centroid) have any effect?
OK, will have a look at it... |
No, areas are always rasterized first. After that, points, lines, boundaries, centroids are rasterized in the order in which they are read from the vector map. |
I don't understand what is the expected output with type="centroid,area". Isn't the centroid inside the area anyway? |
Yes, the centroid is inside the area. If the center point of a cell is not inside the area, that cell is not assigned to that area. However, if the centroid of that area is inside that cell, that cell gets assigned to this centroid. IIUC, this is what this PR is about. |
Right. It makes a difference only (?) in case when a cell center is not inside area but centroid happens to be in that cell. In that case that cell is rasterized. Centroids and boundaries in GRASS GIS is what is constructing areas, so unless the centroid has some special meaning in a specific use case, it is just something hidden. That's why I don't understand why would a user want the behavior as default? Since @ninsbl is suggesting it, I'm guessing there is a case where the behavior is useful, but it does not seem to me it should be the default behavior. Is having centroids in this context an expected default for most users? If centroids, why not boundaries? |
|
Sorry for not being clear i my answer to Markus. My plan is to look at scripts like v.rast.stats to also rasterize centroids (or at least add the option if it is not present already). |
Please do. It will be more clear. PR is meant to be merged or closed. |
|
OK, done. |
Not sure if this should be left for GRASS 8 as it might alter results of existing user scripts, but it would be good to have centroids rasterized e.g. also in v.rast.stats by default. And if areas are rasterized by default, it would make sense to make centroids the default too...