Open MichaelJendryke opened 3 years ago
DGGRID finds the cell (i, j)
bounds of the clipping polygon on each quadrilateral it intersects, and then iterates over those cells. This approach was chosen because it works for all grid topologies (in DGGRID triangles, diamonds and hexagons all use Q2DI
coordinates), and it generates/outputs the intersected cells in sequence number order.
But as you say, this often leads to a low acceptance rate.
Now that version 7 can find spatial hierarchy children the obvious approach is to cull the cells hierarchically. That is on my todo list. I did look at it a bit, and it looks easy for coarse resolutions, but gets trickier for very high resolutions where it is most useful.
I am generating cells for a specific area by using a Shapefile for clipping.
The polygon used for clipping is actually a lower resolution cell, e.g. resolution 12. Doing so will generate cells of resolution 24 that intersect with the single - resolution 12 - cell. However I noticed, that
DGGIRD
generates more cells and tests those cells against the clipping polygon. Often the number of cells that are accepted (those that intersect) is very low, decreasing the acceptance rate to less than 10%. How doesDGGRID
determine the area for which cells are generated for intersection? Is there a way to improve the acceptance rate?