To clarify, what I'm looking for is exactly the inverse operation to "gridding" as defined here:<div><br></div><div><a href="http://www.gdal.org/grid_tutorial.html">http://www.gdal.org/grid_tutorial.html</a></div>
<div><br></div><div>This is much simpler and it's not difficult to write the low-level code myself, but it seems quite strange to not have a high-level interface. Is there really no interface, or am I just inadequate at reading the docs?<br>
<br><div class="gmail_quote">On Tue, Apr 26, 2011 at 01:24, Jed Brown <span dir="ltr"><<a href="mailto:jed@59a2.org">jed@59a2.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
I have a bunch of unstructured points, typically quadrature nodes of an unstructured mesh, and I want to sample a raster field at those points. I have plenty of spatial locality, but not any specific regular structure. Maybe I'm missing something, but I don't see an API for this direction.<div>
<br></div><div>There is GDALCreateTPSTransformer which would seem to be a reasonable way to go from an unstructured mesh to a raster. But it's not easy to go the other way since it would involve, e.g. defining a (bad) quadrature using the raster and then do a Galerkin projection into the unstructured finite element basis, then evaluate that basis at the unstructured quadrature nodes. Alternatively, we could make tiny regular patches in the unstructured grid and do lots of independent projections, but I worry that this would be a performance issue.</div>
<div><br></div><div>Is there a recommended way to sample a raster field at unstructured points?</div>
</blockquote></div><br></div>