I have several rasters with varying Z values and thousands of parcels. I’d like to extract the lowest z value from multiple rasters by parcel. Any help is appreciated!
Look there are many ways. I think it would be fair to say that the RasterMosaicker (set to minumum) would be one of the first things to get one raster that can be used.
I went for a point cloud approach, because if you start mentioning 10k or more parcels, I am thinking there are going to be lots of raster cells. FME ability with pointclouds is pretty good, but it can be confusing, if you don’t use it often.
So I have made a example process, in which I am creating points in a 10cm grid (that would be driven by your pixel size) across each parcel. Each of those points holds the id of the parcel. That is going to be a lot of points, but I know that FME can deal with billions of points in a pointcloud quickly. I have assume a unique NUMBER, being parcelID and then passed it through to get the lowest point on each parcel and then merge it back to the parcel.
A key thing is to understand is the movement of an attribute to a component on each point, and then back out to an attribute at the end.
Should also say I have not tested this..especially the pointcloudmerger retaining the first input point, but t do expect the logic that I am trying to show will work.
Thank you,
The rest of my model is in FME so I thought about having it entirely in there.
Something else I should have mentioned, I want the majority lower value; for example, if a parcel is almost entirely in a raster with a value of 125, but a tiny corner is 85, I want 125. With this addition, the point cloud and grid approach concerns me that the output will include values outside the parcel or a value that isn’t realistic of the parcel as a whole.
Hi,
The polygons going into the pointcloudcombiner will only make points inside the polygon, so no issue with points outside the parcel. You would also make them at a grid of 1m since your raster is 1 m.
As for the frequency aspect of the minimum, you are the best placed to come up with the method that works for you. But that still means the process is valid until the raster component transformer, after that you can either try to accomplish is via pointcloud transformers following (that can be daunting), or you can bring the pointcloudcoercer in straight after the elevation is added and create individual points them, and pass it through additional analysis using standard transformers.