Theodore
Let me look into this - I'll give you a "half answer" right now. We appreciate that for very large datasets, some geoprocessing tools
need improvements for scalability. There is a tool called "LAS Dataset to Tiled Rasters" available here http://links.esri.com/3dSamples that
can take in a large collection of LAS (via LASD) and output raster *tiles* rather than one large raster. You can let the tool define the tiles, or
input your own feature class to define tile boundaries, names, other attributes.
Having said that, I realize you asked about a different GP tool (LAS Points Statistics As Raster) and we don't currently have that
one implemented with a "tiled raster" mode, but I've asked the engineer who built LAS Dataset to Tiled Rasters and it may not be difficult to create something similar, either in Python or with an iterator in ModelBuilder.
Wherever possible, I recommend *against* creating very large single rasters, but instead create more manageable smaller rasters, then manage them as a single mosaic using the Mosaic Dataset. Note If you're seeking to build canopy height, rather than use Z_range in LAS Points Statistics As Raster my recommended workflow would be as follows:
- Run LAS Dataset to Tiled Rasters twice, once filtered for ground points only, 2nd time for First Returns, to
create a bare earth DTM and first return DSM (unless you already have them?) in tiles - Build separate Mosaic Datasets for your two raster collections (DSM and DTM), then create a Referenced Mosaic Dataset based on the
DSM and apply the Arithmetic Function to subtract the DTM. This Referenced Mosaic Dataset will be your canopy height (without taking up any disk space to store it). - If you really need the canopy height as a separate raster, you can export it from the Referenced Mosaic Datase
(This may sound like more work - and I suppose it is - but since you have lidar at different point densities, I would suggest that the ideal processing would be to build DSM and DTM at different resolutions as limited by the point density - specifically, process each lidar collection separately as "projects" and then combine them as recommended here http://esriurl.com/LidarGuidebook Your resulting canopy height map would have variable resolution based on which lidar collection it came from, but the mosaic dataset can easily handle that, and you could then summarize at different resolutions as necessary. )
Last note re: processing speed, I haven’t looked closely at your figures to estimate if there is something wrong or if that’s an appropriate amount of time to expect. Even if you were to iterate through either GP tool to create multiple output tiles, it may not run any faster - the total processing
may still take many hours - but the results should be more maintainable.
Let me know if this is helpful
Cody B.