Groups keyboard shortcuts have been updated
Dismiss
See shortcuts
Advice for working with stream network delineation with large DEM
49 views
Skip to first unread message
Sandy Elliott
Jan 13, 2025, 7:08:44 AMJan 13
to WhiteboxTools
Hi,
I have successfully delineated a stream network with a 1m DEM in Whitebox Workflows for a 17 km2 catchment. Now I am wondering what I need to do to scale this up to a much larger catchment. Can WbW handle this? (when the data would not fit into RAM on my PC). I do have access to HPC, but I suspect that at some point the catchment might need to be broken up into chunks - the trouble being that tiling would not make much sense because streams could run across tiles. Does anyone have advice on a strategy for this?
Jean-François Bourdon
Mar 4, 2025, 1:03:48 PM (13 days ago) Mar 4
to WhiteboxTools
Hi Sandy,
I'm doing exactly this for a territory of over 500 000 km², no HPC needed. I use the Isobasins tool on a very large and arbitrary territory (a 5000 km² rectangle/tile for example) at a corser resolution (let say 5 m DEM resolution). So if I know that I have enough RAM (WbW or WbOC, doesn't matter) to safely process a 200 km² catchment at 1 m resolution, I'll set the size parameter of Isobasin to this value (as number of grid cells so if raster resolution is 5 m, must be 200 * 1000^2 / 5^2). Doing this will generate several catchment of about 200 km² and many more of smaller size. I ditch all those smaller catchments and only reprocess at a 1 m resolution the bigest catchments (plus a small 1km buffer around). On the next 5000 km² tile, I make sure to cover the smaller contiguous catchements that I previously ditched. Of course, the DEMs must be hydrologically corrected at each step using breaching and/or filling.
I must note that this not a perfect workflow. Because each catchment is processed independently of its neighboor, this mean that different and conflicting breaching/filling solutions can be found for the same cell position. This leads to small gaps and overlaps between catchments at the finer resolution. If your goal is simply to extract the bigger vector streams, it might not be an issue but if you want to merge together the individual hydrologically corrected DEMs, you will run into issues not easy to solve with only the current WBT tools.
I'm doing exactly this for a territory of over 500 000 km², no HPC needed. I use the Isobasins tool on a very large and arbitrary territory (a 5000 km² rectangle/tile for example) at a corser resolution (let say 5 m DEM resolution). So if I know that I have enough RAM (WbW or WbOC, doesn't matter) to safely process a 200 km² catchment at 1 m resolution, I'll set the size parameter of Isobasin to this value (as number of grid cells so if raster resolution is 5 m, must be 200 * 1000^2 / 5^2). Doing this will generate several catchment of about 200 km² and many more of smaller size. I ditch all those smaller catchments and only reprocess at a 1 m resolution the bigest catchments (plus a small 1km buffer around). On the next 5000 km² tile, I make sure to cover the smaller contiguous catchements that I previously ditched. Of course, the DEMs must be hydrologically corrected at each step using breaching and/or filling.
I must note that this not a perfect workflow. Because each catchment is processed independently of its neighboor, this mean that different and conflicting breaching/filling solutions can be found for the same cell position. This leads to small gaps and overlaps between catchments at the finer resolution. If your goal is simply to extract the bigger vector streams, it might not be an issue but if you want to merge together the individual hydrologically corrected DEMs, you will run into issues not easy to solve with only the current WBT tools.
Reply all
Reply to author
Forward
0 new messages