I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?
I am trying to extract river network from a 5m DEM with some success using
r.watershed. Has anyone tested this algorithm on high resolution LiDAR
data
for example - 1meter DTM and what kind of results have they obtained?
in the NC sample dateset there is lidar derived 1m x 1m DEM:
r.info map=elev_lid792_1m@PERMANENT
+----------------------------------------------------------------------------+
| Map: elev_lid792_1m@PERMANENT Date: Wed Feb 14 19:45:24 2007
|
| Mapset: PERMANENT Login of Creator: helena
|
| Location: nc_spm_08_grass7
|
| DataBase: D:\grassdata
|
| Title: Rural area: Lidar-based 1m DEM
|
| Timestamp: none
|
I’m happy to report that I’ve modeled overland water flow with r.watershed for over a quarter million acres, consisting of several large project sites, at 1 meter DEM resolution. The data source was LiDAR points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other artificial flow control features, to achieve water flowing through a road. Otherwise, water is routed along roads until a lowest point is reached for crossing.
I also use r.terraflow outputs as ancillary data to help drop in culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of topographic depressions known as wetlands. These depressions interrupt the stream network continuity in reality, but r.watershed does a fantastic job making a continuous drainage network model, especially in these difficult areas.
I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?
Thanks for your reply! Sounds great.
How did you add culverts or other artificial flow control features to achieve water flowing through roads?
I have a rivers layer and I compared it the streams I’ve obtained from r.watershed and r.watershed appears to not match these streams (which were accurately digitised) and I was wondering if I had a better resolution DTM would it solve this problem?
Also, why is sink filling needed for terraflow and not watershed?
Thanks for your reply! Sounds great.
How did you add culverts or other artificial flow control features to achieve water flowing through roads?
I have a rivers layer and I compared it the streams I’ve obtained from r.watershed and r.watershed appears to not match these streams (which were accurately digitised) and I was wondering if I had a better resolution DTM would it solve this problem?
Also, why is sink filling needed for terraflow and not watershed?
Thanks for your help
On Máirt 5 Noll 2017 at 23:43, Mark Seibel <mseibel@gmail.com> wrote:
Hi Shane.
I’m happy to report that I’ve modeled overland water flow with r.watershed for over a quarter million acres, consisting of several large project sites, at 1 meter DEM resolution. The data source was LiDAR points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other artificial flow control features, to achieve water flowing through a road. Otherwise, water is routed along roads until a lowest point is reached for crossing.
I also use r.terraflow outputs as ancillary data to help drop in culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of topographic depressions known as wetlands. These depressions interrupt the stream network continuity in reality, but r.watershed does a fantastic job making a continuous drainage network model, especially in these difficult areas.
I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?
I just used the tutorials from this page https://grasswiki.osgeo.org/wiki/Creating_watersheds to extract the stream network - however it could be more accurate so was wondering is r.terraflow a better option for me. I was using r.watershed originally
Do you know why sink filling is needed for r.terraflow and not for r.watershed?
Thanks for your reply! Sounds great.
How did you add culverts or other artificial flow control features to achieve water flowing through roads?
I have a rivers layer and I compared it the streams I’ve obtained from r.watershed and r.watershed appears to not match these streams (which were accurately digitised) and I was wondering if I had a better resolution DTM would it solve this problem?
Also, why is sink filling needed for terraflow and not watershed?
Thanks for your help
On Máirt 5 Noll 2017 at 23:43, Mark Seibel <mseibel@gmail.com> wrote:
Hi Shane.
I’m happy to report that I’ve modeled overland water flow with r.watershed for over a quarter million acres, consisting of several large project sites, at 1 meter DEM resolution. The data source was LiDAR points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other artificial flow control features, to achieve water flowing through a road. Otherwise, water is routed along roads until a lowest point is reached for crossing.
I also use r.terraflow outputs as ancillary data to help drop in culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of topographic depressions known as wetlands. These depressions interrupt the stream network continuity in reality, but r.watershed does a fantastic job making a continuous drainage network model, especially in these difficult areas.
I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?
How did you add culverts or other artificial flow control features to achieve water flowing through roads?
Culverts can be added by digitizing the connection points (line) across the road from high accumulation to high accumulation on the other side of the road. It can be a long iterative process, that can be encapsulated in a script. I’ve wondered if machine learning could be used to automate what seems to be such repetitive and easy culvert connections.
Also, why is sink filling needed for terraflow and not watershed?
I am not into the code, but I know r.watershed use different code to produce different model results. r.watersheed seeks the lowest points in the terrain, thus internal sinks get modeled with exit points rather than internally drained. On the other hand, r.terraflow hydrologically fills the terrain, then models flow accumulation. The results are similar, but the differences can be used to add another check on reality from another model; especially in really flat, depressional terrains.
Mark
Thanks for your help
On Máirt 5 Noll 2017 at 23:43, Mark Seibel <mseibel@gmail.com> wrote:
Hi Shane.
I’m happy to report that I’ve modeled overland water flow with r.watershed for over a quarter million acres, consisting of several large project sites, at 1 meter DEM resolution. The data source was LiDAR points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other artificial flow control features, to achieve water flowing through a road. Otherwise, water is routed along roads until a lowest point is reached for crossing.
I also use r.terraflow outputs as ancillary data to help drop in culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of topographic depressions known as wetlands. These depressions interrupt the stream network continuity in reality, but r.watershed does a fantastic job making a continuous drainage network model, especially in these difficult areas.
I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?
wiki/Creating_watersheds to extract the stream network - however it could
be more accurate so was wondering is r.terraflow a better option for me. I
was using r.watershed originally
I've found r.watershed to be the most accurate, for our work. We've field
verified the streams and its really solid data. I like that r.watershed
doesnt alter the terrain since the LiDAR point data captures terrain
nuances so well.
Mark
Do you know why sink filling is needed for r.terraflow and not for
r.watershed?
Thanks.
On Wed, Dec 6, 2017 at 11:42 AM, Marco Alicera <marco.alicera@gmail.com>
wrote:
Thanks for your reply! Sounds great.
How did you add culverts or other artificial flow control features to
achieve water flowing through roads?
I have a rivers layer and I compared it the streams I've obtained from
r.watershed and r.watershed appears to not match these streams (which were
accurately digitised) and I was wondering if I had a better resolution DTM
would it solve this problem?
Also, why is sink filling needed for terraflow and not watershed?
Thanks for your help
On Máirt 5 Noll 2017 at 23:43, Mark Seibel <mseibel@gmail.com> wrote:
Hi Shane.
I'm happy to report that I've modeled overland water flow with
r.watershed for over a quarter million acres, consisting of several large
project sites, at 1 meter DEM resolution. The data source was LiDAR
points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other
artificial flow control features, to achieve water flowing through a road.
Otherwise, water is routed along roads until a lowest point is reached for
crossing.
I also use r.terraflow outputs as ancillary data to help drop in
culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of
topographic depressions known as wetlands. These depressions interrupt the
stream network continuity in reality, but r.watershed does a fantastic job
making a continuous drainage network model, especially in these difficult
areas.
Happy Modeling!
Mark
On Tue, Dec 5, 2017, 3:49 PM Shane Carey <careyshan@gmail.com> wrote:
Hi,
I am trying to extract river network from a 5m DEM with some success
using r.watershed. Has anyone tested this algorithm on high resolution
LiDAR data for example - 1meter DTM and what kind of results have they
obtained?
Thanks
--
Le gach dea ghui,
*Shane Carey*
*GIS and Data Solutions Consultant*
Yep, I like that also - for me though - you can see the big rivers in the hillshade but it is clearly not picking them out! Any ideas why this is? Is the threshold value very important do you know?
Thanks
···
On Wed, Dec 6, 2017 at 1:11 PM, Mark Seibel <mseibel@gmail.com> wrote:
–
Hi.
I just used the tutorials from this page https://grasswiki.osgeo.org/wiki/Creating_watersheds to extract the stream network - however it could be more accurate so was wondering is r.terraflow a better option for me. I was using r.watershed originally
I’ve found r.watershed to be the most accurate, for our work. We’ve field verified the streams and its really solid data. I like that r.watershed doesnt alter the terrain since the LiDAR point data captures terrain nuances so well.
Mark
Do you know why sink filling is needed for r.terraflow and not for r.watershed?
Thanks for your reply! Sounds great.
How did you add culverts or other artificial flow control features to achieve water flowing through roads?
I have a rivers layer and I compared it the streams I’ve obtained from r.watershed and r.watershed appears to not match these streams (which were accurately digitised) and I was wondering if I had a better resolution DTM would it solve this problem?
Also, why is sink filling needed for terraflow and not watershed?
Thanks for your help
On Máirt 5 Noll 2017 at 23:43, Mark Seibel <mseibel@gmail.com> wrote:
Hi Shane.
I’m happy to report that I’ve modeled overland water flow with r.watershed for over a quarter million acres, consisting of several large project sites, at 1 meter DEM resolution. The data source was LiDAR points to make the DEMs.
At this resolution, it becomes necessary to add culverts, or other artificial flow control features, to achieve water flowing through a road. Otherwise, water is routed along roads until a lowest point is reached for crossing.
I also use r.terraflow outputs as ancillary data to help drop in culvert locations and help provide guidance in problem areas.
My geographic area is central Florida, which is very flat and full of topographic depressions known as wetlands. These depressions interrupt the stream network continuity in reality, but r.watershed does a fantastic job making a continuous drainage network model, especially in these difficult areas.
I am trying to extract river network from a 5m DEM with some success using r.watershed. Has anyone tested this algorithm on high resolution LiDAR data for example - 1meter DTM and what kind of results have they obtained?