Skip to end of metadata
Go to start of metadata

Call #4: DEM Geomorphology Toolbox and ArcGOS Pro Spatial Analyst Toolbox for high res hydrologic models

Date: 12/02/2019

Time: 11:00 am - 12:00 pm MT / 1:00 - 2:00 pm ET

Recording: 191202-CDI-Geomorph.mp4


  1. Introductions
  2. Presentation by Jennifer Cartwright: Lidar-derived digital elevation models (DEMs) enable detailed assessments of stream-channel geomorphology but require careful preparation to produce accurate stream geography and geomorphic metrics. The DEM Geomorphology Toolbox is a suite of ArcGIS tools to automate the identification of sites of geomorphic instability that may represent sediment sources and sinks in stream-channel networks. This toolbox enables users to modify input DEMs based on stormwater-infrastructure locations and derive stream networks at user-specified resolutions. The toolbox also identifies steep stream banks, abrupt changes in channel slope, and areas of rough terrain, which may be associated with channel incision, head cuts, and gully formation. Spatial clusters of outputs from multiple tools can help prioritize field efforts to assess and restore eroding stream reaches.
  3. Presentation by Luke Sturtevant: Advances in lidar technology and lidar post processing technology have made leaps and bounds in the past decade. We are harnessing the power of python and the Spatial Analyst toolbox in ArcGIS Pro to derive high resolution hydrologic models on a HUC8 scale from the most current lidar terrain at full resolution. Breaching algorithms have been developed and are still being refined to locate flow dams in the terrain such as culverted roads that would otherwise divert the natural flow course of streams. The breached digital elevations models are then passed through custom scripts to calculate flow accumulation, flow direction, and stream order grids to derive stream lines, watershed boundaries, and pour point layers with flow connectivity attributes and unique identifiers. Finally, linked value added attribute tables are built for drainage area, peak flows, and other basin characteristics for each stream segment, pour point, and/or associated catchment area.  
  4. Wrap up discussion and questions

Call #3: Channel Incision Mapping using Lidar

Date: 10/31/2019

Recording: 191031-CDI-Geomorph.mp4


  1. Introductions
  2. Presentation by Marina Metes (USGS), Remotely detecting channel incision in headwater streams using lidar and topographic openness
  3. Wrap up discussions and questions

Meeting notes:

  1. Introductions
    1. Leslie Hsu (CDI coordinator)
  2. Presentation by Marina Metes: A method was developed to map reach-scale incision from lidar-derived digital elevation models using topographic openness, a landscape metric measuring the enclosure of an area (i.e. channel bottoms) relative to the surrounding landscape (i.e. stream banks). The method was validated with field surveys and local photogrammetric models of stream banks
    1. Topographic openness - Landscape metric to measure how open or enclosed an area is
      1. Positive and negative openness, can use short or long search radius
      2. Search radius can identify local vs. regional features
      3. Each pixel, radius searches in 8 directions and is constrained by “line-of-site”-
        1. Uses average of the 8 directions
        2. Upstream and downstream direction might be a flat surface, which could lower the overall value. Field comparisons-pulled out the two largest values (removing the upstream, downstream values) which doesn’t change the overall distribution
      4. Values of 90 degrees indicate a flat surface
      5. Result: Grids based on openness, one for negative, one for positive
        1. Highlights streams very well
        2. Currently being used to look at channel head
      6. Color ramp on openness: 36-98 but 84-88 could be riparian
        1. Really low values in positive openness inside the channel could be restrained by steep banks
        2. Interesting variations in channel patterns (incised to deposition)
      7. Research questions: Can openness detect channel incision? What does this reveal about spatial patterns of incision?
      8. Study area:
        1. Maryland Piedmont in Chesapeake Bay: urban watersheds
        2. Agriculture history, to urban landscapes
      9. Methods
        1. Overlap openness values with stream network
        2. values <81 indicate severe incision, 81 - 84 indicate moderate incision, and values > 84 indicate no incision
        3. Check against survey data
          1. Accuracy between 70 - 75%; Cohen’s Kappa=0.536 which means moderate agreement
      10. Applications
        1. Look at incision changes over time for urbanizing watershed
        2. Raw differences in openness changes to look at magnitude of change
        3. Slope of channel as plotted against incision
        4. Openness for sediment modeling, bank erosion
        5. GRASS GIS has geomorphon tools that uses each of the 8 measurements to look at the landforms, helps determine if on hillslope, channel, etc. Might be able to use for floodplains, toe slopes, etc. Allows to deep dive into dissecting the landscape.
  3. Q&A
    1. Faith Fitzpatrick: It looks like it is picking up the confined/unconfined valley type and the channel, and when it moves into the intermediate area it picks up indications of an entrenched valley where the meander belt width abuts the valley width. The entrenched valleys usually have a lot of geomorphic and sediment activity.
    2. Peter McCarthy: How performant are these tools to evaluate openness? Can it be scaled to much larger basins with pretty good performance?
      1. Could be scaled up. Used a 36-meter search radius and ran in 10-15 minutes
    3. Krissy Hopkins: Done for whole state of NC on 3-meter DEM
    4. How did statewide dataset compare for smaller first order streams to larger wide rivers?
      1. Similar patterns could be seen at larger scales with larger 3-meter
    5. Marina’s study used 1.8 meter resolution
    6. Scalability: Can we use resampled coarser lidar data

Call #2: Riparian Corridors and Rapid Flood Inundation Mapping

Date: 10/1/19


  1. Introductions
  2. Presentation by Sinan Abood, USDA Forest Service National Riparian areas project (20 minutes)
  3. Presentation by Greg Petrochenkov (USGS) and Fernando Aristizabal (NOAA), Developing quick estimates of flood inundation for CONUS using an optimized GIS Flood Tool (20 minutes) 
  4. Wrap up discussions and questions (20 minutes)

Attendees: Peter McCarthy (StreamStats), Jeremy Newson, Sinan Abood, Luke S., Greg Noe, Sam Lamont, Marina Metes, Mike Wieczorek, Pete Steeves, Andrea Medenblik (SAWSC), Faith Fitzpatrick (joined late), Roland Viger (joined very late), Linda Ann Spencer, Caelan Simeone

Meeting notes:

  1. Introductions
  2. Presentation by Sinan Abood, USDA Forest Service National Riparian areas project (20 minutes)
    1. Using freely available data
    2. Goal: Develop riparian base maps
    3. Data: 
      1. NHDPlus 100k scale (medium resolution)
      2. NWI for wetlands inventory, 
      3. gSSURGO (model can use gSSURGO for Hydric Rating, drainage class, soil group, etc. 
      4. National map DEMs (model can use several scales (10-meter, 5-meter etc.), 
      5. NLCD-30-meter NLCD; 
      6. Hydrologic data (26,000 USGS gages): 50-year flood heights
    4. 50-year flood height to describe riparian areas
      1. This flood height will intersect with terrace and be more likely to represent riparian areas. 100-year flood height frequently includes upland areas.
      2. Any gage with less than 30 years was not included
      3. Historic gages not included (active gages only)
      4. Most gages on stream order 3+. Fit gage height to stream order.
    5. Python 3 packages in ArcGIS pro
      1. Can utilize high resolution data (i.e. 1-meter DEMs)
      2. 2 toolboxes:
        1. Riparian batch
        2. Data and result prep
    6. Results: Riparian areas for CONUS on 100k streams at 10-meter spatial resolution
    7. Applications:
      1. NLCD2016 correlation with riparian mapping
      2. Increase in cultivation in riparian areas
      3. Land cover change (rate): Sum of loss between 2001 and 2016 was significant (didn’t capture number)
      4. Riparian base maps can be used to mask NAIP imagery and then evaluate high resolution classification in lowland wide floodplains (Upper Rio Grande example)
      5. Areas with high road density in riparian areas 
    8. Accuracy:
      1. Assessment in Hiawatha NF
      2. 88% mapping accuracy without any model corrections
      3. Model correction increased accuracy to 93%
    9. Model Parameters:
      1. Flood value
      2. Stream positional inaccuracies
      3. DEM spatial resolution: 4% increase in using different resolution
      4. Sampling distance
    10. Future considerations
      1. Stream classification and stream order
      2. Hydrologic data debate (50-year vs. 100-year)
      3. Watershed scale
      4. Inherited errors from input data
    11. Tool
      1. Places sample points on map; distance between points is 75% of density (7.5 meter for 10-meter).
      2. Number of transects can be adjusted by sampling distance (250 or 500 meter being used?)
      3. Elevation of stream and 50-year flood height was used to dev
    12. Citations/Publication
      1. Story map 
      2. Presentation
      3. Toolbox found here 
      4. Delineating and mapping riparian areas for ecosystem service assessmen
  3. Presentation by Greg Petrochenkov on rapid flood estimation tools: 
    1. GFT rapid flood inundation tools - paper originally written by Kris Verdin
    2. How to rapidly estimate flood inundation without using computationally expensive modeling tools
    3. Methods
      1. Retrospective NWM output for last 20 years and use maximum output
      2. Originally the method was to create the stream network dynamically
      3. Optimized method is using NHDPlusV2.0 stream network
      4. Use the raw DEM rather than the hydro-enforced DEM for mapping inundation
      5. Mannings’s n values: Using a simplified approach based on stream slope. Other methods can be used or a static value can be applied.
      6. Similar to HAND method (height above nearest drainage). 
      7. Can specify how many per stream (or based on length of stream).
      8. Create a synthetic rating curve for each cross section 
      9. Interpolation between cross sections.
    4. Drawbacks:
      1. Watershed delineated from a D8 grid
    5. Example:
    6. Performance: on YETI machine, CONUS in 2.5 hours
      1. Some steps can be pre-computed
    7. Code review us underway; should be published soon.
  4. Wrap up discussion and questions

Call #1: Discussion and Presentation about FACET

Date: 8/29/19

Attendees: Peter McCarthy (StreamStats).  Greg Noe (WMA ESPD), Krissy Hopkins (SAWSC), Jeremy, Sinan Abood (USFS), Fernando Aristizabal (NWS), Luke S., Roland Viger, Peter Claggett, Mike Wieczorek (NAWQA). Labeeb Ahmed (working with Peter, Krissy, and Greg on FACET)

Meeting notes:

  1. Introductions and short description of tools being used for channel shape/geometry from lidar or DEMs.
    1. Uses USGS gages, soils, and a number of other data to 
    1. Roland Viger and Greg Petrochenkov (WMA-Geo-Intelligence Branch) on GFT tool update related to rapid flood inundation tool (Verdin and others), Manning’s n solutions (w/Pete McCarthy and Mike Wieczorek), collaborating with Fernando A @ NWS comparing GFT w/HAND & improving methods, Sinan Abood @ USFS (w/Mike Wieczorek)
    2. Luke S.: New England WSC; work related to FEMA studies for approximation of flood zones using USGS regression equations to get Q; and automated tools for cutting cross sections from lidar, computing Manning’s n, bankfull width and depth
    3. Sinan: research scientist with Forest Service; national riparian base map; riparian delineation model, free tool to download. ( and for more details regarding USFS National Riparian areas project
    4. Peter C: Started FACET tool in 2014, development started with W. Virginia University. V1 of the tool
    5. Fernando Aristizabal (NWC): Continental scale high-resolution flood inundation mapping with National Water Model forecasts, synthetic rating curves, and HAND. Currently working on techniques to improve limitations with HAND.
  2. FACET discussion/presentation
    1. Working on version 2 now; 
    2. Python based code which will be hosted on (correct address?).
    3. Look at floodplains to quantify and evaluate how floodplains provide services.
    4. Greg lead field effort to collect data at 68 locations. Measure stream channel and floodplain. 
    5. Looking at deposition and bank erosion rates. 
    6. Dendrohydrology
    7. How do we scale local measurements up to basin and reach based estimates?
    8. Using characteristics of channel and floodplain as well as various land use estimates to predict storage and transport of sediments.
    9. Computing costs/value related to sediment erosion and storage
    10. Extracting data from DEM/lidar. Extracting bankfull width and bank height characteristics
    11. Breaching depressions with White Box Tools
    12. West Branch Brandywine Creek
  3. Working group/future meetings?
    1. Sinan has volunteered to present at the next meeting
    2. Pete M. and/or Roland will pursue a Doodle Poll for next meeting.
  4. CDI (Community for Data Integration) group: Develop a list of tools and links to various efforts


  1. Bathymetry data?
    1. Did not have bathymetry data, used channel surface as bottom of channel, so likely underestimating height, but focus was smaller streams. Max DA was 3,000 sq. kilometers
  2. Complete Delaware basin (West Branch Brandywine only?)
    1. Completed for entire Delaware basin and part (~⅓ of Chesapeake bay)
    2. Lidar: Combination of sources, some from national map some from local sources, using 3 meter DEM. Tested at 1-meter as well, found that at 3-meter resolution the results were similar and ran much faster.
  3. Toolboxes developed in python (available as toolbox in ESRI?)
    1. Great interaction with (name?) developer of White Box. Very responsive. Have not tested White Box against TauDem.
    2. Is WhiteBox multi-threaded? Can it be set up on larger mega systems?
    3. Conditioned DEM before breaching
      1. Roads and railroads have been breached
    4. Best breaching algorithm they’ve used so far was developed by Univ. of Maryland, using r.geomorphon tool in GRASS
    5. Current breaching model for HUC10 take 5-10 minutes now
    6. Luke: Has algorithm using NumPy, pretty good but maybe not quite as fast
    7. Identified points of intersection by using roads and streams layer, created a buffer, and breached through bridges and/or other structures easily. 
    8. Group has interest in hearing more about the Fill-n-Spill algorithm from Al Rea and others.
    1. Luke: not in scope
    2. Pete: SStats 
    3. Roland: yes, GFT is being sent out for code review in a few weeks
    4. Rich Barnes had some interesting stuff but wasn’t able to maintain support. (”RichDem” git site).
    1. Standalone and open-source tools but currently reliant on Tau-DEM tools
    2. White Box tools: has potential to replace Tau-DEM
    3. Publishing code?
  4. Scaling processing to deal with large (or national) data sets?
  5. Ideas about how to compare alternate realizations with observations? With each other?
    1. In DRB, field work on cross-sections, field delineation for mapping 2-year floodplain
    2. Definition of “flood plain” or whatever the intended feature is very important for assessing quality of results, testing predictive value of derived metrics. “Active flood plain” is a particularly tricky one.
  6. Publishing software:
    1. Hydrologic condition tool is what they’re currently calling the tools
    1. Roland: Encourage everyone to let Roland and Pete M. about code so we can help publish and/or incorporate into other programs/workflow
    2. Luke on tools from NEWSC for FEMA work: Gave a presentation to NHD folks, also included RichDEM and NumPy, has not tested on anything less than 2-meter. Runs HUC12 in ~10 minutes.
    3. Iowa State University (Brian Gelder): CUTTER tools (link to presentation)
    4. Scott Haag (Drexel Univ) has developed a rapid watershed delineation tool that created the Mississippi drainage basin in under a minute from a 10m DEM
  7. Move to Community for Data Integration focus group? (e.g., Water Group under Earth-Science Themes Working Group (ETWG)) and Riparian Mapping and Methods for Floodplain Delineation.
  8. Paraphrase of workflow: clean up the DEM, derive some DEM stuff. Analyze DEM to delineate and characterize 2-year bankfull width

Another resource that might be of interest: Liu, Y. (2018). Height Above Nearest Drainage (HAND) for CONUS, HydroShare,

  • No labels