3. HAND Methodology
The baseline flood estimation method employed by the National Water Center (NWC) FIM services is based upon the Height Above Nearest Drainage (HAND) method. As explained by Zheng et al. (2018) and Liu et al. (2018), HAND is a geoprocessing technique that converts a Digital Elevation Model (DEM) to a Relative Elevation Model (REM) depicting the elevation of the surrounding terrain above the river to which it drains. HAND was first introduced by Rennó et al. in 2008 and since then has been used to support several geospatial applications around the world.
As seen in Figure 1, HAND is defined as the height of each grid cell with respect to the nearest stream cell it drains to. The HAND value of each grid cell indicates the water depth needed to inundate that cell. The inundated extent corresponding to a given water depth may then be determined by selecting all the cells with a HAND value less than or equal to the given depth.
Figure 1: Height Above Nearest Drainage (HAND) Method
Note that water depth, in this case, is relative to the elevation of the channel thalweg represented by the DEM and any value-added product provided by hydroconditioning the DEM. Hydroconditioning of the DEM involves processes that ensure surface flow patterns are consistent for the HAND grid calculation. An example of hydroconditioning is pit removal (or filling) within the DEM. “HAND depth” will be used to describe water depth throughout the rest of this document because the estimated water depth is relative to the elevation of the channel thalweg and does not represent an absolute water depth.
Figure 1 illustrates this concept, where terrain elevation above some common datum is converted into a HAND REM by subtracting the channel thalweg elevation of the nearest drainage cell from the terrain elevation of the terrain cell of interest. An example of the resulting HAND grid is pictured on the right. In this example, if we assume a HAND depth of 2 meters, using the HAND grid on the right, we can determine the inundation extent by finding all values in the HAND grid that are less than or equal to 2 meters. The light blue area in the HAND grid represents the resulting HAND depth. The HAND Schematic in Figure 2 also depicts the process of creating the HAND model REM using HAND Grids grouped by the nearest drainage cells.
Figure 2: HAND Schematic depicting how the HAND model REM is created using HAND Grids grouped by the nearest drainage cells
The NWC FIM services implementation of HAND relies upon a hydrographic network, such as the NHDPlus, to define the local drainage. A hydrographic network consists of stream reaches (i.e. river segments) and associated catchments. Catchments define local drainage boundaries for stream reaches while stream reaches define the geometry of the river and the conveyance for surface and subsurface runoff from the catchments. Since HAND is a geoprocessing technique that operates on a grid, a grid representation of the reach is assumed. Catchments, reaches, and a DEM are the primary inputs used to generate a HAND grid.
The hydrographic network is also used to convert streamflow into HAND depth. Using the HAND grid, reach-averaged channel properties are derived for incremental HAND depth values as seen in Figure 3:
-
The Reach Length (L) and Channel Bed Slope (S0) come from the attribute table of the flowline feature class in the hydrographic network.
-
Channel Top Width (W) is derived by dividing the Inundated Area (As) by the Reach Length (L).
-
Cross Section Area (A) is derived by dividing the Volume of the Inundated Area (V) by the Reach Length (L).
-
Wetted Perimeter (P) is derived by dividing the Inundated Bed Area (Ab) by the Reach Length (L).
-
Hydraulic Radius (R) is derived by dividing the Cross Section Area (A) by the Wetted Perimeter (P).
Figure 3: Reach-averaged channel hydraulic properties used in HAND model.
Using these reach-averaged hydraulic properties, a channel bed slope approximation, and a constant roughness coefficient for each reach in the hydrographic network, Manning’s Equation can be applied to compute a relationship between discharge and average HAND depth for each reach-catchment pair (Figure 4).
Figure 4: Manning’s Equation applied to compute a relationship between discharge and average HAND depth for each reach-catchment pair.
Figure 5 depicts the process of HAND synthetic rating curve (SRC) development. Over a range of HAND depth values, a single synthetic rating curve is developed for each reach relating a range of discharge values to a range of HAND depth values as seen in Figures 5 and 6. Please note that at this time, a constant roughness coefficient is assumed for the entire range of flows at a given reach. Additional development is needed to define roughness as a function of HAND depth. Channel bed slope is derived either from provided NHDPlus attributes or maximum and minimum channel thalweg elevations along the reach of interest.
Figure 5: Synthetic Rating Curve (SRC) Development Process.
Figure 6: Synthetic rating curve (SRC) developed for each reach.
The HAND-derived rating curves are generated and applied to what we will refer to as target reaches, which are derived from a hydrographic network. The blue pixelated feature pictured in Figure 6 is an example of a target reach. The RFC Replace & Route and NWM-based streamflow forecasts are generated on a separate hydrographic network of what we refer to as source reaches. The yellow line feature pictured in Figure 7 is an example of a source reach. Leveraging both the HAND grid and associated synthetic rating curves, inundation extent can be derived by converting forecast streamflow from the source reach associated with the hydrologic model (i.e. RFC Replace & Route and National Water Model (NWM)) to forecast HAND depth using the synthetic rating curve of the target reach. Because the rating curves and streamflow forecasts occur on distinct hydrographic networks a translation between source and target reaches is necessary. Once the HAND depth has been computed, the inundation extent for the catchment of interest is derived by selecting the HAND values less than or equal to the forecast HAND depth. This process is repeated for every reach-catchment pair in the domain.
Figure 7: Example of target reach vs. source reach.
For more information about HAND methodology, please refer to: Fernando Aristizabal, Fernando Salas, Gregory Petrochenkov, Trevor Grout, Brian Avant, Bradford Bates, Ryan Spies, Nick Chadwick, Zachary Wills, Jasmeet Judge. 2023. "Extending Height Above Nearest Drainage to Model Multiple Fluvial Sources in Flood Inundation Mapping Applications for the U.S. National Water Model.' Water Resources Research.