Academia.eduAcademia.edu

Figure 1 – uploaded by Dirceu L Severo

See full PDFdownloadDownload figure
Figure 1. Using methods described by Nobre ef al. (201 1a), the local drain directions (LDD) and drainage network (a) are extracted from a digita elevation model (DEM) (c). The LDD with the drainage network (a) are used to generate (b) the closest drainage map (each drainage cell is spatially associated with all DEM cells that drain into it), and then, the original DEM (c) is processed using the height above the nearest drainage (HAND operator (the height of the corresponding drainage-outlet DEM cell is subtracted from the height of each hillslope DEM cell) and the nearest drainags map (which guides the operator, denoting subtraction cell pairs), producing the HAND model (d), in which each cell height represents the difference it level to its respective closest drainage cell. The drainage network is then converted into a normalized topographic reference so that the HAND model n¢ longer retains a reference to sea level (diagram modified from Nobre et al., 201 1a)

Figure 1 Using methods described by Nobre ef al. (201 1a), the local drain directions (LDD) and drainage network (a) are extracted from a digita elevation model (DEM) (c). The LDD with the drainage network (a) are used to generate (b) the closest drainage map (each drainage cell is spatially associated with all DEM cells that drain into it), and then, the original DEM (c) is processed using the height above the nearest drainage (HAND operator (the height of the corresponding drainage-outlet DEM cell is subtracted from the height of each hillslope DEM cell) and the nearest drainags map (which guides the operator, denoting subtraction cell pairs), producing the HAND model (d), in which each cell height represents the difference it level to its respective closest drainage cell. The drainage network is then converted into a normalized topographic reference so that the HAND model n¢ longer retains a reference to sea level (diagram modified from Nobre et al., 201 1a)

Related Figures (11)

Figure 2. The height above the nearest drainage (HAND) model raster (a) is interpolated, and vector lines of equal height HAND (HAND contours) are generated, producing a HAND contour map (b)
Figure 2. The height above the nearest drainage (HAND) model raster (a) is interpolated, and vector lines of equal height HAND (HAND contours) are generated, producing a HAND contour map (b)
Figure 3. Hazard-ranked height above the nearest drainage contours (two classes) overlaid on (a) height above the nearest drainage-shaded relief topology (drainage-normalized terrain) and on (b) height above sea level-shaded relief topography. Data from the Sao Paulo metropolitan region (Nobre et al., 2010)
Figure 3. Hazard-ranked height above the nearest drainage contours (two classes) overlaid on (a) height above the nearest drainage-shaded relief topology (drainage-normalized terrain) and on (b) height above sea level-shaded relief topography. Data from the Sao Paulo metropolitan region (Nobre et al., 2010)
Figure 4. Santa Catarina State of Brazil; Itajai-Acu river valley (purple) and Blumenau municipality (red)
Figure 4. Santa Catarina State of Brazil; Itajai-Acu river valley (purple) and Blumenau municipality (red)
Figure 5. (a) Shaded relief, obstruction-free topography within the study area (DTMpjumenau: !-m grid, 1-m altimetric precision) and (b) satellite image of Blumenau city crossed by the Itajai-Agu river (Google Earth) and overlaid with the surveyed 2011 inundation map (Refosco et al., 2013)
Figure 5. (a) Shaded relief, obstruction-free topography within the study area (DTMpjumenau: !-m grid, 1-m altimetric precision) and (b) satellite image of Blumenau city crossed by the Itajai-Agu river (Google Earth) and overlaid with the surveyed 2011 inundation map (Refosco et al., 2013)
Figure 6. (a) Height above sea-level topographic contours and (b) height above sea-level hypsometric map for the Blumenau 2011 flooded area (including bodies of water)
Figure 6. (a) Height above sea-level topographic contours and (b) height above sea-level hypsometric map for the Blumenau 2011 flooded area (including bodies of water)
Figure 7. (a) Height above the nearest drainage (HAND) contours (HAND contour map) and (b) HAND hypsometric map for the Blumenau 2011 flooded area (including bodies of water) A hypothetical flood is defined as the probable flooding progression of a particular river. We produced HAND The frequency histograms illustrate the distribution of raster heights within the flooded area (Figure 8). The actual topography distribution of heights above sea level (DTMgiumenau) reveals that primarily channels (from 0 to 3m) and terraces (from 8 to 17m) cover the surface area occupied by floodwaters. The normalized topology distribution of the heights above the nearest drainage reveals that the relative topography is more homoge- neous. The HAND topology equates cells with the same relative vertical distance to the stream, regardless of whether those cells are positioned across different topographical heights.
Figure 7. (a) Height above the nearest drainage (HAND) contours (HAND contour map) and (b) HAND hypsometric map for the Blumenau 2011 flooded area (including bodies of water) A hypothetical flood is defined as the probable flooding progression of a particular river. We produced HAND The frequency histograms illustrate the distribution of raster heights within the flooded area (Figure 8). The actual topography distribution of heights above sea level (DTMgiumenau) reveals that primarily channels (from 0 to 3m) and terraces (from 8 to 17m) cover the surface area occupied by floodwaters. The normalized topology distribution of the heights above the nearest drainage reveals that the relative topography is more homoge- neous. The HAND topology equates cells with the same relative vertical distance to the stream, regardless of whether those cells are positioned across different topographical heights.
Figure 8. Frequency distribution within the 2011 flooded area for (a) heights above sea level in the DTMpiumenau and for (b) heights above the nearest drainage in the height above the nearest drainage model
Figure 8. Frequency distribution within the 2011 flooded area for (a) heights above sea level in the DTMpiumenau and for (b) heights above the nearest drainage in the height above the nearest drainage model
Figure 9. Inundation-extent predictions computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event. Overestimates (commission) and underestimates (omission) in relation to the actual 2011 observed flood are shown after (a) pruning the excess according to the height above the sea-level cut-off level (17 m) and by (b) pruning the excess by enlarging the channel initiation-accumulated area threshold by ten times (suppressing low-order channels) When diagnosing the spatial distribution of overesti- mation, we experimented with two methods (Figure 9). The first method used an arbitrarily chosen height above sea level to prune areas predicted by HAND contours above a certain altitude. In this case, the height above sea level corresponded to the highest point below the waterfall where inundation was verified. Eliminating predicted areas greater than altitude of 17m reduced the HAND contour overestimation to 54.43%. The predicted flood extent was more accurate for areas near and at the The tributaries in the study plot are primarily first- order Strahler channels that drain local hills high above the Itajai-Acu. Thus, the overestimation of the HAND contour in this case exhibited a treatable dependence on the Strahler order of the drainage network. For the second analysis, we increased the channel initiation threshold from 0.405 to 4.05km* to alter the Strahler order hierarchy, which suppressed the first-order streams for HAND calculations. This technique reduced the HAND contour overestimation to 25.95% (Table I). With its channel-gradient independence, the HAND contours followed rivers indefinitely (no cut-off points). However, a few affected first-order streams that flow level with the Itajai-Acu flooded channel were excluded from the HAND contours, which resulted in a 7.74% underes- timation (omission). Interestingly, the flooded area above the NW waterfall that was predicted using this method did not occur during the actual 2011 event.
Figure 9. Inundation-extent predictions computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event. Overestimates (commission) and underestimates (omission) in relation to the actual 2011 observed flood are shown after (a) pruning the excess according to the height above the sea-level cut-off level (17 m) and by (b) pruning the excess by enlarging the channel initiation-accumulated area threshold by ten times (suppressing low-order channels) When diagnosing the spatial distribution of overesti- mation, we experimented with two methods (Figure 9). The first method used an arbitrarily chosen height above sea level to prune areas predicted by HAND contours above a certain altitude. In this case, the height above sea level corresponded to the highest point below the waterfall where inundation was verified. Eliminating predicted areas greater than altitude of 17m reduced the HAND contour overestimation to 54.43%. The predicted flood extent was more accurate for areas near and at the The tributaries in the study plot are primarily first- order Strahler channels that drain local hills high above the Itajai-Acu. Thus, the overestimation of the HAND contour in this case exhibited a treatable dependence on the Strahler order of the drainage network. For the second analysis, we increased the channel initiation threshold from 0.405 to 4.05km* to alter the Strahler order hierarchy, which suppressed the first-order streams for HAND calculations. This technique reduced the HAND contour overestimation to 25.95% (Table I). With its channel-gradient independence, the HAND contours followed rivers indefinitely (no cut-off points). However, a few affected first-order streams that flow level with the Itajai-Acu flooded channel were excluded from the HAND contours, which resulted in a 7.74% underes- timation (omission). Interestingly, the flooded area above the NW waterfall that was predicted using this method did not occur during the actual 2011 event.
Table I. Height above the nearest drainage (HAND) contour performance as inundation predictor for two overestimation removal techniques, height cut-off and elimination of first-order channels (total study area 38. 25 km”, channel area 1.94km”, flood 2011 area 6.14 km’, HAND risk-zone area 9.12 km” ) DEM, digital elevation model; DSM, digital surface model.
Table I. Height above the nearest drainage (HAND) contour performance as inundation predictor for two overestimation removal techniques, height cut-off and elimination of first-order channels (total study area 38. 25 km”, channel area 1.94km”, flood 2011 area 6.14 km’, HAND risk-zone area 9.12 km” ) DEM, digital elevation model; DSM, digital surface model.
Figure 10. Map showing the depth of the flood layer as the inverse of the height above the nearest drainage, thus simultaneously rendering this an inundation-extent and flood layer depth map. The inundation-extent prediction was computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event (commission reduced by removing low-order drainage channels) The HAND contours delineate a stepwise local drop to the nearest river channel. A 1-m HAND contour denotes that when the waterline increases by 1m from the channel, the floodwaters will reach and match the 1-m normalized contour. In addition, the flood layer would be 1-m thick at the river where the HAND height is zero and zero (absent) at the 1-m HAND contour. The depth of the flood layer is the arithmetic inverse of the HAND height. Thus, a hypsometric HAND contour map can simulta- High-resolution topographical data, such as those provided in the DTMpiumenau, are rarely available for areas in South America or for other less developed continents. If a new flood-extent prediction method requires access to this type of data, its application may be restricted. After demonstrating the capacity of HAND
Figure 10. Map showing the depth of the flood layer as the inverse of the height above the nearest drainage, thus simultaneously rendering this an inundation-extent and flood layer depth map. The inundation-extent prediction was computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event (commission reduced by removing low-order drainage channels) The HAND contours delineate a stepwise local drop to the nearest river channel. A 1-m HAND contour denotes that when the waterline increases by 1m from the channel, the floodwaters will reach and match the 1-m normalized contour. In addition, the flood layer would be 1-m thick at the river where the HAND height is zero and zero (absent) at the 1-m HAND contour. The depth of the flood layer is the arithmetic inverse of the HAND height. Thus, a hypsometric HAND contour map can simulta- High-resolution topographical data, such as those provided in the DTMpiumenau, are rarely available for areas in South America or for other less developed continents. If a new flood-extent prediction method requires access to this type of data, its application may be restricted. After demonstrating the capacity of HAND
Figure 11. Inundation-extent predictions conducted using low-quality Shuttle Radar Topography Mission (SRTM) digital surface model data (30-m grid, 1-m vertical resolution) computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event (commission reduced by removing low-order drainage channels)
Figure 11. Inundation-extent predictions conducted using low-quality Shuttle Radar Topography Mission (SRTM) digital surface model data (30-m grid, 1-m vertical resolution) computed by height above the nearest drainage contours for a 13-m hypothetical flood equivalent to the 2011 Blumenau event (commission reduced by removing low-order drainage channels)
Related topics:
Environmental EngineeringCivil EngineeringEnvironmental ScienceFlood Inundation ModelingInundation MappingTerrain ModelingFlood-extent predictiontopographical contours
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved