How Flood Depth works

Flood Depth answers one question: for a given flood, which ground goes under water, how deep, how sure are we, and what is there? The only required input is a digital elevation model (DEM), which is freely available for the whole planet. That makes it usable in places with no gauges, surveys or hydraulic models.

1 · Hydrologically conditioned terrain

Depressions are removed with Priority-Flood + ε (Barnes et al., 2014), so every cell drains. Flow follows the steepest of eight neighbours (D8, O'Callaghan & Mark, 1984). Flow accumulation gives each cell its contributing area, and channels begin where that area exceeds the stream threshold. Where the DEM is too flat to resolve channels (deltas, swamps), mapped OpenStreetMap rivers are burned 10 m into the routing surface (Hellweger, 1997). HAND is still measured on the un-burned surface. Sea and lakes below a chosen water level are masked and act as base level.

2 · Height Above Nearest Drainage (HAND)

Each cell's HAND is its elevation minus the elevation of the channel cell it drains to, following the flow path (Rennó et al., 2008; Nobre et al., 2011). HAND normalises terrain by the drainage network. Low HAND means hydrologically close to the river, and it correlates strongly with soil saturation and flood occurrence.

3 · Synthetic rating curves

For each reach, at stage h, every cell in its catchment with HAND < h is inundated. Summing over those cells gives surface area, volume V and bed area A_b. The reach-averaged cross-section is A = V/L, wetted perimeter P = A_b/L, and Manning's equation gives Q = (1/n)·A·(A/P)^2/3·S^1/2 (Zheng, Tarboton et al., 2018). This is the method behind NOAA's National Water Model flood inundation maps (Liu et al., 2018).

4 · Discharge in data-sparse catchments

• Water level: a uniform stage, for scenario testing and matching observed flood marks.
• Rainfall: Rational method Q = C_f·C·i·A/3.6. Time of concentration uses Kirpich (1940) on the longest flow path. Intensity comes from the 24-h depth by the IMD ratio i_t = (P₂₄/24)(24/t)^1/3, and the Leclerc & Schaake (1972) areal reduction factor is applied. The Rational method is best for catchments under ~50 km². Use it as an indicative upper bound for larger ones.
• Discharge: one reported or gauged peak, transferred to ungauged reaches by drainage-area ratio (exponent 0.5–0.9 is typical).

5 · Uncertainty

Global DEMs have metre-scale vertical error. Flood Depth treats HAND error as Gaussian with standard deviation σ, so the inundation probability is P = Φ((h − HAND)/σ). The P = 0.5 contour equals the deterministic extent. Suggested σ: 0.5 m for LiDAR, 1–2 m for Copernicus/FABDEM, 2–4 m for SRTM in vegetated or urban terrain.

6 · Exposure

Buildings and facilities come from OpenStreetMap. In much of the Global South they are available thanks to Humanitarian OpenStreetMap Team mapping. Each one is sampled for depth and probability.

Limitations

• HAND is a static, “bathtub-along-the-network” method. It ignores backwater, levees below DEM resolution, flood dynamics and coastal surge.
• DEMs record the water surface, not the river bed, so channel conveyance is under-estimated at low stages.
• The default global DEM (SRTM-derived) includes tree canopy and buildings. Upload FABDEM or LiDAR where possible.
• Only catchment area inside the analysis box is counted. For a large river entering from outside, use Discharge or draw the box around the whole catchment.
• Use the results for screening, prioritisation and early action, not for engineering design.

References

1. Barnes, R., Lehman, C., Mulla, D. (2014). Priority-Flood: An optimal depression-filling and watershed-labeling algorithm. Computers & Geosciences 62, 117–127.
2. Rennó, C.D. et al. (2008). HAND, a new terrain descriptor using SRTM-DEM. Remote Sensing of Environment 112, 3469–3481.
3. Nobre, A.D. et al. (2011). Height Above the Nearest Drainage – a hydrologically relevant new terrain model. Journal of Hydrology 404, 13–29.
4. Zheng, X., Tarboton, D.G., Maidment, D.R., Liu, Y.Y., Passalacqua, P. (2018). River channel geometry and rating curve estimation using height above the nearest drainage. JAWRA 54(4), 785–806.
5. Liu, Y.Y., Maidment, D.R., Tarboton, D.G. et al. (2018). A CyberGIS integration and computation framework for high-resolution continental-scale flood inundation mapping. JAWRA 54(4), 770–784.
6. O'Callaghan, J.F., Mark, D.M. (1984). The extraction of drainage networks from digital elevation data. CVGIP 28, 323–344.
7. Kirpich, Z.P. (1940). Time of concentration of small agricultural watersheds. Civil Engineering 10(6), 362.
8. Leclerc, G., Schaake, J.C. (1972). Derivation of hydrologic frequency curves. MIT Report 142.
9. Hellweger, F. (1997). AGREE – DEM surface reconditioning system. University of Texas at Austin.
10. Horn, B.K.P. (1981). Hill shading and the reflectance map. Proc. IEEE 69(1), 14–47.