Watershed Delineation

Open Watershed Delineation

Delineate catchment boundaries, extract upstream river networks with Strahler stream ordering, and trace the longest flowpath — for any location on Earth between 60°S and 85°N — using the global MERIT-Hydro hydrography dataset. The outputs (area, perimeter, outlet coordinates, longest watercourse length) plug directly into downstream HydroDesign tools such as the Watercourse Profile, Rational Method, and SCS Unit Hydrograph calculators.

Overview

Watershed delineation (also called catchment delineation or basin delineation) is the process of determining the drainage area upstream of a given outlet point — sometimes called the pour point. Every drop of rain that falls within the delineated boundary eventually drains through the outlet, making this the foundational step for virtually every hydrological analysis: runoff estimation, flood frequency analysis, reservoir yield, water-quality modelling, and environmental flow assessment.

HydroDesign’s Watershed Delineator uses the Global Watersheds API (Heberger, 2022), which combines pre-computed vector unit catchments from MERIT-Basins with on-the-fly raster delineation on the MERIT-Hydro flow grid. The result is a catchment polygon, a Strahler-ordered river network, and a longest-flowpath line — delivered in seconds for most outlets worldwide.

Free to use

Watershed delineation requires no credits. It is available to all registered HydroDesign users at no cost, with unlimited re-runs while you refine the outlet.

Setting the Outlet Point

The outlet (pour point) is the single point through which all water in the catchment must pass. It defines the catchment: move the outlet upstream and the contributing area shrinks; move it downstream and it grows. You can set the outlet two ways:

• Click the map — the easiest method. Click anywhere and the coordinates are captured and filled into the input fields automatically. Recommended for most workflows: you can see the terrain and river network as you position the point.
• Enter coordinates — type latitude and longitude directly into the input fields (WGS-84 / EPSG:4326, decimal degrees). Useful when you have a surveyed structure location, a GPS waypoint, or a coordinate from another dataset.

Valid ranges: latitude between -60° and +85°; longitude between -180° and +180°. Points in the ocean, on small islands outside MERIT-Hydro coverage, or in the high Arctic/Antarctica will fail with a clear error.

Why pour-point snapping matters

The Global Watersheds API automatically snaps your clicked point to the nearest cell on the MERIT-Hydro flow-accumulation grid. This is critical: the delineation algorithm traces upstream along the grid, so if the outlet sits one pixel off the channel the tool would follow a hillslope instead of the river, returning a tiny, wrong catchment. Snapping corrects for click precision and minor DEM-vs-OSM alignment errors. If the river is visible on the basemap, click near it — the snapper will handle the rest.

Coastal and estuarine outlets

Along tidal estuaries and deltas, the apparent river channel on satellite imagery may not match MERIT-Hydro’s hydrologic flow grid. If you get an unexpectedly small catchment at a coastal outlet, try moving the point 100 – 500 m upstream to the clearly defined freshwater channel before the estuary.

Data Source (MERIT-Hydro)

The underlying elevation data come from MERIT-Hydro (Multi-Error-Removed Improved-Terrain Hydro), a global hydrography dataset at 3-arcsecond (≈ 90 m at the equator) resolution developed by Yamazaki et al. (2019) at the University of Tokyo.

MERIT-Hydro starts from the MERIT DEM — itself built from SRTM v4.1, AW3D-30 m, and Viewfinder Panoramas and cleaned of vegetation bias, stripe noise, speckle, and absolute-bias errors — and then applies hydrologic conditioning to produce a globally consistent set of:

• Flow direction grid (D8 — each pixel drains to exactly one of its eight neighbours along the steepest descent; Jenson & Domingue, 1988).
• Flow accumulation grid (number of upstream cells draining through each pixel).
• River network vectors derived from the flow-accumulation grid.
• Unit catchments (MERIT-Basins) — small pre-computed sub-catchments used to accelerate delineation of large basins.

MERIT-Hydro vs. SRTM — when to use which

Aspect	MERIT-Hydro (this tool)	Raw SRTM / NASADEM
Resolution	3 arc-sec (~90 m)	1 arc-sec (~30 m) for SRTM v3 / NASADEM
Hydrologic conditioning	Yes — pits filled, flat areas resolved, channels carved	No — raw DEM needs pre-processing
Global coverage	60°S – 85°N, land only	60°S – 60°N (SRTM); NASADEM same
Vegetation / building bias	Removed	Present — adds several metres of height in forests and cities
Best for	Automated delineation, river networks, regional studies	Detailed site-level terrain, 2D hydraulic models on cleaned DEM

For most design-flood and planning work the ~90 m MERIT-Hydro resolution is more than adequate, and the hydrologic conditioning means you get correct drainage networks without manual pit-filling. For very small catchments (< 1 km²) or detailed site hydraulics, a locally sourced higher-resolution DEM is often warranted.

Coverage

MERIT-Hydro covers land surfaces between 60°S and 85°N — essentially the entire globe except Antarctica and parts of the high Arctic. Very small islands and atolls, and some isolated coastal promontories, may have limited coverage; these will return an error rather than an incorrect result.

Precision Options

Two precision levels are available. Select the one that matches the accuracy and turnaround you need.

Low precision (fast)

Uses pre-computed MERIT-Basins vector unit catchments. The algorithm walks upstream through the unit-catchment graph and unions every unit feeding the outlet. Fast (typically 1 – 3 seconds) but the boundary near the outlet is the outer edge of whichever unit catchment contains it — so some downstream area may be included beyond the exact pour point.

Best for: quick surveys, regional comparisons, large catchments, and situations where the absolute catchment area is not the single critical output.

High precision (detailed)

Combines vector unit catchments upstream with raster-based delineation (~90 m resolution) for the terminal unit catchment that contains the outlet. The terminal boundary follows the exact raster flow grid to the pour point, giving a much more accurate boundary at the outlet itself.

Best for: engineering design, culvert and bridge sizing, regulatory submissions, and any study where the area at a specific outlet must be reported accurately.

Large watersheds auto-revert to low precision

Catchments exceeding 50,000 km² automatically revert to low precision regardless of the selected option. This is a server-side limit to keep processing times reasonable — a raster delineation of the full upstream area of, say, the Orange or Vaal would take minutes to tens of minutes.

Simplify & Beautify

Two optional post-processing steps are available on the polygon output:

Simplify reduces the number of vertices using the Douglas-Peucker algorithm. A raster-derived boundary at 90 m resolution can have tens of thousands of vertices following every pixel edge — simplification collapses straight runs into single segments, cutting file size by 10× or more with no perceptible change in the boundary position. Recommended for web maps, reports, and GIS overlays where exact pixel fidelity is not required.

Beautify applies curve-fitting smoothing to the boundary to remove the characteristic staircase effect of raster-derived polygons. The result is a visually appealing, cartographically pleasing polygon. The boundary position may shift by up to one pixel (~90 m); this is acceptable for presentation maps and reports, but for strict area calculations leave beautify off. See Beautify below for details.

Beautify

Beautify smooths the watershed boundary by applying curve-fitting to remove the staircase effect inherent in raster-derived polygons. It produces a more visually appealing, rounded polygon that reads better on maps and in reports, but it may slightly alter the boundary position — generally by less than one pixel (~90 m). For engineering reports where the area and perimeter must tie exactly to the underlying DEM delineation, leave beautify off; for presentation and concept-stage work, turn it on.

Data Source Selector

The Watershed Delineator supports multiple delineation backends. Availability depends on the region of your outlet:

Source	Coverage	Notes
Global Watersheds (mghydro)	Worldwide (60°S – 85°N)	Default. MERIT-Hydro based. Fast, reliable, unlimited.
ESRI ArcGIS Watershed	Worldwide	Alternative backend using HydroSHEDS. Useful for cross-check.
USGS StreamStats	USA only	Highest resolution in the US; provides regional flood-frequency estimates as a by-product.
USGS NLDI	USA only	Network-navigation API — good for gauge-based delineation.

US-only sources are disabled (greyed out) when the outlet is outside the USA. For South African, African, or most non-US work, use Global Watersheds or ESRI ArcGIS.

Catchment Area

The reported catchment area (in km²) is computed from the delineated polygon using a geodesic area calculation on the WGS-84 ellipsoid. This accounts for the Earth’s curvature and is more accurate than the naive planar-coordinate calculation that would underestimate area at low latitudes and overestimate at high latitudes.

For a polygon with vertices $(\varphi_i, \lambda_i)$ in latitude and longitude, the geodesic area is evaluated by Karney’s algorithm (implemented by the Global Watersheds API):

$$A \;=\; \oint R(\varphi)^{2}\,\cos(\varphi)\,\mathrm{d}\lambda \;\mathrm{d}\varphi$$

Geodesic area on the WGS-84 ellipsoid — schematic

where $R(\varphi)$ is the ellipsoidal radius at latitude $\varphi$. In practice this is evaluated numerically segment-by-segment around the polygon boundary.

The perimeter is likewise reported as the geodesic distance around the boundary, and the outlet is reported as the snapped pour-point coordinates (WGS-84, decimal degrees) — which may differ slightly from the coordinates you clicked due to snapping.

Quality-check the area

Always compare the computed area against known or published catchment areas — from DWS gauging station metadata, the SANRAL Drainage Manual, previous studies in the same basin, or a planimeter estimate from a 1:50 000 topographic map. Discrepancies greater than ~10% usually indicate that the outlet was placed slightly upstream or downstream of the intended location, or that the catchment straddles a poorly defined divide in flat terrain.

River Network & Stream Order

The tool extracts all upstream river segments within the catchment and classifies them using Strahler stream ordering (Strahler, 1957), a classical geomorphic index of river-network complexity:

• Order 1 — headwater streams with no tributaries.
• Order 2 — formed by the junction of two order-1 streams.
• Order 3 — formed by the junction of two order-2 streams.
• In general, order $n$ requires the junction of two order-$(n-1)$ tributaries. Joining a lower-order stream to a higher-order stream does not raise the order.

A catchment’s highest stream order is a compact description of its branching complexity: a Strahler-1 catchment is a single gully; a Strahler-4 catchment has substantial network structure; a Strahler-7+ system is a major river basin.

On the map, river width is rendered proportionally to stream order for visual clarity. Hover over any river segment to see its stream order and name (where the underlying dataset carries a name). For export, each river segment carries its order as an attribute in the shapefile and GeoJSON outputs.

Engineering use of stream order

Stream order provides a quick indication of river size and branching complexity — useful for preliminary screening of road-crossing locations (higher orders likely need larger structures), culvert sizing (order-1 streams usually take pipes or small box culverts; order-3+ often require bridges), and flood-risk prioritisation. It is not a substitute for a proper discharge estimate, but it is a good first filter.

Longest Flowpath

The longest flowpath (also called the longest watercourse or principal flow path) traces the route from the most hydraulically distant point on the catchment divide to the outlet. This is the path that a raindrop takes the longest time to travel — it is typically along the main channel upstream, then up the largest tributary, then up sub-tributaries until the divide is reached.

The flowpath length $L$ is a key input for estimating the time of concentration $T_c$ — the time for runoff to travel from the most remote point to the outlet — via empirical formulas such as Kirpich, SCS Lag, and SANRAL’s SA-SCS equation. $T_c$ directly controls the design rainfall intensity used in the Rational Method and the unit hydrograph lag used in SCS methods.

$$T_c \;=\; 0.0195 \cdot \frac{L^{0.77}}{S^{0.385}}$$

Kirpich formula for time of concentration (SI form)

where $T_c$ is in minutes, $L$ is flowpath length in metres, and $S$ is the average slope along the flowpath in m/m.

Flowpath slope — average and 10-85

The average slope along the flowpath is:

$$S_\text{avg} \;=\; \frac{\Delta H}{L} \;=\; \frac{H_\text{source} - H_\text{outlet}}{L}$$

Average flowpath slope

where $\Delta H$ is the total elevation drop and $L$ is the flowpath length. This is the simplest and most commonly reported measure.

The 10-85 slope — preferred by SANRAL and widely used in flood-frequency analysis — excludes the most and least steep portions of the flowpath by taking the elevation difference between the 10% and 85% distance points:

$$S_{10\text{-}85} \;=\; \frac{H_{85\%} - H_{10\%}}{0.75 \, L}$$

10-85 slope (USGS / SANRAL)

The 10-85 slope is less sensitive to local anomalies at the headwater (a single steep cliff near the divide) and near the outlet (a flat estuarine reach), and generally gives more realistic $T_c$ estimates for natural catchments.

On the map, the longest flowpath is rendered as a red dashed line. The length, drop, and both slope measures are reported in the results panel, and the full 3-D profile (distance vs. elevation) is available in the Watercourse Profile tool — export the watershed result and open it directly in Watercourse Profile for the downstream workflow.

Export Formats

Results can be downloaded in three GIS-standard formats. All exports are in WGS-84 (EPSG:4326) coordinates.

Format	File extension	Best for	Notes
GeoJSON	.geojson	Web maps, QGIS, programmatic use	Open format, human-readable, single file.
Shapefile	.zip (containing .shp, .dbf, .shx, .prj)	ArcGIS, QGIS, legacy GIS workflows	Zipped archive. 2 GB file-size limit; attribute names truncated to 10 chars.
KML	.kml	Google Earth, 3D visualisations, presentations	XML-based; includes styling and optional elevation.

All three formats include the watershed boundary polygon, the river network (with stream order as an attribute on each segment), and the longest flowpath line in a single download. File sizes range from a few hundred KB for a small urban catchment to a few MB for a large regional basin — simplify-on-export is applied by default.

Downstream workflow

Export the watershed as GeoJSON and open the companion Watercourse Profile tool to visualise the longest-flowpath elevation profile, compute reach-by-reach slopes, and generate inputs for the Rational Method or SCS / Unit Hydrograph flood estimation. The watershed area, longest-flowpath length, and average slope are all that the Rational Tc formula needs.

Use Cases

• Design flood estimation — determine catchment area and longest flowpath as inputs for the Rational Method, SCS, Unit Hydrograph, SDF, or other flood-estimation techniques supported by HydroDesign.
• Culvert and bridge sizing — identify the upstream catchment for a road crossing to estimate design flows for culvert or bridge hydraulic design.
• Environmental and water-quality studies — delineate catchments for sediment transport, nutrient-loading, or ecological-flow analysis.
• Dam-yield and reservoir studies — catchment area is the starting point for MAR estimation, regional yield analysis, and dam-safety evaluation.
• GIS data preparation — export catchment boundaries and river networks in standard GIS formats for further analysis in QGIS, ArcGIS, or other spatial tools.
• Teaching and visualisation — quickly show students or clients the upstream area and drainage network of any point on Earth without setting up a DEM workflow.

Limitations

Every automated delineation tool has failure modes. Be aware of these when interpreting results:

• Resolution (~90 m) — MERIT-Hydro has 3-arcsec resolution. Small catchments (< 1 km²), narrow valleys, and terrace features may not be accurately captured. For detailed site-level work, consider supplementing with a higher-resolution local DEM.
• Flat terrain — in very flat areas (coastal plains, pans, deltas, endorheic basins), the D8 algorithm may produce unexpected flow directions due to the DEM’s limited vertical precision. Delineations in the Kalahari, along the KZN coast, or on the Highveld pans should be sanity-checked against satellite imagery.
• Urban areas and man-made drainage — stormwater pipes, kerb inlets, detention ponds, canals, and road cross-drainage are not represented in the DEM. The delineated boundary reflects natural topographic drainage, not engineered drainage networks. For urban design, always overlay the delineated catchment with the municipal stormwater layout and adjust manually where infrastructure diverts flow between natural sub-catchments.
• Endorheic basins — closed basins that do not drain to the ocean may produce unexpected results, as the MERIT-Hydro conditioning sometimes forces artificial drainage connectivity to an external outlet. Cross-check against the HydroSHEDS endorheic-basin mask for known closed systems.
• Coastal and estuarine outlets — see the outlet-setting section above — move the outlet upstream of tidal influence.
• Large watersheds (> 50,000 km²) — automatically downgraded to low precision; processing may be slow for very large basins.
• API availability — the delineation depends on the external mghydro.com API. If the service is temporarily unavailable, delineation requests will fail — retry after a few minutes, or switch to the ESRI ArcGIS backend.

Common Pitfalls

Three recurring mistakes produce most bad delineations:

1. Outlet on the wrong side of a divide. In flat or rolling terrain the channel on imagery can be ambiguous. Click on an obviously incised channel downstream of your point of interest, confirm the area is sensible, then move upstream step by step to your design location — a sudden jump in area when you cross a subtle divide usually means you’ve hopped catchments.
2. Outlet in a culvert / cross-drainage location. If your design point is where a road crosses a stream, the natural catchment delineation is correct only if the road’s drainage structures respect the natural divide. On urban or peri-urban sites always verify with the municipal stormwater layout.
3. Using the raw boundary for a presentation map. The raster-derived boundary looks staircased at high zoom. Enable Simplify and Beautify for cartographic outputs; leave them off for strict engineering area calculations.

References

• Yamazaki, D., Ikeshima, D., Sosa, J., Bates, P.D., Allen, G.H. & Pavelsky, T.M. (2019). MERIT Hydro: A high-resolution global hydrography map based on latest topography datasets. Water Resources Research, 55(6), 5053 – 5073. doi:10.1029/2019WR024873
• Heberger, M. (2022). Delineator: Fast, accurate global watershed delineation using hybrid vector/raster methods. Zenodo. doi:10.5281/zenodo.7314287
• Jenson, S.K. & Domingue, J.O. (1988). Extracting topographic structure from digital elevation data for geographic information system analysis. Photogrammetric Engineering and Remote Sensing, 54(11), 1593 – 1600. (Foundational D8 flow-direction algorithm.)
• Strahler, A.N. (1957). Quantitative analysis of watershed geomorphology. Transactions, American Geophysical Union, 38(6), 913 – 920. (Original definition of stream ordering.)
• Farr, T.G. et al. (2007). The Shuttle Radar Topography Mission. Reviews of Geophysics, 45, RG2004. (SRTM — the precursor elevation dataset on which MERIT DEM and MERIT-Hydro build.)
• Lehner, B., Verdin, K. & Jarvis, A. (2008). New global hydrography derived from spaceborne elevation data. Eos, Transactions American Geophysical Union, 89(10), 93 – 94. (HydroSHEDS — companion global hydrography dataset used by the ESRI ArcGIS delineation backend.)
• Global Watersheds web application: mghydro.com/watersheds — help documentation: mghydro.com/watersheds/help.html — GitHub: github.com/mheberger/delineator.

---

Open Watershed Delineation