⚠️ Under Development — Expect Mistakes and Errors · Pre-Feasibility Draft, October 2026
Notebook
21
InfrastructureModelled

East Berbera Watershed Delineation

45.1027°E, 10.5084°N · Copernicus DEM 30m · Rust/GDAL D8 · A=299 km² · Q₁₀₀=207 m³/s · 1.7× Calinuur · ocean-corrected

299 km²
1.7× Calinuur (180 km²) · ocean-corrected
Catchment area
207 m³/s
0.4× Calinuur (560 m³/s) · EB smaller
Q₁₀₀ flood
149 m³/s
0.4× Calinuur (380 m³/s)
Q₂₅ flood
1,290 m
Golis highland divide
Max elevation
~800 m
Phase 3 est. (similar to CC)
Gallery collector
11 km
east — vs 37km west for CC
Distance Berbera
0 km
trunk goes south direct
T1 shared wayleave
Site identified from Google Earth Pro · 10.508437°N, 45.102742°E · imagery 4 April 2025 · Rust/GDAL analysis, ocean-corrected

The user identified this coastal location east of Berbera from Google Earth, noting a dry riverbed (wadi) visible at the beach. A native Rust watershed tool (berbera-watershed, Priority-Flood + D8) delineated the land-only catchment at 299 km² from the Golis highlands — 1.7× the Calinuur watershed. An earlier pysheds estimate of 2,066 km² was an artefact: the Copernicus GLO-30 DEM stores ocean cells as 0.0 m with no NoData flag, causing uncorrected tools to include coastal sea area in the watershed. The corrected area is smaller than the earlier pysheds figure by 6.9×, and Q₁₀₀ drops from 2,330 m³/s to 207 m³/s — below CC Calinuur's 560 m³/s. This reverses the earlier wadi catchment advantage and changes the MCA result.

DEM Watershed Results — EB vs CC

ParameterEast Berbera (NB21)Calinuur CC (NB19)Ratio / Advantage
Site coordinates45.1027°E, 10.5084°N44.680°E, 10.476°N—
Snapped pour point (Rust/DEM)45.132°E, 10.428°N44.680°E, 10.476°N—
Catchment area (ocean-corrected)299 km²180 km²EB is 1.7× LARGER
Q₂₅ flood discharge149 m³/s380 m³/sCC is 2.6× LARGER
Q₁₀₀ flood discharge207 m³/s560 m³/sCC is 2.7× LARGER
Previous (pysheds, uncorrected)2,066 km² / 2,330 m³/s—Artefact — ocean cells included
Alluvial fan size (est.)~15–20 km² (5–7% catchment)~14 km²Similar fan sizes
Gallery collector length (Ph.3)~800 m est.~800 m est.Similar gallery length
Gallery intake feasibility✓ Feasible — fan size confirms✓ Confirmed at CCBoth feasible
Max catchment elevation1,290 m ASL~2,000 m ASLCC higher — longer wadi
Min catchment elevation65 m ASL~6 m ASL—
Mean catchment elevation386 m ASL~650 m ASLEB lower mean
Relief H1,225 m~1,994 mCC has greater relief
Distance from Berbera city11 km east37 km westEB 3.4× closer
D-Berbera spur length11 km (westward)15 km (eastward)EB −4 km
T1 trunk estimated length~36 km (direct south)42 km (east then south)EB −6 km
T1 shared wayleave needed0 km20 kmEB cleaner topology
Brine disposal (nearshore)Similar — Gulf of Aden coastConfirmed SDI<0.8Similar
Q₁₀₀ flood protection bermSmaller — Q₁₀₀=207 m³/sLarger — Q₁₀₀=560 m³/sEB easier berm
Water quality (SDI)Not measured — clear water visibleConfirmed <0.8CC confirmed

Delineation Method

The watershed was delineated using a native Rust tool (berbera-watershed) built on the gdal crate (GDAL 3.8.4) against the Copernicus GLO-30 DEM (30m, EPSG:4326). The tool runs at 4.7 s on the full regional DEM. A key fix: the Copernicus DEM stores ocean cells as 0.0 m with no NoData flag — an ocean flood-fill from the northern border masks these before any hydrological processing.

1. Ocean maskingFlood-fill from north border: any cell ≤0.5 m reachable from coast is set to NODATA before processing. Removes ~1.9M ocean cells that pysheds included in the watershed.
2. Priority-FloodBarnes (2014) priority-flood fills depressions using a min-heap. Seeds from border cells and NODATA cells at elev−1.0 so coastal cells drain to sea.
3. Flat resolutionBarnes (2014) two-gradient method: BFS h₂ (distance from outlets) and h₁ (distance from high boundaries). Effective elev = base + (2h₂−h₁)×1e⁻⁶.
4. D8 flow direction8-direction steepest descent. NODATA neighbours treated as outlets at −1.0 m so coastal cells always drain outward.
5. Flow accumulationTopological sort (BFS from in-degree=0 cells). 100% cells processed — 0 sinks after ocean masking + Priority-Flood.
6. Pour point snapSearched 0.05° radius around 45.1027°E for highest accumulation cell. Snapped to 45.132°E, 10.428°N (acc = 320,364 cells).
7. Catchment delineationBFS upstream from snapped pour point. Result: 320,364 cells = 299.3 km². Runtime: 4.7 s total.
8. HydrologyKirpich tc = 0.0195 × L(m)⁰·⁷⁷ × S⁻⁰·³⁸⁵. Chen (1983) IDF scaling. Rational method Q = 0.2778 × C × I × A.

Rational Method Hydrology

Same method as NB19 Calinuur analysis — rational method Q = 0.2778 × C × I × A. Same runoff coefficient (C=0.65, arid rocky Golis geology) and same regional IDF parameters (P₁₀₀,₂₄ₕᵣ = 65mm for Berbera coast).

ParameterValueNote
Catchment area A299.3 km²Rust/GDAL D8 — ocean cells masked
Basin length L~24 kmEquivalent circular basin L = √(A/π) × 2.5
Total relief H1,225 mMax 1,290 m − min 65 m (Rust DEM stats)
Concentration time tᶜ2.46 hrKirpich: 0.0195 × L(m)⁰·⁷⁷ × S⁻⁰·³⁸⁵
P(100yr, 24hr)65 mmRegional IDF, Berbera coast — same as NB19
Intensity I(100yr, tᶜ)3.83 mm/hrChen (1983) IDF scaling n=0.477 at tc=2.46hr
Runoff coefficient C0.65Arid rocky catchment — same as Calinuur
Q₂₅149 m³/sP₂₅=47mm → I₂₅=2.76mm/hr
Q₅₀178 m³/sP₅₀=56mm → I₅₀=3.29mm/hr
Q₁₀₀207 m³/sRust/GDAL D8 — 299 km² land-only catchment

EB tc=2.46 hr is shorter than Calinuur (2.46 vs ~3 hr) because the EB basin is smaller and steeper (H/L ratio higher). The lower area (299 vs 180 km²) combined with comparable intensity gives Q₁₀₀=207 m³/s — less than CC's 560 m³/s. EB actually produces a smaller peak flood than CC.

Gallery Intake Feasibility

The alluvial fan at the EB wadi mouth is estimated at ~15–20 km² (5–7% of catchment), formed by 299 km² of highland drainage over geological time — comparable in scale to the Calinuur fan (~14 km²). Both sites should sustain a horizontal gallery intake with similar collector lengths.

Gallery intake feasible — similar to CC Calinuur
  • Phase 1 (45,000 m³/d): ~150m of collector
  • Phase 2 (110,000 m³/d): ~370m of collector
  • Phase 3 (182,500 m³/d): ~610m of collector (2 × 305m arms)
  • Assumed infiltration: 3,000 m³/d per 100m of horizontal collector at 15m depth in medium-coarse alluvial gravel
  • Gallery length similar to CC Calinuur (~800m) — fan size estimate corrected from earlier 165 km² artefact
Field verification needed

Alluvial fan thickness and permeability must be confirmed by GPR (Ground Penetrating Radar) survey and 2–3 test bores before committing to gallery design. The DEM confirms the fan is large enough; bore logs will confirm the saturated thickness and hydraulic conductivity.

Flood protection — EB berm is smaller than CC

Q₁₀₀ = 207 m³/s — significantly less than CC Calinuur's 560 m³/s. The EB flood berm is actually easier to design than CC's, reversing the earlier assessment. Using normal arid-channel stage-discharge relationships (Manning n=0.025, wide channel):

ParameterEast BerberaCalinuur CCDesign basis
Q₁₀₀207 m³/s560 m³/s100-yr return period
Channel width (est.)50–80 m80–120 mProportional to Q⁰·⁵
Flood depth (est.)1.5–2.0 m2.5 mManning backwater calc
Flood berm height2.0–2.5 m ASL2.5 m ASL+1m freeboard above flood depth
FFL (finished floor)3.8–4.3 m ASL4.3 m ASLFFL = berm + 1.8m
Berm cost vs CCSimilar or lessBaselineEB Q₁₀₀ 2.7× smaller than CC

Pipeline Topology Improvement

CC Calinuur — current NB19 topology

T1 trunk starts at 44.680°E and travels 20km east sharing the coastal corridor with D-Berbera before branching at 44.870°E. Only then does it turn south up the Golis escarpment. The shared-wayleave section is necessary because CC is west of Berbera — the trunk must pass through the Berbera zone to reach the escarpment.

Total: 42km trunk (20km coastal + 22km escarpment). 20km shared wayleave.

East Berbera — cleaner topology

T1 trunk starts at 45.103°E and turns directly south up the escarpment — Berbera city is to the west so it's never in the trunk's path. D-Berbera becomes a short 11km westward spur from the plant. No shared-wayleave section needed.

Total: ~36km trunk (direct south). 0km shared wayleave. D-Berbera 11km westward.

Multi-Criteria Analysis — Preliminary Update

NB19 9-criterion MCA updated with Rust/GDAL-confirmed values for EB. The wadi catchment criterion now scores 1.85/5.0 for EB (Q₁₀₀=207 m³/s < CC's 560 m³/s — proportional scoring 5.0×207/560 = 1.85). Geotech improves to 4.5 (EB berm is actually smaller than CC). Distance to Berbera and pipeline complexity both score 5.0. SDI is unconfirmed so EB scores 4.5 vs CC's confirmed 4.8.

CriterionWeightCC ScoreEB ScoreEB vs CCNotes
Water quality (SDI)15%4.84.5−0.3CC confirmed SDI<0.8; EB unconfirmed
Intake type feasibility15%5.04.0−1.0CC confirmed; EB fan ~same size, less tested
Wadi catchment (Q₁₀₀)12%5.01.85−3.15EB Q₁₀₀=207 < CC 560 → score 5×(207/560)=1.85
Distance to Berbera10%2.55.0+2.5EB 11km vs CC 37km — major advantage
Distance to Highland Res10%4.04.5+0.5EB T1 ~36km vs CC 42km
Pipeline complexity10%3.55.0+1.5EB: direct south, 0km shared wayleave
Brine disposal8%4.54.5=Similar Gulf of Aden coast depth
Land access & acquisition10%4.04.5+0.5EB east of city — less urban pressure
Geotech (seismic/flood)10%4.54.5=EB Q₁₀₀=207 m³/s → smaller berm, similar cost
WEIGHTED TOTAL100%4.284.21−0.07CC narrowly leads — wadi criterion reversal
MCA result reversed by corrected hydrology

Calinuur CC now scores 4.28/5.0 vs East Berbera at 4.21/5.0 — a narrow margin of −0.07 points. The reversal is driven entirely by the wadi catchment criterion: correcting the pysheds ocean artefact shows EB Q₁₀₀=207 m³/s, substantially less than CC's 560 m³/s, dropping that score from 5.0 to 1.85. EB's pipeline and proximity advantages are real and significant, but insufficient to compensate at 12% criterion weight. The most critical unknown remains SDI at EB: confirmed SDI<0.8 would restore the intake score from 4.0 to 5.0 (adding 0.15 weighted points, flipping the result to EB=4.36 vs CC=4.28).

Regional context

Toggle satellite to see the coastal wadi mouths east of Berbera. The EB site at 45.103°E is visible as the first significant wadi east of Berbera city. The CC site at 44.680°E is the Calinuur wadi mouth 37km west.

Recommended Next Steps

1
SDI measurement at EB site
Critical

Deploy SDI test kit (Millipore 0.45µm) at 1m depth in nearshore zone 500m offshore during calm conditions. Repeat 3× across the tidal cycle. This is both the binary gate AND the MCA swing criterion: SDI<0.8 → gallery confirmed → intake score rises from 4.0 to 5.0 → EB total becomes 4.36 vs CC 4.28 (EB leads). SDI>2.0 → gallery infeasible → full UF pre-treatment required.

2
Wadi local name + community confirmation
Critical

Confirm the local toponym for the wadi at 45.103°E with Berbera community members. Verify no active quarrying, irrigation extraction, or other uses of the alluvial fan that would affect gallery intake recharge.

3
GPR survey — alluvial fan thickness
High

Ground Penetrating Radar transects across the alluvial fan (3 lines, 500m each). Confirms saturated thickness and depth to bedrock. If <8m saturated, gallery yield per metre reduces and longer collectors needed.

4
T1 escarpment route reconnaissance (east alignment)
High

Walk the new T1 route south from EB site toward Highland Reservoir. The escarpment at 45.1°E may differ significantly from the known route at 44.87°E (Berbera-Sheikh road). Identify passes, cliffs, and any better routing.

5
Update pipeline GeoJSON with EB topology
Medium

Replace CC SWRO coordinates in swro_plant.geojson and pipeline.geojson. Redesign T1 as direct-south from EB, D-Berbera as 11km westward spur. Re-run NB20 DEM validation against new topology.

6
Brine diffuser bathymetry — east of Berbera
Medium

Obtain nearshore bathymetry chart for the coast east of Berbera. Confirm 20m water depth is achievable within 300–500m offshore for the Phase 3 brine diffuser (7.07 m³/s at 70 g/L).

7
Update NB19 MCA to include EB as 5th candidate
Low

Add EB alongside W1, W2, W3, CC in the NB19 multi-criteria analysis with confirmed Q₁₀₀=207 m³/s. EB scores 4.21 overall; CC 4.28. The margin is narrow — SDI confirmation at EB would flip the result. Update radar chart and summary table.

8
Pipeline comparison — EB vs CC topology
Low

EB has shorter pipeline (−6km trunk, −4km D-Berbera vs CC), and flood berm is comparable or smaller (Q₁₀₀=207 vs 560 m³/s). The pipeline and civil work advantage is a meaningful differentiator to include in the feasibility study scope comparison.

0/1000 characters

Your feedback is read by the Berbera planning team. All submissions are treated with respect.