Two Rivers Catchment: Reading a Changed Hydrograph

Three linked changes are shown: peak discharge rose from 12 to 31 m³/s (about 2.6 times), lag time fell from 6.0 to 2.5 hours, and the rising limb became far steeper.

Why: about 19 km² of 48 km² - roughly 40% of the catchment - was stripped of forest. Trees intercept rain, and their roots and litter keep soil open. Infiltration fell from 22 to 6 mm/h, a 73% loss, so most of a 55 mm storm that once soaked away now runs over the surface. Overland flow travels faster than throughflow, which shortens the lag time.

Compounding factors: 4 km² of roads and settlement are effectively impermeable and drain straight into gullies; the quarry's settling is inadequate, so silt blocks soil pores (lowering infiltration further) and settles in the channel, raising water levels for the same discharge; the 40 gullies deliver water quickly downslope.

Consequence downstream: San Pedro's intake now receives a higher, faster, muddier flood. Higher turbidity raises treatment costs and can force shutdowns, while the sharp peak erodes banks and threatens the 320 farmers' irrigation intakes. The catchment has shifted from a 'sponge' to a 'drainpipe'.

Recommendation: check dams ($2.1m) plus reforestation ($3.2m) = $5.3m, leaving $0.7m for quarry enforcement and monitoring. Do not build the dam now.

Evaluation:

• Reforestation (5 km² of the 19 km² cleared): restores interception and infiltration at the source and improves dry-season baseflow for San Pedro. But it treats only about a quarter of the cleared area and takes 5–10 years to mature, so it will not deliver protection this season.
• Check dams (40): cheap per unit, trap sediment before it reaches the intake, and slow gully flow, cutting the peak. Weaknesses: they fill with silt and need clearing, and they do not fix the cleared slopes above them.
• Flood-control dam ($5.5m): gives the largest single reduction in peak at the outlet and could regulate supply, but it consumes 92% of the budget, displaces land, traps sediment (losing storage capacity), starves downstream of silt, and leaves the cause - 19 km² of bare slope - untouched. A dam also creates a false sense of security that encourages further clearing.

The combination works at two scales: gullies and slopes. The leftover $0.7m should enforce the quarry's discharge permit, since silt is the cheapest problem to fix and undermines both other measures.

The figures are described as storms 'of about 55 mm in three hours', but equal totals do not mean equal storms or equal catchments.

• Antecedent conditions: a storm falling on soil already wet from a previous day will generate far more runoff than the same total on dry soil. We are not told the preceding rainfall for either event.
• Rainfall intensity pattern: 55 mm as a 20-minute cloudburst produces a much sharper peak than 55 mm spread over three hours, even though the totals match.
• The catchment itself changed: by 2023, 19 km² had been cleared, so the two events are being routed through different landscapes. The comparison shows change over time, not a controlled experiment.
• Channel and gauge: siltation raises the stage for a given discharge, and the rating curve used to convert level to m³/s may no longer hold. Backwater from high tide at the outlet could also distort the peak.
• Spatial coverage: one gauge represents 48 km², yet a convective storm may cover only part of it, so the recorded peak depends on where the rain fell relative to the gauge.

The direction of change (higher peak, shorter lag) is credible because it matches the infiltration data, but the exact 2.6-fold figure should be treated as indicative, not precise.