A vegetated riparian buffer removes pollutants from overland runoff mainly two ways: physical filtration/settling (stems and litter slow the sheet flow so suspended sediment drops out) and biogeochemical uptake + denitrification (roots take up dissolved nitrate, and anoxic buffer soils host bacteria that convert it to N₂ gas). Both processes remove a roughly constant *fraction* of the remaining load per metre of travel, which integrates to first-order exponential decay — the standard model used in USDA-NRCS and EPA riparian-buffer design guidance:
C_out / C_in = exp( -k · W / v )
W = buffer width (m)
v = runoff velocity through the buffer (m/s)
k = removal-rate coefficient (1/m), set by vegetation type
W / v = residence time τ — how long a parcel spends inside the buffer
Sediment settles faster than nitrate is denitrified, so k_sediment > k_nitrate for the same vegetation — sediment trapping saturates at a fairly narrow buffer width while nitrate keeps improving out to wider buffers. Denser vegetation raises both k values by slowing flow and adding root/microbial surface area.
- Buffer width — more metres of travel inside the buffer, so more exponential decay of both loads.
- Vegetation density — sets k for sediment and nitrate; dense trees + shrub understory filter and denitrify fastest.
- Slope & rainfall intensity — both raise runoff velocity v. Faster flow means less residence time (τ = W/v), so the *same* buffer removes less — the real trade-off that limits how much a fixed-width buffer can do during an intense storm.
This is why real conservation programs (US NRCS buffer standards, EU agri-environment schemes) specify minimum widths and vegetation type together — a narrow grass strip on a steep slope in heavy rain barely helps, while a wide, densely-vegetated buffer on gentle ground can remove the great majority of both sediment and nitrate before it reaches a stream.