Every shape here was carved by the runoff upstream — see what the ground is doing →
Everything in this atlas comes down to one move: cut a line across the valley and measure the ground along it. That cut is a , and its shape answers the question that matters most for stream health — when big water comes, where can it go?
is the water level that just fills the channel — roughly the flow that returns every year or two. It does more work shaping the channel than any other flow, so it’s the reference line for everything else. The is drawn at twice the bankfull depth: a proxy for where moderate floods reach.
Divide the width at flood-prone stage by the width at bankfull and you get the — the single most telling health number in this atlas. Big ratio: floods spread out, slow down, drop their energy gently. Ratio near 1: floods are trapped in the channel, and all that energy goes into the bed and banks.
Apples to apples: both panels drain the same watershed — 2 mi², ~15% impervious, curve number 70, time of concentration 1 hr, a typical small SC Piedmont basin, with 24-hr rainfall from NOAA Atlas 14 (Clemson area) — and they share an identical bankfull channel (~36 ft wide, 3 ft deep). The valleys are idealized but shaped after real benched Piedmont sections like the field-taped Berkeley Orchard reach: the C sits in the broad alluvial valley its letter implies, the F/G in a confined valley with banks about 2.5× bankfull depth — the same bank-height ratio the crew taped in the field. The three rows of houses sit at the same distances from the channel in both panels; what differs is the valley each row was built in. Each panel solves its own water level from the same peak flow (Manning’s equation, 1% slope, channel n 0.045, wooded floodplain n 0.10; drawn with ~10× vertical exaggeration, standard for cross-section plots). The variable on display is capacity: at the 1,000-year storm the broad floodplain passes the whole flow at about 9 ft of stage and wets three houses; the confined, incised valley stacks the same flow past 11 ft and takes all six. The trench itself swallows every storm up to roughly the —-year event — and then fails all at once, onto the stranded floodplain where people built because it never seemed to flood. The slider is spaced by return period (log scale); 500- and 1,000-year marks are extreme illustrations, not design values.
Streams get hurt in a particular sequence. Something changes upstream — more pavement sending flashier floods, a straightened channel, a removed beaver dam — and the stream starts . A step called a forms and migrates upstream like a slow-motion waterfall, lowering the bed as it goes.
Once the bed drops, the old floodplain is stranded above flood reach. The stream is now entrenched: every flood stays in the trench, so the trench erodes wider. Banks collapse, trees fall in, and the channel converts — in the used on these pages, from a connected C or benched B form to an entrenched F or gullied G. Recovery is possible, but it usually takes decades of the stream rebuilding a new, lower floodplain — or a restoration project that reconnects the old one.
George Aull Creek is this story in progress. Three LiDAR flights (2011, 2019, 2026) catch its erosion rate doubling, a headcut chewing upstream, and eight sections converting from B toward F — the atlas’s featured patient.
— laser scanning from aircraft — gives an elevation model of the ground, good to about a foot. We cut sections from it every couple hundred feet along each stream. One honest limit: the laser bounces off water, so wherever a stream was flowing on flight day, the “bed” in the data is actually the water surface, and true depth is understated. On the Berkeley Orchard tributary a crew also taped ten sections by hand — those include the real bed and anchor the LiDAR reaches.
Bankfull size at each section comes from — published relationships between drainage area and channel size. Flood sizes come from USGS regression equations and an NRCS runoff model, cross-checked against each other; the “Hydrology basis” box on every stream page shows every input so the numbers can be independently rebuilt.
Be clear about the hierarchy, though. Everything on this site is screening — the level of analysis you can do before getting wet, so you know where getting wet matters most. An actual restoration design starts where this site stops: a full field survey by a Rosgen-trained professional in the channel — rod and tape on real bankfull indicators, pebble counts on the real bed, sinuosity and slope walked, not flown. Nothing here replaces that day in the stream. Everything here exists to aim it.
All of this is screening-grade desktop analysis meant to target field work, not replace it. FEMA FIS governs regulatory flood values.