Lynxce · Post-wildfire source-water screening

A wildfire burned the slopes above a town's water intake.

PWSW turns a fire and an intake into a defensible post-fire monitoring brief. Below is the entire workflow — six steps in the real app — walked through on the North Umpqua / Archie Creek (2020) pilot.

Archie Creek Fire, 2020 → Roseburg intake → Vulnerability: HIGH
Low severity Moderate High Steep & severe (≥23°)
View the PWSW-tool code on GitHub ↗ Open source · Apache-2.0

Why this matters

A storm-driven problem that routine sampling misses.

The pain point, what the tool does about it, and the science that backs it — before you walk the six steps.

The problem

After a wildfire, source-water quality degrades sharply and episodically — turbidity/sediment, DOC & disinfection-byproduct precursors, nutrients, and metals spike on the first post-fire storms.

Response is storm-driven: medians barely move while peaks spike, so routine calendar sampling misses the events that matter. Small, under-resourced utilities and watershed councils face this with limited budgets and no fast, defensible way to prioritize intakes or decide when and where to sample.

What PWSW does

Turns a fire perimeter + an intake or assessment point into a fire-specific monitoring brief — what constituents to expect (prioritized), and where and when to sample — plus an at-a-glance map and GIS layers.

It delineates the contributing basin (USGS NLDI/NHDPlus), overlays best-available burn severity (RAVG→MTBS→LANDFIRE→dNBR), computes physical drivers, and combines them into an Intake Vulnerability Index (IVI, 0–100 relative rank). A screening + monitoring-design aid — not a concentration prediction or a regulatory model. Free, open source (Apache-2.0), runs locally on free US public data.

The science

IVI bands are calibrated to rank-agree with the USGS Staley et al. (2017) “M1” post-fire debris-flow likelihood model — Spearman ρ = 0.853–0.916 across a 12-site PNW/CA panel (config v2.1-phase3). It orders intakes the way an accepted federal model does; it does not reproduce a number.

Inputs are authoritative US federal sources: USGS (NLDI, terrain/pfdf), USFS RAVG + MTBS, NOAA/PRISM precipitation, USDA STATSGO (KFFACT).

Honest limits — screening, not prediction; the headline is a deliberately conservative upper-bound (C=1) band; connectivity uses a hillslope-delivery proxy; PNW design-storm rainfall uses interim NOAA Atlas 2 pending NOAA Atlas 15 (CONUS ~2027).

01PICK

Pick your fire

Type a fire name + year, search the NIFC / WFIGS incident database right on the map, or draw the perimeter yourself. The tool resolves the official burn perimeter — no GIS files to prepare.

Screenshot · the live app
The fire picker: NIFC search plus name and year fields, set to Archie Creek 2020
02INTAKE

Set your intake or assessment point

Click the map with the Place-intake tool, or type a lat / lon. PWSW snaps it to the USGS stream network and delineates the contributing basin upstream of the intake automatically from USGS hydrography (NLDI / NHDPlus).

Screenshot · the live app
The interactive map picker with the intake pin over Roseburg, Oregon
03RUN

Run the screen

One button. PWSW assembles burn severity (RAVG→MTBS→LANDFIRE→dNBR), slope, soil erodibility (KFFACT) and precipitation over the burned upstream basin, then computes an Integrated Vulnerability Index (IVI, 0–100).

Screenshot · the live app
The Run screen button with an honest first-run wait-time notice
The IVI is calibrated to rank-agree with the USGS Staley (2017) M1 post-fire debris-flow model — Spearman ρ ≈ 0.85. Minutes on cached terrain; a cold first run downloads the DEM mosaic (~15 min), and the app streams a live progress checklist.
04MAP

Read the screening map

One map ties it together: the contributing basin, fire perimeter, RAVG burn severity, the steep-and-severe hotspots (slope ≥ 23° ∩ moderate–high severity), tributaries, and the intake.

PWSW screening map of the North Umpqua / Archie Creek basin

The burn (red / purple) sits on steep ground in the upper basin, directly upstream of the Roseburg intake at the western outlet.

05BRIEF

Read the monitoring brief

Validated headlineHIGH
Index scoreIVI 45.8 / 100  ·  bands: Low<25 · Mod 25–30 · High 30–75 · Very High ≥75
Burn severity load
9.6%
HIGH · mod–high burn
Steep & severe ground
4.0%
STRONGEST INDICATOR
Precipitation
1710 mm/yr
HIGH · erosive
Soil erodibility (KFFACT)
0.255
MODERATE

Constituents to expect, keyed to the physical burn drivers:

Turbidity · HIGH DOC / DBP precursors · HIGH Nutrients · MODERATE Metals · MODERATE Cyanotoxins · CONTEXTUAL

Action: stand up event-triggered source-water monitoring focused on the first post-fire storm season — post-fire response is storm-driven, so a fixed calendar misses the peaks.

06EVIDENCE

See the evidence — observed, before vs. after

The screen also pulls the actual monitoring record for the basin (WQP / NWIS) and splits it at fire onset. Medians barely move while peaks spike — the case for event-triggered sampling.

Observed turbidity and DOC at the intake, before vs. after the fire
1310 FNU
post-fire peak at a burned tributary  (~5 NTU is typical)
~160 NTU
at the Roseburg intake downstream
120
monitoring stations found in-basin

The same screen, across the Pacific Northwest

One run per fire — and it discriminates

Four real runs on the current calibration. It is not an always-red alarm: the screen weighs where the burn sits (steep, severe, connected, upstream) — not just how much burned. Retreat and Airplane Lake burned nearly the same basin fraction (~3.5%), yet screen Low vs High.

Four PWSW screening maps across Oregon and Washington with vulnerability badges
FireState / yearContributing basinMod–high burnedIVIScreen
Archie CreekOR 20203,529 km²9.6%45.8HIGH
Airplane LakeWA 2023710 km²3.5%38.1HIGH
PioneerWA 20242,361 km²4.1%34.1HIGH
RetreatWA 20242,515 km²3.5%18.0LOW