WiMac internal lab · live third-party feeds · unverified outputs · not an operational product
WiMac Lab — Resilience Suite
WiMac Infrastructure Advisory · lab.wimac.co.nz · staging · WIP datasets
Live feed
checking…

District readout

observed from NRC telemetry · forecast from MET Norway · ARI against NIWA HIRDS v4
Gauges reporting
waiting for /api/live
Peak observed ARI
last 72 h
Peak forecast ARI
next 120 h
Wettest 24 h ahead
any gauge
Antecedent (k 0.98)
district mean · slope memory

Network Resilience & Detour Tool

the delivered tool, running live below — click any road to close it Open full screen →
◄ Looking back what happened · how rare · what it cost
Looking forward ► what the next five days may do

Event Qualifier — Emergency Works

Observed rainfall from live NRC telemetry tested against real HIRDS v4 design depths — peak ARI by duration and gauge, with the current event tracked as it accumulates toward the 1-in-10 yr qualifying line.

Live: NRC Hilltop hourly totals, per-gauge HIRDS ladder
Solid lines — observed. NRC Hilltop telemetry · NIWA HIRDS v4 (historical, no RCP uplift)
Qualifying standard: 1-in-10 yr (NZTA EW) · damage cost attribution still to come (M3)
Open the Event Qualifier →

Forward Risk Screening

Live MET Norway forecast rainfall screened into flood and slip exposure across the network for the next 120 hours — ranked links with their drivers, and a mechanism toggle between surface flooding and slope instability.

Screening, not prediction — it says where to look, never what will close.
Dashed lines — forecast. MET Norway locationforecast · antecedent API (k 0.90 / 0.98)
Horizon: hourly to +48 h, 6-hourly to +120 h · per-link scores await forecast.json (M2)
Open Forward Screening → Geolocated network map →
Shared engine

Network Resilience & Detour Tool

The routing model every instrument stands on — close any road, route the detour, see what severs and what isolates. Unchanged from the tool delivered July 2026.

Open the tool →

Not yet built

placeholders · each tile is a module in the staged delivery plan · nothing here is live
M4

Slip Risk

Slope instability separated from surface flooding. Long-memory antecedent index drives the colouring and the ranked list; the lag between the two mechanisms is made legible — flood peaks with the rain band, slip keeps climbing for a further day and decays over several.

Next after M2/M3
Needs long-API percentile calibrated against the gauge record, not assumed
M5

Closure & Detour Scenarios

The differentiator. Forecast-red links populate the routing tool's existing closure engine; it routes detours and reports what severs and what isolates. User-initiated by design — a button you press, never an automatic display, because auto-closure would read as a prediction the screening cannot support.

Depends on M2
Needs forecast.json per-link scores joined by lid
M6

Bridge Scour Watch

Scoped as a watch list, not a scour model: bridges on catchments under a severe forecast, cross-referenced with known scour history and the weight-restriction schedule. Hakaru writes its own first entry — load-restricted since May 2026 and already on a flood site.

Data-gated — confirm before scoping
Needs waterway opening, pier form, deck level, founding detail. Deck levels are usually the item that does not exist in usable form.
M1

EW Close-out Export

The week-after-the-storm artefact: rainfall, peak ARI by gauge and duration, antecedent state and event window, formatted for an NZTA Emergency Works application. Reuses the run archive, needs no new analysis, and replaces work the council currently does by hand.

Highest value per hour in the plan
Needs per-run archive accruing (already running) + NZTA EW form template
M3

Evidence Weighting

Converts the product from a model's opinion into what has actually happened here before. Damage register and the 10-year WC141 claim record geocoded to links, activating w_fail so rankings cite the council's own cost history.

Parallel track — non-blocking
Needs road + RP geocoding, 30 m snap, and an unattributed.csv published every build
new

Fordcast

Ford crossings as their own instrument. Each ford keyed to its catchment's live rainfall and antecedent state, reporting passable / marginal / impassable rather than a rainfall depth — the question a grader driver or a school bus operator actually asks.

Scope to confirm
Needs ford inventory with catchment and deck level, plus a stage proxy — river gauge or rainfall-runoff
new

Seismic

The other way this network fails. Shaking intensity against the same road links — GeoNet ShakeMap on an event, liquefaction and slope susceptibility as standing layers, feeding the same closure engine so a quake scenario routes detours exactly as a storm does.

Scope to confirm
Needs GeoNet feed choice (ShakeMap vs FDSN), susceptibility source, and whether this is screening or post-event triage
M2

Per-link forecast scores

The join that turns the map from illustrative to real: forecast.json v0.1, 29 timesteps, flood and slip scores on 1,200–1,600 selected links, published as a small separate file so the base tool stays cached and the forecast refreshes alone.

In progress — the current blocker
Needs link selection tiers A–C and the governing-duration rule per spec §7.1
Screening tools — not a warning service. These are network-scale screening indicators to help WiMac decide where and when to look. They are not hydraulic or flood models, carry no service-level guarantee, and do not replace official advice. MetService remains the authoritative source for weather warnings; NZTA determines Emergency Works eligibility. Every output requires manual review before any operational decision. HIRDS v4 depths are used here for development and testing; the licence position for a hosted commercial service is unresolved.
Environment lab · WiMac internal only Data MET Norway / yr.no (CC BY 4.0) · NRC Hilltop · NIWA HIRDS v4 (CC BY) · LINZ (CC BY 4.0) · © OpenStreetMap contributors Contact willis@wimac.co.nz