Loading gauge…
02AB017 · ON · Drainage area 226 km² (HYDAT)
Loading verification report...
| Stage | Years | KGE |
|---|---|---|
| Calibration | – | – |
| Validation | – | 0.838 |
Checking your access…
The forecast level is the forecast flow converted to water level with a stage-discharge rating curve that FlowAi fits to this station's own record of daily levels and flows (Water Survey of Canada approved and provisional data, last 10 years, recent years weighted more, refitted monthly; current to 2026-10-10). Days with ice, estimated values or failed quality checks are left out of the fit.
Accuracy. The curve reproduces observed daily-mean level to within 1.6 cm (root-mean-square; 2.0 cm on the most recent year held out, 2.5 cm on the highest 10% of flows). It was fitted on levels from 0.41 to 1.93 m (flows 0.1 to 69.9 m3/s). Above that range the curve is extended and the forecast is marked "beyond observed range": error grows quickly and the level is more likely too low than too high.
Datum. Levels are metres above the gauge datum.
Daily mean. The daily forecast is a daily-mean level. The highest level reached during the day can be higher, often by about 10 cm and by 25 cm or more on small, fast rivers. The 6-hourly forecast carries each 6-hour mean flow through the same curve (it matches 5-minute readings within about 2 cm), so its peak is closer to the highest level of the day.
Ice. Winter ice changes the level-flow relationship. No level forecast is shown while the latest three days carry an ice flag or a level with no flow, and a level forecast is withdrawn if the rating stops passing FlowAi's accuracy checks, which are rechecked monthly. The gauge then shows flow only.
Checking your access…
Physical hydrological model (Raven), corrected by a machine learning model (per-gauge XGBoost) trained on the residuals. Where the correction stops before the last day, it fades into Raven over two days. Current validation KGE 0.838 and NSE 0.740 at a 7 day horizon, measured on the record used to calibrate and select this model, so it is not independent evidence: every live gauge clears the gate because clearing it is what makes a gauge live. These are the figures the gauge must keep clearing to stay live, re-checked at each daily revalidation on the current observation record, not a fixed score from the day it was promoted (scored at promotion; not yet re-checked). Over its 5 validation years the model scored KGE 0.684 to 0.792 at 7 days, each year on its own. Operational skill is worse than these figures, because the forecast starts from a modelled river state that is systematically off at issue time. This forecast's correction reads yesterday's observed flow. On a day it has not arrived by the forecast run, the forecast is corrected instead by this gauge's previous corrector, which reads the flow observed two days before and must pass the same daily gate; such a forecast is flagged.
Skill may drop in extreme events outside the calibration range.
Powered by AI-developed physically-based hydrological models
Data sources: Water Survey of Canada, Environment and Climate Change Canada. View attribution details