Classification: no_prediction Confidence: Model confidence remains high due to prior calibration events, but the failure to predict today's significant event is a major gap. The model has seen saturated events and correctly predicted small events, but missed generating any output for this medium-large event.
A significant rainfall event of 0.816 inches produced a substantial peak flow of 792 CFS, but no physics prediction was generated and no empirical headline was issued.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak CFS | N/A | 792 CFS | N/A |
| Total rise | — | 713.9 CFS | — |
| Band | Zone | Precip | Predicted Rise | Intensity |
|---|---|---|---|---|
The physics model failed to generate a prediction for a meaningful hydrological event. Despite 0.816 inches of rainfall occurring in a wet antecedent condition (WET tier, multiplier 1.5), the system did not produce a peak CFS estimate. This is a critical data point because the actual peak reached 792 CFS, crossing the 'medium' threshold (700 CFS). The absence of a prediction suggests either a failure in the input data processing pipeline (the gauge did not register the rainfall input) or a threshold issue where the event was not triggered for calculation.
Historically, the model has struggled with large events (e.g., the false negative on 2026-06-08 where it underestimated a 1520 CFS flood by 63%). While the previous adjustment on 2026-06-13 reduced the coefficient by 15% to correct an overprediction, this 'no prediction' error prevents any calibration of the response coefficient for today's event. Without a predicted value, we cannot calculate an error percentage to guide adjustment. Therefore, conservative protocol dictates maintaining the current coefficient.
The empirical forecast also issued no headline. Given the substantial rise to 792 CFS, an empirical headline targeting 'medium' or 'high' interest would have been warranted. The lack of both physics and empirical output for this significant event highlights a need to review the data ingestion and trigger thresholds for the prediction engine.
No changes made.