Classification: false_positive Confidence: Confidence in Band 1 remains 'high' based on historical event count, but this severe error suggests the underlying coefficient is unstable. Further small-event testing is required after this large correction.
The physics model drastically overpredicted the peak flow for a small 0.8-inch rainfall event, predicting 209.7 CFS when the actual peak was only 12.4 CFS.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak CFS | 210 CFS | 12 CFS | +197 CFS |
| Total rise | — | 7.6 CFS | — |
| Band | Zone | Precip | Predicted Rise | Intensity |
|---|---|---|---|---|
| 1 | boxley | 0.81" | 292 CFS | INTENSE |
The physics prediction was a severe false positive. With 0.8 inches of rain in NORMAL antecedent conditions, the model predicted a peak of 209.7 CFS, whereas the observed peak was just 12.4 CFS. This represents an error of over 1500%, indicating a significant disconnect between the current response coefficient and the actual watershed behavior for small-to-moderate events. The model appears to be treating the rainfall as much more impactful than it actually was, likely due to an inflated response coefficient or incorrect intensity scaling.
The empirical forecast did not issue a headline, which is consistent with the lack of significant rise (only 7.6 CFS above baseflow, well below the 500 CFS low threshold). While the empirical model correctly avoided alerting users to non-existent conditions, the physics engine's failure to suppress the response highlights a critical calibration issue.
Given the massive overprediction, the response coefficient for Band 1 must be reduced. The current coefficient of 258.1 CFS/inch is far too high for this type of event. A reduction is necessary to align predictions with observed low flows during normal moisture conditions. No adjustments to moisture or intensity tiers are needed as the classification logic appeared correct (NORMAL intensity), but the magnitude scaling is fundamentally broken for this event.
| Band | Change | Reason |
|---|---|---|
| 1 | -20% | Severe overprediction (209.7 CFS predicted vs 12.4 CFS actual) requires maximum negative adjustment to lower the response coefficient. |