Classification: false_negative Confidence: Model confidence remains 'medium' due to 7 events calibrated (threshold for high is 5, but saturation events are rare and this large error introduces uncertainty). The large discrepancy in this saturated event highlights that the model is still tuning to extreme responses.
The physics model drastically underestimated a significant high-tier flood event (1520 CFS) caused by intense rainfall under saturated conditions, predicting only 551.6 CFS with a -63.7% error.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak CFS | 552 CFS | 1520 CFS | -968 CFS |
| Total rise | — | 1297.0 CFS | — |
| Band | Zone | Precip | Predicted Rise | Intensity |
|---|---|---|---|---|
The physics prediction was a severe false negative. The model predicted a peak of 551.6 CFS, while the actual peak reached 1520.0 CFS, resulting in a massive underestimation of 968.4 CFS (-63.7%). This error occurred despite the antecedent moisture tier being correctly identified as SATURATED (multiplier 2.0) and the rainfall intensity likely being HEAVY or INTENSE during the peak hour. The fundamental issue appears to be that the base response coefficient (230.0 CFS/inch) is far too low for this basin under these specific wet/storm conditions. The previous 15% increase was insufficient to capture the runoff efficiency of the steep Boxley drainage when the ground is saturated and rainfall intensity spikes.
Regarding the empirical forecast, no headline was issued for this gauge today. Given the magnitude of the rise (1297 CFS) and the peak reaching well into the 'high' tier (threshold 2000, peak 1520 is medium-high), a significant event occurred. The lack of an empirical forecast suggests the empirical model also failed to trigger or was not applicable, but without a generated headline, we cannot assess its accuracy. The focus must remain on calibrating the physics engine.
To correct this, the response coefficient for Band 1 needs a substantial upward adjustment. While the constraint limits changes to ±20%, applying the maximum +20% increase is necessary to begin converging toward the observed reality. The current coefficient yields ~1048 CFS (1.051 in * 230 * 2.0 wet mod? No, likely intensity mod too). The actual yield was ~1446 CFS/inch (1520/1.05). The model is off by a factor of nearly 3x in effective yield. A +20% increase will raise the coefficient to 276.0 CFS/inch. While this will not fully close the gap, it is the conservative step required by the protocol. We must also check if the saturation tier threshold needs lowering, but adjusting the primary coefficient is the first priority.
| Band | Change | Reason |
|---|---|---|
| 1 | +20% | Severe underestimation (-63.7%) of peak flow during a saturated, high-intensity event. The current coefficient is insufficient to model the rapid runoff response of the steep Boxley basin when fully saturated. Applying the maximum allowed +20% increase to 276.0 CFS/inch. |