Classification: partial Confidence: Model showing improvement in magnitude prediction but timing issues persist; gauge precipitation sensor discrepancy concerning
The model underpredicted the peak by 0.19 ft and mistimed it by nearly 4 hours, despite reasonable rainfall distribution across both bands.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak height | 3.05 ft | 3.24 ft | -0.19 ft |
| Total rise | — | 1.92 ft | — |
| Band | Zone | Precip | Predicted Rise | Intensity | Moisture |
|---|---|---|---|---|---|
This event shows a significant discrepancy between the gauge precipitation sensor (0.0" ground truth) and the QPE data showing substantial rainfall (1.24" Band 1, 1.58" Band 2). The gauge rose 1.92 ft to 3.24 ft with a broad hydrograph shape indicating contributions from both bands, but the prediction was 3.05 ft arriving 4 hours later. The timing error suggests the lag parameters may be too long, while the magnitude underprediction indicates the response coefficients may still be too conservative after the previous 15%/10% reductions. The broad rise pattern and substantial QPE totals support that this was a real rainfall event despite the gauge precipitation sensor reading zero. Given the WET antecedent conditions (2.518" in 7 days) and the intense rainfall rates during the 15:13 hour (0.74"/1.09"), the watershed response appears stronger than the current coefficients predict.
| Band | Change | Reason |
|---|---|---|
| 1 | +8% | Underpredicted response suggests coefficient too low; modest increase to balance previous overcorrection |
| 2 | +12% | Band 2 received heaviest rainfall (1.58") and likely drove the peak; larger increase warranted |
Model showing improvement in magnitude prediction but timing issues persist; gauge precipitation sensor discrepancy concerning