Classification: false_positive Confidence: QPE data quality issues prevent meaningful calibration adjustments - gauge vs QPE discrepancy indicates upstream data problems rather than model coefficient issues
Model significantly overpredicted a 1.01 ft rise, forecasting 4.74 ft peak when actual peak was only 3.1 ft, representing a 52.9% magnitude error.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak height | 4.74 ft | 3.1 ft | +1.64 ft |
| Total rise | — | 1.01 ft | — |
| Band | Precip | Predicted Rise | Intensity | Moisture |
|---|---|---|---|---|
| 4 | 1.73" | 1.28 ft | HEAVY | WET |
| 5 | 1.95" | 0.99 ft | HEAVY | WET |
The model predicted a peak of 4.74 ft but the actual peak was only 3.1 ft, representing a substantial overprediction of 1.64 ft (52.9% error). The QPE data shows significant rainfall across all bands (1.43" Band 1, 1.46" Band 2, 1.66" Band 3, 1.73" Band 4, 1.95" Band 5), but the gauge recorded 0" precipitation, indicating a major QPE overestimation. The very broad hydrograph shape with 9.5-hour rise duration suggests the actual rainfall was much lighter and more dispersed than QPE indicated.
The discrepancy between QPE rainfall totals (1.4-1.9" across bands) and gauge measurement (0") is extreme and points to a fundamental QPE calibration issue rather than model coefficient problems. The WET moisture conditions (1.786" 7-day average) would amplify any rainfall input through the 1.5x multiplier, but the core issue appears to be QPE overestimation during this weather pattern.
Given that the gauge provides ground truth for Band 1 rainfall and recorded zero precipitation while QPE showed 1.43", this represents a clear case where QPE data quality is compromised. Making coefficient adjustments based on faulty input data would degrade model performance for future events with accurate QPE readings.
No changes made.
QPE data quality issues prevent meaningful calibration adjustments - gauge vs QPE discrepancy indicates upstream data problems rather than model coefficient issues