Classification: partial Confidence: The model is performing well on magnitude for dry-day events but shows a persistent tendency to lag in timing predictions. The confidence in the magnitude prediction is high, while confidence in precise timing for minor, broad events remains moderate.
The model accurately predicted the magnitude of the minor gauge rise (1.78 ft vs 1.81 ft) but significantly mispredicted the timing, placing the peak nearly 4 hours late relative to the observed hydrograph.
| Metric | Predicted | Actual | Error |
|---|---|---|---|
| Peak height | 1.78 ft | 1.81 ft | -0.03 ft |
| Total rise | — | 0.85 ft | — |
| Band | Zone | Precip | Predicted Rise | Intensity | Moisture |
|---|---|---|---|---|---|
| 1 | ? | 0.42" | 0.01 ft | MODERATE | DRY |
| 2 | ? | 0.71" | 0.00 ft | HEAVY | DRY |
Headline: No rise indicated by recent rainfall Settled outcome: None (reached None) LLM said headline was correct: True Notes: The empirical model predicted 'No rise indicated by recent rainfall' (target tier: low, p_rise: 0.0027). The actual peak was 1.81 ft, which is below the 'low' threshold of 3.0 ft. The empirical call was correct in indicating that the gauge would not reach the low_floatable threshold.
The event was a minor hydrological response (0.85 ft rise) in a DRY watershed, driven by sparse QPE inputs (0.42" Band 1, 0.71" Band 2) that were not corroborated by the gauge's precipitation sensor (0.0" recorded). The magnitude prediction was excellent, with a -1.7% error, indicating that the current response coefficients and runoff scales are appropriately calibrated for low-magnitude, dry-day events.
However, the timing error of 3.8 hours classifies this event as 'partial'. The observed peak occurred at 18:30, while the predicted peak was issued for 03:20 UTC (roughly 22:20 CDT), suggesting the model overestimated the lag or dispersion for this specific hydrograph shape. The 'very_broad' shape of the hydrograph indicates a slow, distributed response, which may not align perfectly with the current lag assumptions for these bands under DRY conditions. The empirical model correctly predicted 'No rise' given the lack of gauge rainfall and low probability metrics, which is consistent with the actual outcome of a sub-3.0 ft peak.
Predictions issued today (archive): 11 active issuance(s) from 2026-09-12T18:20:02.306880+00:00 to 2026-09-13T04:20:02.315037+00:00; issued peaks ranged 0.96 to 1.78 ft. The prediction graded below is the HIGHEST issued peak of the day (issued 2026-09-13T01:20:02.070649+00:00), not the last card of the day.
Grade-only mode: no adjustments suggested; calibration is fit offline.
The model is performing well on magnitude for dry-day events but shows a persistent tendency to lag in timing predictions. The confidence in the magnitude prediction is high, while confidence in precise timing for minor, broad events remains moderate.