← 2026-10-01  |  2026-11-03 →

Cossatot River — Daily Analysis: 2026-10-31

Classification: correct Confidence: Model continues to perform well on quiet days and correctly suppresses false positives under dry conditions. Confidence in dry-state predictions remains high.

Event Summary

The model correctly predicted a quiet day with a peak near baseline, matching the negligible actual rise of 0.07 ft despite low-intensity rainfall.

Prediction vs Actual

Metric Predicted Actual Error
Peak height 2.06 ft 1.89 ft +0.17 ft
Total rise — 0.07 ft —

Band Contributions

Band Precip Predicted Rise Intensity Moisture
1 0.10" 0.00 ft LIGHT VERY DRY
2 0.12" 0.00 ft LIGHT VERY DRY
3 0.11" 0.00 ft LIGHT VERY DRY
4 0.10" 0.00 ft LIGHT VERY DRY
5 0.11" 0.00 ft LIGHT VERY DRY

Empirical Forecast

Headline: No rise indicated by recent rainfall Settled outcome: None (reached None) LLM said headline was correct: True Notes: Empirical headline 'No rise indicated by recent rainfall' was correct as the gauge remained at baseline (1.82-1.89 ft).

Analysis

The event on 2026-10-31 involved very light rainfall (approx. 0.41" total, spread thinly over all bands) on VERY DRY antecedent moisture. The physics-based prediction issued a peak of 2.06 ft, while the actual gauge peak was 1.89 ft. This represents a peak error of 9.0% (0.17 ft), which falls well within the 'correct' threshold of ±15-20%. The model correctly identified that the combination of low intensity and dry ground would not generate a meaningful hydrological response above the baseline.

The empirical forecast also correctly classified the day as 'No rise indicated by recent rainfall,' with a p_rise of 0.0016, which was borne out by the flat hydrograph. The band contributions were all calculated to be near zero, reflecting the initial abstraction and lack of runoff efficiency under VERY DRY conditions. The timing error of 1.8 hours is negligible given the flat nature of the hydrograph (max-min spread only 0.07 ft).

Graded Prediction

Predictions issued today (archive): 18 active issuance(s) from 2026-10-31T11:20:02.340541+00:00 to 2026-11-01T04:20:02.132621+00:00; issued peaks ranged 2.0 to 2.06 ft. The prediction graded below is the HIGHEST issued peak of the day (issued 2026-11-01T02:20:02.332805+00:00), not the last card of the day.

Coefficient Adjustments

Grade-only mode: no adjustments suggested; calibration is fit offline.

Notes

Model continues to perform well on quiet days and correctly suppresses false positives under dry conditions. Confidence in dry-state predictions remains high.

← 2026-10-01  |  2026-11-03 →