Richland Creek — Daily Analysis: 2026-05-05

Classification: partial Confidence: First meaningful calibration event shows systematic overprediction, but provides valuable data to begin tuning seed coefficients toward observed watershed behavior

Event Summary

The model significantly overpredicted the peak (2.13 ft vs 1.79 ft actual) and mistimed it by nearly 10 hours, suggesting the response coefficients are too high for this rainfall pattern.

Prediction vs Actual

Metric	Predicted	Actual	Error
Peak height	2.13 ft	1.79 ft	+0.34 ft
Total rise	—	0.26 ft	—

Band Contributions

Band	Zone	Precip	Predicted Rise	Intensity	Moisture
1	upper_richland	0.34"	0.34 ft	MODERATE	NORMAL
2	falling_water	0.36"	0.25 ft	MODERATE	NORMAL

Analysis

This event shows a clear overprediction problem with both magnitude and timing. The model predicted a 2.13 ft peak arriving around 4:39 AM on May 6th, but the actual peak of 1.79 ft occurred much earlier at 1:45 PM on May 5th. The 0.34 ft overprediction (19% error) combined with the 9.9-hour timing error suggests the response coefficients are calibrated too aggressively. The rainfall pattern shows two distinct pulses - an early morning burst (0.61" in Band 1, 0.41" in Band 2 around 5 AM) and an evening pulse (0.22" and 0.23" around 8-9 PM). The actual hydrograph peaked after the first pulse, indicating the model may be double-counting or overweighting the rainfall response. The very broad hydrograph shape suggests both bands contributed, but the response was more muted than predicted. Given this is only the first calibrated event with meaningful data, conservative adjustments are warranted to begin tuning the seed coefficients toward reality.

Coefficient Adjustments

Band	Change	Reason
1	-15%	Band 1 received the heaviest rainfall (0.61") but the actual response was much lower than predicted, suggesting the 1.0 ft/inch coefficient is too high
2	-10%	Band 2 showed more balanced rainfall distribution but still contributed to overprediction, warranting a smaller reduction in the 0.7 ft/inch coefficient

Notes

First meaningful calibration event shows systematic overprediction, but provides valuable data to begin tuning seed coefficients toward observed watershed behavior