A strong Vietnam streamflow score is not a forecast of next month
WHAT THE EVIDENCE SAYS
A comparison of monthly-flow estimators using chronological validation reported ridge-regression holdout R² of 0.863 and rolling-origin R² of 0.831. Inputs include climate information from the target month and preceding discharge, so the output is within-month estimation rather than advance forecasting. Authors highlight uncertain gauge provenance and limitations in socioeconomic inputs.
WHY IT MATTERS / OUR READ
Skill metrics are decision-relevant only if the model's inputs would be available when managers actually need the answer. Test true forecast lead time and verify discharge measurements before deployment.
CAVEATS / UNCERTAINTY
The flow-gauge reference and transferred drainage-area assumptions introduce uncertainty; the socioeconomic screening is not measured basin-specific yearly change.
SOURCE
Water Science — Pham et al., 29 October 2026
View original evidence ↗
The best model uses weather information from the month whose flow it estimates.
Read the Wild Lands assessment →