Joint distribution of upstream runoff governs downstream river-discharge prediction uncertainty in distributed ML models
2026-07-03 • Machine Learning
Machine Learning
AI summaryⓘ
The authors explain that predicting river flow with uncertainty is important for water management. They show that new machine learning models usually predict flow at one basin outlet but don't work well when predicting flow across many locations in a river network. When trying to predict flow everywhere, the uncertainty from all upstream areas must be considered together, not independently, or else the predictions become too certain. Using data from Japan, they demonstrate a method to keep the uncertainty realistic when spreading predictions across a river system.
Hydrological predictionsUncertainty quantificationProbabilistic modelingLumped modelsDistributed modelsLSTM (Long Short-Term Memory)Runoff routingHayami routing schemeJoint distributionQuantile matching
Authors
Karan Ruparell, Tristan Hascoet, Takemasa Miyoshi, Kieran M. R. Hunt, Hannah L. Cloke, Christel Prudhomme, Florian Pappenberger
Abstract
Uncertainty quantification of hydrological predictions is necessary to inform operational decisions. Recent generative machine-learning methods have advanced probabilistic streamflow prediction, but have remained confined to lumped models that predict a basin outlet directly. At the same time, deterministic LSTM runoff models are increasingly applied at grid or catchment scale and routed through river networks to produce spatially continuous, physically consistent discharge fields. This technical note argues that moving probabilistic prediction from lumped to distributed models introduces a specific new requirement: the joint distribution of upstream runoff generation must be sampled jointly. In lumped inference, the model predicts the outlet distribution directly and can modulate spread from basin attributes. In distributed inference, downstream discharge is obtained by routing many upstream runoff predictions, so independent local sampling averages uncertainty away. Using Japan as a case study, we train two probabilistic basin-scale runoff LSTMs and route their runoff through a Hayami routing scheme. Randomly matching upstream ensemble members produces severely under-dispersed downstream ensembles, whereas a simple quantile matching strategy restores much of the spread of the direct basin-scale reference. The shift from lumped to distributed probabilistic hydrology therefore requires explicit attention to the spatial joint structure of runoff uncertainty.