HydroDiffusion: Diffusion-Based Probabilistic Streamflow Forecasting with a State Space Backbone

TL;DR

HydroDiffusion introduces a diffusion-based probabilistic streamflow forecasting model with a state space backbone, improving temporal coherence and long-range dependency capture over previous LSTM-based models, and demonstrating superior performance across US watersheds.

Contribution

The paper presents HydroDiffusion, a novel diffusion-based framework with a state space model backbone for probabilistic streamflow forecasting, addressing limitations of prior LSTM-based models.

Findings

HydroDiffusion outperforms LSTM-based diffusion models in accuracy.

It maintains performance across full forecast horizons.

It surpasses the DRUM model in probabilistic forecasting skill.

Abstract

Recent advances have introduced diffusion models for probabilistic streamflow forecasting, demonstrating strong early flood-warning skill. However, current implementations rely on recurrent Long Short-Term Memory (LSTM) backbones and single-step training objectives, which limit their ability to capture long-range dependencies and produce coherent forecast trajectories across lead times. To address these limitations, we developed HydroDiffusion, a diffusion-based probabilistic forecasting framework with a decoder-only state space model backbone. The proposed framework jointly denoises full multi-day trajectories in a single pass, ensuring temporal coherence and mitigating error accumulation common in autoregressive prediction. HydroDiffusion is evaluated across 531 watersheds in the contiguous United States (CONUS) in the CAMELS dataset. We benchmark HydroDiffusion against two diffusion baselines with LSTM backbones, as well as the recently proposed Diffusion-based Runoff Model (DRUM). Results show that HydroDiffusion achieves strong nowcast accuracy when driven by observed meteorological forcings, and maintains consistent performance across the full simulation horizon. Moreover, HydroDiffusion delivers stronger deterministic and probabilistic forecast skill than DRUM in operational forecasting. These results establish HydroDiffusion as a robust generative modeling framework for medium-range streamflow forecasting, providing both a new modeling benchmark and a foundation for future research on probabilistic hydrologic prediction at continental scales.