Fabian et al. (2026) Climate-informed streamflow forecasting based on a Bayesian autoregressive exogenous stochastic volatility model and its use for hydrological drought forecasting

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-08-16
• Authors: Pamela Sofia Fabian, Hemie Cho, Yoo‐Geun Ham, Hyun‐Han Kwon
• DOI: 10.1016/j.jhydrol.2026.136255

Research Groups

• Department of Civil Engineering, University of Seoul, Seoul, Republic of Korea
• Water Resources & Environmental Management Division, Korea Institute of Hydrological Survey, Goyang-si, Republic of Korea
• Department of Environmental Management, Graduate School of Environmental Studies, Seoul National University, Seoul, Republic of Korea

Short Summary

This study presents a Bayesian autoregressive exogenous stochastic volatility model for season-ahead streamflow forecasting in the context of South Korea's Soyang Dam. The proposed approach incorporates large-scale climate predictors and explicitly models time-varying variance through stochastic volatility.

Objective

• To investigate the potential of pre-season large-scale climate variables to provide skillful season-ahead forecasts of summer (JJA) streamflow in a monsoon-dominated basin.
• To explore the influence of teleconnections in the North Atlantic region on the Soyang Dam inflow and examine their impact on streamflow patterns.

Study Configuration

• Spatial Scale: The study focuses on the Soyang Dam watershed, which is located in South Korea.
• Temporal Scale: The analysis covers a period of 44 years (1979–2022) with daily dam inflow data obtained from the Water Management Information System (WAMIS).

Methodology and Data

• Models used: Bayesian autoregressive exogenous stochastic volatility model (ARXSV)
• Data sources:
  • Daily dam inflow data from WAMIS
  • Climate state variables, including Sea Surface Temperature (SST) from HadISST and Sea Level Pressure (SLP) from NCEP/NCAR reanalysis data

Main Results

• The proposed ARXSV model outperforms conventional autoregressive models in predicting JJA streamflow.
• The study demonstrates the importance of incorporating large-scale climate predictors, such as North Atlantic SST and SLP, into streamflow forecasting models.
• The results highlight the potential of using Bayesian modeling to quantify uncertainty in parameter estimation and predictive distributions.

Contributions

• This study provides a novel approach to season-ahead streamflow forecasting by integrating large-scale climate information with statistical modeling.
• The proposed framework offers broader utility for supporting season-ahead water-resources decision-making, especially during critical periods such as the summer monsoon season.

Funding

• This research was funded by [project/program name and reference code]

Citation

@article{Fabian2026Climateinformed,
  author = {Fabian, Pamela Sofia and Cho, Hemie and Ham, Yoo‐Geun and Kwon, Hyun‐Han},
  title = {Climate-informed streamflow forecasting based on a Bayesian autoregressive exogenous stochastic volatility model and its use for hydrological drought forecasting},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136255},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136255}
}