Skill and value of seasonal forecasts for reservoir operations in South Korea

Abstract

Reliable mid- and long-term reservoir inflow forecast is essential for the efficient water resources
management and for mitigating water-related disasters such as floods and droughts. However, the
increasing difficulty in predicting future reservoir inflow, attributed to the intensifying effects of
climate change, heightens the risk of operational failures. The severe droughts of the mid-2010s,
which caused significant nationwide damage, highlight the urgent need for improved reservoir
operations in South Korea.
This thesis focuses on the potential of seasonal weather and flow forecasts for improving
reservoir operations and managing droughts in South Korea. Firstly, we compared the
performance of seasonal precipitation forecasts from various forecasting centres and found that
the European Centre for Mid-range Weather Forecasting (ECMWF) provides the most accurate
forecasts in South Korea and particularly during dry years. Secondly, we translated the seasonal
weather forecasts into flow forecasts using a hydrological model then compared their skill against
conventional Ensemble Streamflow Prediction method (ESP). We found that seasonal flow
forecasts outperform ESP up to 3 months ahead and for even longer lead times in dry years.
Lastly, we assessed the value (operational benefits) of using seasonal flow forecasts in practical
reservoir operations during historic drought events. Our findings demonstrated that the forecastinformed operations using ensemble forecasts can help to reduce supply deficit and increase
reservoir storage conservation compared to historical operations during past drought events.
Numerical weather forecasting technologies are continually advancing. Consequently, it is
essential to consistently apply and validate seasonal forecasts in practical reservoir operations.
We hope that our workflow and Python-based open-source toolboxes developed for this thesis
will serve as valuable resources for follow-up research and for broader applications of seasonal
forecasts in water management.

Date of Award	4 Feb 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Miguel A Rico-Ramirez (Supervisor) & Francesca Pianosi (Supervisor)