Daily streamflow prediction with reservoir operation

The booms in reservoir construction mainly take place in emerging economies, including Southeast Asia. The vigorous debate on the reservoir existence leads to the urgent need in more research focusing on their impacts on the hydrological regimes (and thus on the environment and society) and their operation modeling, particularly in real time. Presenting the reservoir operation in hydrological modeling systems is authentically challenging due to the current limitations in mathematical ability and data availability. However, recent advancements in global data and remote sensing as well as potentials in computation and modeling techniques provide an opportunity for the model development and integration, so that we may step closer to the goal. Therefore, the main objective of this PhD thesis is to explore simulation-based reservoir operation models on the basis of maximizing the use of available data (from in-situ to global) and available modeling techniques (process-driven and data-driven). The research context includes the model developments, the model performance examinations and the model applications. The chosen process-driven model is the full-distributed wflow_sbm model. The chosen data-driven models are several machine learning models. The wflow_sbm model was used to investigate the model realism in representing the reservoir operation in the model. It was also applied to quantify effects of the reservoir operation on the water balance, daily streamflow regime and extreme flows. The machine learning models was used to investigate the roles of available reservoir-related data as their inputs to estimate the real-time reservoir outflow, and thus the model accuracy. It was also applied to improve the simulation and multi-step reforecast of the real-time reservoir operation and outflow using global forecast data. This thesis focuses on the real-time operation of major, multi-purpose and over-year storage reservoirs, using the Greater Chao Phraya River Basin in Thailand as a case study to represent a reservoir-dominated basin in the vulnerable region. Findings of this thesis can enhance the understanding of the reservoir operations, their effects, and their modeling for future research and can support reservoir operators in the real-time management of water resources.