Seasonal design floods reflecting seasonal variation information are very important for reservoir operation and management. The seasonal design flood estimation method currently used in China is based on univariate frequency analysis and assumes that the seasonal and annual design frequencies are equal, which neither satisfies flood prevention standards nor considers the interdependence between different seasonal floods. The Danjiangkou reservoir located in the Han River basin, the first pilot basin of most regulated water resources management policy in China, was selected as a case study. After dividing the entire flood season into subseasons, a dependent structure of summer and autumn floods was established by copula functions. Three bivariate flood quantile selection methods, namely the equivalent frequency combination (EFC) method, conditional expectation combination (CEC) method, and conditional most likely combination (CMLC) method, were performed to estimate unique seasonal design floods to meet the needs in engineering and compared with the univariate design values. The evaluation criteria and a boundary identification method were used to assess the rationality of these combination methods. Compared with the CEC and EFC methods, the CMLC method has smaller root-mean square error (RMSE) and bias values by 12.9–34.8% and 22.6–36.4%, respectively. The CMLC and EFC methods are within the feasible regions, whereas the CEC estimators are beyond the feasible range. The results of economic analysis show that the CMLC method can enhance the floodwater use rate from 79.7 to 82.8% for the wet year and from 91.8 to 93.7% for the dry year in comparison with the EFC method. The CMLC method is more rational in physical realism and recommended for estimating seasonal design floods in the Danjiangkou reservoir, which would provide rich information as references for flood risk assessment, reservoir scheduling, and management.

Standard impoundment operation rules (SIOR) are pre-defined guidelines for refilling reservoirs before the end of the wet season. The advancement and availability of the seasonal flow forecasts provide the opportunity for reservoir operators to use flexible and early impoundment operation rules (EIOR). These flexible impoundment rules can significantly improve water conservation, particularly during dry years. In this study, we investigate the potential application of seasonal streamflow forecasts for employing EIOR in the upper Yangtze River basin. We first define thresholds to determine the streamflow condition in September, which is an important period for decision-making in the basin, and then select the most suitable impoundment operation rules accordingly. The thresholds are used in a simulation–optimization model to evaluate different scenarios for EIOR and SIOR by multiple objectives. We measure the skill of the GloFAS-Seasonal forecast, an operational global seasonal river flow forecasting system, to predict streamflow condition according to the selected thresholds. The results show that: (1) the 20th and 30th percentiles of the historical September flow are suitable thresholds for evaluating the possibility of employing EIOR; (2) compared to climatological forecasts, GloFAS-Seasonal forecasts are skillful for predicting the streamflow condition according to the selected 20th and 30th percentile thresholds; and (3) during dry years, EIOR could improve the fullness storage rate by 5.63% and the annual average hydropower generation by 4.02%, without increasing the risk of flooding. GloFAS-Seasonal forecasts and early reservoir impoundment have the potential to enhance hydropower generation and water utilization.

The controlled outflows from a reservoir are highly dependent on the decisions made by the reservoir operators who mainly rely on available hydrologic information, such as past outflows, reservoir water level and forecasted inflows. In this study, Random Forests (RF) algorithm is used to build reservoir outflow simulation model to evaluate the value of hydrologic information. The Three Gorges Reservoir (TGR) in China is selected as a case study. As input variables of the model, the classic hydrologic information is divided into past, current and future information. Several different simulation models are established based on the combinations of these three groups of information. The influences and value of hydrologic information on reservoir outflow decision-making are evaluated from two different perspectives, the one is the simulation result of different models and the other is the importance ranking of the input variables in RF algorithm. Simulation results demonstrate that the proposed model is able to reasonably simulate outflow decisions of TGR. It is shown that past outflow is the most important information and the forecasted inflows are more important in the flood season than in the non-flood season for reservoir operation decision-making.

The change in water level in the upper reaches of the Yangtze River is of great significance to flood control and navigation. As the first water control station for the mainstream of the Yangtze River after the Minjiang, Hengjiang, Tuojiang River, and other important tributaries flow into the Yangtze River, it is imperative to forecast the water level of Zhutuo Station accurately. The present study utilizes Microsoft Azure's automated machine learning platform (AutoML) and recurrent neural network (RNN) model to predict water levels at Zhutuo Station. The AutoML approach demonstrates certain advantages over RNN methodologies in terms of operability, resource utilization, computational efficiency, and hardware configuration requirements when predicting the water level of Zhutuo. The results show that the future 1-h forecast performance is similar, the mean absolute error (MAE) and root mean square error (RMSE) of the AutoML platform are 0.0098 and 0.012, respectively, and the MAE and RMSE of the RNN model are 0.0088 and 0.011, respectively. The current study's outcomes contribute valuable insights for the real-time monitoring and predictive analytics of water levels, thereby enabling waterway managers to obtain the balance between model complexity and modeling convenience.