Streamflow forecasts are traditionally effective in mitigating water scarcity and flood defense. This study developed an artificial intelligence (AI)-based management methodology that integrated multi-step streamflow forecasts and multi-objective reservoir operation optimization for water resource allocation. Following the methodology, we aimed to assess forecast quality and forecast-informed reservoir operation performance together due to the influence of inflow forecast uncertainty. Varying combinations of climate and hydrological variables were input into three AI-based models, namely a long short-term memory (LSTM), a gated recurrent unit (GRU), and a least-squares support vector machine (LSSVM), to forecast short-term streamflow. Based on three deterministic forecasts, the stochastic inflow scenarios were further developed using Bayesian model averaging (BMA) for quantifying uncertainty. The forecasting scheme was further coupled with a multi-reservoir optimization model, and the multi-objective programming was solved using the parameterized multi-objective robust decision-making (MORDM) approach. The AI-based management framework was applied and demonstrated over a multi-reservoir system (25 reservoirs) in the Zhoushan Islands, China. Three main conclusions were drawn from this study: (1) GRU and LSTM performed equally well on streamflow forecasts, and GRU might be the preferred method over LSTM, given that it had simpler structures and less modeling time; (2) higher forecast performance could lead to improved reservoir operation, while uncertain forecasts were more valuable than deterministic forecasts, regarding two performance metrics, i.e., water supply reliability and operating costs; (3) the relationship between the forecast horizon and reservoir operation was complex and depended on the operating configurations (forecast quality and uncertainty) and performance measures. This study reinforces the potential of an AI-based stochastic streamflow forecasting scheme to seek robust strategies under uncertainty.

Baseflow plays a vital role in protecting the environment and ensuring a stable water supply for farming. There are still gaps in the current understanding of baseflow convergence rates in the humid region due to the abundance of rainfall and the high‐water table. Therefore, this study focused on the evolution and hysteresis characteristics of baseflow in humid basins of southeastern China. The baseflow ensemble simulation (BES) method was established to improve the reliability and applicability of baseflow simulation. We suggest a way of differentiating the wet and dry seasons based on the multi‐year average monthly baseflow index (BFI) to determine the intra‐annual distribution of water effectively and simply. The hydrological hysteresis effect of baseflow on precipitation is revealed by characterizing baseflow response to precipitation under precipitation events during wet and dry seasons. A methodology for assessing the performance of baseflow simulation was proposed from observations of streamflow and precipitation. We found that the BES method performed better in baseflow simulation than other single separation methods. Using the BES method, the lag time of baseflow to precipitation during the wet and dry seasons was found to be 3.09 and 4.04 days after utilizing the BFI to divide the hydrological situation into wet and dry seasons. Additionally, precipitation had nearly twice as much intensity influence on baseflow during the dry season compared to the wet season. These findings have significant ramifications for the use, management, and planning of water resources in humid areas of China.

Baseflow estimation and evaluation are two critical and essential tasks for water quality and quantity, drought management, water supply, and groundwater protection. Observed baseflows are rarely available and are limited to focused pilot studies. In this study, an exhaustive evaluation of four different baseflow separation methods (HYSEP, WHAT, BFLOW, and PART) using surrogates of observed baseflows estimated with the conductivity mass balance (CMB) method is carried out using data from several streamflow gauging sites from the South Atlantic-Gulf (SAG) region comprised of nine states in the Southeastern U.S. Daily discharge data from 75 streamflow gauging sites for the period 1970–2013, located in the least anthropogenically affected basins in the SAG region were used to estimate the baseflow index (BFI), which quantifies the contribution of baseflow from streamflows. The focus of this study is to compare the four different baseflow separation methods and calibrate and validate these methods using CMB method based estimates of baseflows to evaluate the variation of BFI values derived from these methods. Results from the study suggest that the PART and HYSEP methods provide the highest and lowest average BFI values of 0.62 and 0.52, respectively. Similarities in BFI values estimated from these methods are noted based on a strong correlation between WHAT and BFLOW. The highest BFI values were found in April in the eastern, western, and central parts of the SAG region, and the highest contribution of baseflow to the streamflow was noted in October in the southern region. However, the lowest BFI values were noted in the month of September in all regions of SAG. The calibrated WHAT method using data from the CMB method provides the highest correlation as noted by the coefficient of determination. This study documents an exhaustive and comprehensive evaluation of baseflow separation methods in the SAG region, and results from this work can aid in the selection of the best method based on different metrics reported in this study. The use of the best method can aid in the short and long term management of low flows at a regional level that supports a sustainable aquatic environment and mitigates the effects of droughts effectively.