عنوان مقاله [English]
Daily streamflow prediction is very important for many hydrological applications in providing information for optimal use of water resources. Developing an efficient predictive technique for both long- and short-term streamflow is a challenge in hydrology which is crucial for resource planning and management. This is because streamflow is influenced by various dynamic nonlinear processes, such as rainfall, runoff yield and confluence, evaporation, topography, and anthropic activities. In addition, streamflow forecasting has attracted more attention because of reservoir operations and irrigation management decisions. Over the past decades, researchers have carried out different attempts to forecast daily streamflow. Artificial intelligence modeling has been widely used for streamflow forecasting in recent years because of the availability of long-term gauging data, the ever-increasing computational power.
In this study, according to the nonlinear, random distribution, complex characteristics of hydrological parameters such as streamflow, an integrated method including decomposition technique based on the ensemble empirical mode decomposition (EEMD) combined with model tree (MT) was carried to forecast daily streamflow. To assess the validity of the proposed ensemble EEMD-MT model, a hydrometry gagging station, Gachsar station located on Karaj river, was considered for a 28 years period (1984-2012) at daily scale. Accordingly, a total of 9672 daily streamflow time series dataset given from Gachsar gagging station is employed for developing ensemble EEMD-MT model for daily steamflow forecasting. Among total daily streamflow dataset, 75% as calibration dataset were selected to construct the model and remaining of them were selected as validation dataset. One of the important steps in hydrological molding is to determine the optimum number of time delays from the river flow. In this way, two popular metrics, partial autocorrelation function (PACF) and auto-correlation function (ACF) for the time series dataset were calculated to detect the important input variables which have the highest effect on the target variable for modeling. In this study, for Gachsar station, three antecedent values were selected as the input paramater to simulate daily streamflow. Then, all input and output variables should be decomposed by EEMD into several intrinsic mode functions (IMFs) and one residual. IMFs, then, were modeled by MT model separately and all the forecasted results related to each IMF were aggregated.
MT and adaptive neural-fuzzy inference system (ANFIS) as the benchmark models are compared with an ensemble EEMD-MT model. Several evaluation metrics such as correlation coefficient, root mean square error, relative square error, mean absolute error, and relative absolute error are considered to check the accuracy of the standalone and proposed ensemble models for streamflow forecasting at Gachsar station. The results obtained from the proposed standalone and ensemble models showed that the accuracy of EEMD-MT method compared to MT method increased by about 7.5% and modeling error for this station in the term of RMSE error index decreased by 6.8% in the validation phase. At this station, the MT with the combination of EEMD algorithm has the best accuracy (R = 0.96) in predicting the daily streamflow of Karaj river. According to the scatter plots, MT model shows an under-predicted performance in the validation stage for Gachsar station, although this drawback improved by considering decomposition process of streamflow integrating with EEMD algorithm.
This reflects that a hybrid-based AI approach is a robust and useful tool for simulating daily streamflow over the mountainous region. However, most of the models could not successfully predict the extreme (high and low) flow events. The reason is the developed models are not predominantly trained on extreme events. This can be overcome in the future by the flexibility to incorporate auxiliary information and soft data, such as expert knowledge, into the algorithmic framework, which could provide more flexibility in simulation and assist water managers and dam operators. In summary, this research presents a novel study of testing various AI-based algorithms for streamflow prediction and gives a comprehensive comparison among popular AI methods in hydrologic simulation. The authors believe this unique feature of the AI methods, especially the EEMD-MT algorithm, is to be able to be further employed in the study region and provide more flexibility by adding desired decision variables for reservoir management. However, the proposed methodologies in this study are universally applicable to other mountainous regions, and are flexible to incorporate and test for other hydrologic time series data, such as flood records.