Flood similarity identification has important practical significance for floodwater utilization, flood control of reservoirs, and river ecological restoration. In this study, the observations of 125 flood events at the 16 hydrological stations in the middle and upper reaches of the Huai River Basin from 2006 to 2015 were collected, and the metrics of flood magnitude, timing, and rate of changes and patterns were adopted to characterize the entire flood events. Multivariate statistical analysis—principal component analysis and hierarchical clustering method—were adopted to identify the representative flood event types. Finally the temporal and spatial distributions of each flood event type were identified. The results show that: 1) there are five types of representative flood events in the upper and middle reaches of the Huai River Basin, including long duration and extreme variability type, multiple peaks and long duration type, thin and short duration type, fat and short duration type, and conventional type. 2) From the perspective of temporal distribution, the number of flood event types showed a decreasing trend during 2006-2015, and the proportion of conventional floods gradually increased. More flood event types were found in the high flow years (
) and normal flow years (
), and fewer types were in the low flow years (2011-2013) with high frequency of conventional and fat and short duration flood event types. 3) From the perspective of spatial distribution, many flood event types appeared at the source regions, and the flood event types at the middle reaches and downstream regions were relatively few. The flood event type of thin and short duration gradually changed to fat and short duration due to the increased water source conservation capacity, reservoirs' storage capacity, and precipitation diversity in the basin. The study provides some reference for flood information mining and characteristics analysis at the basin scale, and provides scientific foundations for decision makers in flood event analysis, reservoir flood control, and floodwater utilization in the Huai River Basin.
Based on the hydrological data from 2003 to 2014, Long-Short Term Memory (LSTM) was used to construct a daily runoff prediction model for the Ankang discharge station in the upper reaches of the Hanjiang River. The accuracy of daily runoff prediction was evaluated under different input conditions. The result shows that when the foreseeing period is one day, the efficiency coefficient of the LSTM in the calibration period and the validation period can reach 0.68 and 0.74 respectively under the condition that only the previous daily runoff of the Ankang Station is used as input. When the previous areal rainfall of the catchment and the previous daily runoff of the upstream Shiquan Station were added to the LSTM model as input variables, the daily runoff prediction precision was improved. The efficiency coefficient of the training period and the validation period could reach 0.83 and 0.84, respectively. The root mean square error was also significantly reduced. The accuracy of the main flood peak flow forecasting also increased. The LSTM can effectively avoid the problem of over-fitting, and has better generalization performance. However, when the foreseeable period is extended from one day to two or three days, the performance of LSTM is significantly reduced.
The mid- and long-term runoff prediction with satisfactory accuracy plays an important role as basic information in water resources planning & management and optimal operation of water conservancy projects. Combination and bias reduction are two common post-processing approaches in runoff forecast. Applying them in turn, considering the complicated non-stationary and nonlinear characteristics of runoff, a new mid- and long-term runoff prediction method by connecting time-varying weight combination and Bayesian correction is proposed. This method was used to study the annual and monthly inflow prediction of the Longjiang Reservoir in Yunnan Province. The results show that time-varying weight combination balances the performance difference of the established random forest (RF) and support vector machine (SVM) models in the modeling period and the test period. As a consequence of Bayesian correction, the prediction metrics are close to or better than the best of the predictions of the two individual stages. The proportion of correctly classified hydrological year type reaches 77.2% by employing the forecasted annual runoff, and the Nash-Sutcliffe efficiency coefficient of predicted monthly runoff series is over 0.90. Overall, the method put forward in this study has achieved positive effects in improving the forecast performance.
Manning roughness coefficient is the key parameter of flow velocity calculation. Overland flow is significantly different from open channel flow. In this study, we focused on the application of Manning formula in calculating the velocity of overland flow. Compared with open channel flow, the depth of overland flow is very shallow, sometimes only a few millimeters. Thus, vegetation, soil, surface roughness, and other factors have more obvious impact on overland flow. Therefore, the existing open channel flow Manning roughness coefficient cannot be directly used in overland flow. In order to determine the Manning roughness coefficient of overland flow, in this study we developed an indoor experimental system with variable roughness on slope, which includes a water supply system, an experimental tank, and an observation and data recording system. In this system, we used uniform river sand on the flat plate to simulate different roughness of the underlying surface, and placed it in a water tank. The stability and accuracy of the water supply system were verified by 87 pre-experiments. The results show that when the water supply was stable, the discharge was consistent with the data displayed by the electronic flow meter. The 87 groups of weighing data are relatively stable and consistent with normal distribution, and the data are within the 95% confidence interval. Then we designed 166 experiment scenes through a combination of different slopes, surface roughness, and water supply flow to explore the relationship between experimental conditions and Manning roughness coefficient. Among the 166 experiment scenes, a total of six kinds of roughness were designed. The water supply flow ranged from 1 to 25 m 3/h. The slope was between 4°-25°. We used the volume method to calculate the average diameter of the river sand and the chain method to calculate the surface roughness. The experiment data were processed for Support Vector Machine (SVM) training and forecasting, which used root mean square error (RMSE) and coefficient of determination (R 2) as the evaluation indices, considered slope, measured flow, measured depth, average diameter of the river sand, and surface roughness as the independent variables, and Manning roughness coefficient as the dependent variable. The results show that no matter how many kinds of factors were considered, it was difficult to predict the Manning roughness coefficient of laminar flow and transitional flow by the training results of turbulent flow, which indicates a different influence mechanism in different flow patterns. In order to predict the Manning roughness coefficient accurately, we need three factors at least, and measured water depth must be included. When considering four or more factors at the same time, the Manning roughness coefficient could be accurately predicted in turbulent flow.