A combined rotated general regression neural network method for river flow forecasting

ABSTRACT

This study focused on the performance of the rotated general regression neural network (RGRNN), as an enhancement of the general regression neural network (GRNN), in monthly-mean river flow forecasting. The study of forecasting of monthly mean river flows in Heihe River, China, was divided into two steps: first, the performance of the RGRNN model was compared with the GRNN model, the feed-forward error back-propagation (FFBP) model and the soil moisture accounting and routing (SMAR) model in their initial model forms; then, by incorporating the corresponding outputs of the SMAR model as an extra input, the combined RGRNN model was compared with the combined FFBP and combined GRNN models. In terms of model efficiency index, R², and normalized root mean squared error, NRMSE, the performances of all three combined models were generally better than those of the four initial models, and the RGRNN model performed better than the GRNN model in both steps, while the FFBP and the SMAR were consistently the worst two models. The results indicate that the combined RGRNN model could be a useful river flow forecasting tool for the chosen arid and semi-arid region in China.

Editor D. Koutsoyiannis; Associate editor not assigned     

Data preprocessing for river flow forecasting using neural networks: Wavelet transforms and data partitioning

The evaluation of surface water resources is a necessary input to solving water management problems. Neural network models have been trained to predict monthly runoff for the Tirso basin, located in Sardinia (Italy) at the S. Chiara section. Monthly time series data were available for 69 years and are characterized by non-stationarity and seasonal irregularity, which is typical of a Mediterranean weather regime. This paper investigates the effects of data preprocessing on model performance using continuous and discrete wavelet transforms and data partitioning. The results showed that networks trained with pre-processed data performed better than networks trained on undecomposed, noisy raw signals. In particular, the best results were obtained using the data partitioning technique.     

Comparing the performance of empirical black‐box models for river flow forecasting in the Heihe River Basin,Northwestern China

For many practical reasons, the empirical black‐box models have become an increasingly popular modelling tool for river flow forecasting, especially in mountainous areas where very few meteorological observatories exist. In this article, precipitation data are used as the only input to estimate river flow. Using five empirical black‐box models—the simple linear model, the linear perturbation model, the linearly varying gain factor model, the constrained nonlinear system model and the nonlinear perturbation model–antecedent precipitation index—modelling results are compared with actual results in three catchments within the Heihe River Basin. The linearly varying gain factor model and the nonlinear perturbation model yielded excellent predictions. For better simulation accuracy, a commonly used multilayer feed‐forward neural network model (NNM) was applied to incorporate the outputs of the individual models. Comparing the performance of these models, it was found that the best results were obtained from the NNM model. The results also suggest that more reliable and precise predictions of river flow can be obtained by using the NNM model while also incorporating the combined outputs of different empirical black‐box models. Copyright © 2012 John Wiley & Sons, Ltd.     

Estimation of reservoir flood control operation risks with considering inflow forecasting errors

A method for quantifying inflow forecasting errors and their impact on reservoir flood control operations is proposed. This approach requires the identification of the probability distributions and uncertainty transfer scheme for the inflow forecasting errors. Accordingly, the probability distributions of the errors are inferred through deducing the relationship between its standard deviation and the forecasting accuracy quantified by the Nash–Sutcliffe efficiency coefficient. The traditional deterministic flood routing process is treated as a diffusion stochastic process. The diffusion coefficient is related to the forecasting accuracy, through which the forecasting errors are indirectly related to the sources of reservoir operation risks. The associated risks are derived by solving the stochastic differential equation of reservoir flood routing via the forward Euler method. The Geheyan reservoir in China is selected as a case study. The hydrological forecasting model for this basin is established and verified. The flood control operation risks in the forecast-based pre-release operation mode for different forecasting accuracies are estimated by the proposed approach. Application results show that the proposed method can provide a useful tool for reservoir operation risk estimation and management.     

M5 model tree application in daily river flow forecasting in Sohu Stream, Turkey

This study investigate the potential of M5 model tree in predicting daily stream flows in Sohu river located within the municipal borders of Ankara, Turkey. The results of the M5 model tree was compared with support vector machines. Both modelling approaches were used to forecast up to 7-day ahead stream flow. A comparison of correlation coefficient and root mean square value indicates that M5 model tree approach works equally well to the SVM for same day discharge prediction. The M5 model tree also works well up to 7-day ahead discharge forecasting in comparison of SVM with this data set. An advantage of using M5 model tree approach is the availability of simple linear models to predict the discharge as well use of less computational time.     

A non-linear perturbation model considering catchment wetness and its application in river flow forecasting

A non-linear perturbation model for river flow forecasting is developed, based on consideration of catchment wetness using an antecedent precipitation index (API). Catchment seasonality, of the form accounted for in the linear perturbation model (the LPM), and non-linear behaviour both in the runoff generation mechanism and in the flow routing processes are represented by a constrained non-linear model, the NLPM-API. A total of ten catchments, across a range of climatic conditions and catchment area magnitudes, located in China and in other countries, were selected for testing daily rainfall-runoff forecasting with this model. It was found that the NLPM-API model was significantly more efficient than the original linear perturbation model (the LPM). However, restriction of explicit non-linearity to the runoff generation process, in the simpler LMP-API form of the model, did not produce a significantly lower value of the efficiency in flood forecasting, in terms of the model efficiency index R².     

River flow time series prediction with a range-dependent neural network

Abstract

Artificial neural networks provide a promising alternative to hydrological time series modelling. However, there are still many fundamental problems requiring further analyses, such as structure identification, parameter estimation, generalization, performance improvement, etc. Based on a proposed clustering algorithm for the training pairs, a new neural network, namely the range-dependent neural network (RDNN) has been developed for better accuracy in hydrological time series prediction. The applicability and potentials of the RDNN in daily streamflow and annual reservoir inflow prediction are examined using data from two watersheds in China. Empirical comparisons of the predictive accuracy, in terms of the model efficiency R² and absolute relative errors (ARE), between the RDNN, back-propagation (BP) networks and the threshold auto-regressive (TAR) model are made. The case studies demonstrated that the RDNN network performed significantly better than the BP network, especially for reproducing low-flow events.     

Application of the Takagi–Sugeno fuzzy system for combination forecasting of river flow in semiarid mountain regions

Because of high spatial heterogeneity and the degree of uncertainty about hydrological processes in large‐scale catchments of semiarid mountain areas, satisfactory forecasting of daily discharge is seldom available using a single model in many practical cases. In this paper the Takagi–Sugeno fuzzy system (TS) and the simple average method (SAM) are applied to combine forecasts of three individual models, namely, the simple linear model (SLM), the seasonally based linear perturbation model (LPM) and the nearest neighbour linear perturbation model (NNLPM) for modelling daily discharge, and the performance of modelling results is compared in five catchments of semiarid areas. It is found that the TS combination model gives good predictions. The results confirm that better prediction accuracy can be obtained by combining the forecasts of different models with the Takagi–Sugeno fuzzy system in semi‐arid mountain areas. Copyright © 2009 John Wiley & Sons, Ltd.     

Using hydrochemical,stable isotope,and river water recharge data to identify groundwater flow paths in a deeply buried karst system

The deeply buried river‐connected Xishan karst aquifer (XKA) in western Beijing, China, has been suffering from diminishing recharge for several decades, which in turn leads to the disappearing of spring water outflows and continuously lowering of groundwater level in the area. Thus, it is important to correctly recognize the groundwater recharge and flow paths for the sustainable development of the XKA. To investigate these issues, the hydrochemical and isotopic compositions are analysed for both surface water and groundwater samples collected over an area of about 280 km². Results show that (a) the river water is characterized by high Na contents; (b) the δ²H and δ¹⁸O values in the river water are distinctively higher than those of groundwater samples, after experiencing the long‐time evaporative enrichment in the upstream reservoir; (c) the Sr concentrations and ⁸⁷Sr/⁸⁶Sr ratios of groundwater clearly indicated the interaction between water and carbonate minerals but excluded the water–silicate interaction; and (d) the groundwater samples in the direct recharge area of the XKA have the lowest Na concentrations and the δ²H and δ¹⁸O values. Based on the large differences in the Na contents and ¹⁸O values of groundwater and surface water, a simple two‐component mixing model is developed for the study area and the fractions of the river water are estimated for groundwater samples. We find that the distribution pattern of the river water fractions in the XKA clearly shows a change of directions in the preferential flow path of the groundwater from its source zone to the discharge area. Overall, our results suggest that the recharged surface water can be a useful evidence for delineating the groundwater flow path in river‐connected karst aquifer. This study improves our understanding of the heterogeneity in karst groundwater systems.     

地震数据处理系统中交互作业编辑与运行

地震勘探数据处理系统的作业编辑与运行是整个系统中重要的部分,本文针对地震交互作业的编辑及运行,对地震数据处理模块及其参数进行了规范和整理分类,运用Linux下的Qt作为GUI开发工具,把负责整个系统调度的主程序与作业编辑程序分开设计,同时运用了数据库技术和动态连接库的技术对模块及其参数进行管理和调用,实现了作业的交互编辑和运行,还提供了地震数据的显示、编辑以及一些参数的拾取,同时还提供了批处理作业和其他交互处理模块的调用接口.为进一步的处理系统开发提供了参考.     

Modeling river/reservoir system management, water allocation, and supply reliability

The state of Texas has implemented a modeling system for assessing the availability and reliability of water resources that consists of a generalized simulation model called the Water Rights Analysis Package (WRAP) and input datasets for the state's 23 river basins. Reservoir/river system management and water allocation practices are simulated using historical naturalized monthly streamflow sequences to represent basin hydrology. Institutional systems for allocating streamflow and reservoir storage resources among numerous water users are considered in detail in evaluating basinwide impacts of water management decisions. The generalized WRAP model is a flexible tool that may be applied to river basins anywhere. The Texas experience in implementing a statewide modeling system illustrates issues that are relevant to water management in many other regions of the world.     

River flow forecasting using wavelet and cross‐wavelet transform models

In this study, short‐term river flood forecasting models based on wavelet and cross‐wavelet constituent components were developed and evaluated for forecasting daily stream flows with lead times equal to 1, 3, and 7 days. These wavelet and cross‐wavelet models were compared with artificial neural network models and simple perseverance models. This was done using data from the Skrwa Prawa River watershed in Poland. Numerical analysis was performed on daily maximum stream flow data from the Parzen station and on meteorological data from the Plock weather station in Poland. Data from 1951 to 1979 was used to train the models while data from 1980 to 1983 was used to test the models. The study showed that forecasting models based on wavelet and cross‐wavelet constituent components can be used with great accuracy as a stand‐alone forecasting method for 1 and 3 days lead time river flood forecasting, assuming that there are no significant trends in the amplitude for the same Julian day year‐to‐year, and that there is a relatively stable phase shift between the flow and meteorological time series. It was also shown that forecasting models based on wavelet and cross‐wavelet constituent components for forecasting river floods are not accurate for longer lead time forecasting such as 7 days, with the artificial neural network models providing more accurate results. Copyright © 2008 John Wiley & Sons, Ltd.     

Climatic control on river discharge simulations,Zackenberg River drainage basin,northeast Greenland

A spatially distributed, physically based, hydrologic modeling system (MIKE SHE) was applied to quantify intra‐ and inter‐annual discharge from the snow and glacierized Zackenberg River drainage basin (512 km²; 20% glacier cover) in northeast Greenland. Evolution of snow accumulation, distribution by wind‐blown snow, blowing‐snow sublimation, and snow and ice surface melt were simulated by a spatially distributed, physically based, snow‐evolution modelling system (SnowModel) and used as input to MIKE SHE. Discharge simulations were performed for three periods 1997–2001 (calibration period), 2001–2005 (validation period), and 2071–2100 (scenario period). The combination of SnowModel and MIKE SHE shows promising results; the timing and magnitude of simulated discharge were generally in accordance with observations (R² = 0·58); however, discrepancies between simulated and observed discharge hydrographs do occur (maximum daily difference up to 44·6 m³ s^?1 and up to 9% difference between observed and simulated cumulative discharge). The model does not perform well when a sudden outburst of glacial dammed water occurs, like the 2005 extreme flood event. The modelling study showed that soil processes related to yearly change in active layer depth and glacial processes (such as changes in yearly glacier area, seasonal changes in the internal glacier drainage system, and the sudden release of glacial bulk water storage) need to be determined, for example, from field studies and incorporated in the models before basin runoff can be quantified more precisely. The SnowModel and MIKE SHE model only include first‐order effects of climate change. For the period 2071–2100, future IPCC A2 and B2 climate scenarios based on the HIRHAM regional climate model and HadCM3 atmosphere–ocean general circulation model simulations indicated a mean annual Zackenberg runoff about 1·5 orders of magnitude greater (around 650 mmWE year^?1) than from today 1997–2005 (around 430 mmWE year^?1), mainly based on changes in negative glacier net mass balance. Copyright © 2007 John Wiley & Sons, Ltd.     

Design and evaluation of SVR,MARS and M5Tree models for 1, 2 and 3-day lead time forecasting of river flow data in a semiarid mountainous catchment

Accurate and reliable river flow forecasts attained with data-intelligent models can provide significant information about future water resources management. In this study we employed a 50-model ensemble of three data-driven predictive models, namely the support vector regression (SVR), multivariate adaptive regression spline (MARS) and M5 model tree (M5Tree) to forecast river flow data in a semiarid and ecologically significant mountainous region of Pailugou catchment in northwestern China. To attain stable and accurate forecast results, 50 different models were trained by randomly sampling the entire river flow data into 80% for training and 20% for testing subsets. To attain a complete evaluation of the ensemble-model based results, the global mean of six quantitative statistical performance evaluation measures: the coefficient of correlation (R), mean absolute relative error (MAE), root mean squared error (RMSE), Nash–Sutcliffe efficiency coefficient (NS), relative RMSE, and the Willmott’s Index (WI), and Taylor diagrams, including skill scores relative to a persistence model, were selected to assess the performances of the developed predictive models. The results indicated that all of the averaged R value attained was higher than 0.900 and all of the averaged NS values were higher than 0.800, representing good performance of the SVR, MARS and M5Tree models applied in the 1-, 2- and 3-day ahead modeling horizon, and this also accorded with the deductions made through an assessment of the Willmott’s Index. However, the M5Tree model outperformed both the SVR and MARS models (with NS?=?0.917 vs. 0.904 and 0.901 for 1-day, 0.893 vs. 0.854 and 0.845 for 2-day, and 0.850 vs. 0.828 and 0.810 for 3-day forecasting horizons, respectively), which was in concurrence with the high value of WI. Therefore, based on the ensemble of 50 models, the performance of the M5Tree can be considered as superior to the SVR and MARS models when applied in a problem of river flow forecasting at multiple forecast horizon. A detailed comparison of the overall performance of all three models evaluated through Taylor diagrams and boxplots indicated that the 1-day ahead forecasting results were more accurate for all of the predictive models compared to the 2- and 3-day ahead forecasting horizons. Data-intelligent models designed in this study indicate that the M5Tree method could successfully be explored for short-term river flow forecasting in semiarid mountainous regions, which may have useful implications in water resources management, ecological sustainability and assessment of river systems.     

SPM response to tide and river flow in the hyper-turbid Ems River

In this paper, we analyse the behaviour of fine sediments in the hyper-turbid Lower Ems River, with focus on the river’s upper reaches, a stretch of about 25 km up-estuary of Terborg. Our analysis is based on long records of suspended particulate matter (SPM) from optical backscatter (OBS) measurements close to the bed at seven stations along the river, records of salinity and water level measurements at these stations, acoustic measurements on the vertical mud structure just up-estuary of Terborg and oxygen profiles in the lower 3 m of the water column close to Leerort and Terborg. Further, we use cross-sectionally averaged velocities computed with a calibrated numerical model. Distinction is made between four timescales, i.e. the semi-diurnal tidal timescale, the spring–neap tidal timescale, a timescale around an isolated peak in river flow (i.e. about 3 weeks) and a seasonal timescale. The data suggest that a pool of fluid/soft mud is present in these upper reaches, from up-estuary of Papenburg to a bit down-estuary of Terborg. Between Terborg and Gandersum, SPM values drop rapidly but remain high at a few gram per litre. The pool of fluid/soft mud is entrained/mobilized at the onset of flood, yielding SPM values of many tens gram per litre. This suspension is transported up-estuary with the flood. Around high water slack, part of the suspension settles, being remixed during ebb, while migrating down-estuary, but likely not much further than Terborg. Around low water slack, a large fraction of the sediment settles, reforming the pool of fluid mud. The rapid entrainment from the fluid mud layer after low water slack is only possible when the peak flood velocity exceeds a critical value of around 1 m/s, i.e. when the stratified water column seems to become internally supercritical. If the peak flood velocity does not reach this critical value, f.i. during neap tide, fluid mud is not entrained up to the OBS sensors. Thus, it is not classical tidal asymmetry, but the peak flood velocity itself which governs the hyper-turbid state in the Lower Ems River. The crucial role of river flow and river floods is in reducing these peak flood velocities. During elongated periods of high river flow, in e.g. wintertime, SPM concentrations reduce, and the soft mud deposits consolidate and possibly become locally armoured as well by sand washed in from the river. We have no observations that sediments are washed out of the hyper-turbid zone. Down-estuary of Terborg, where SPM values do not reach hyper-turbid conditions, the SPM dynamics are governed by classical tidal asymmetry and estuarine circulation. Hence, nowhere in the river, sediments are flushed from the upper reaches of the river into the Ems-Dollard estuary during high river flow events. However, exchange of sediment between river and estuary should occur because of tide-induced dispersion.     

Model averaging versus model selection: estimating design floods with uncertain river flow data

ABSTRACT

This study compares model averaging and model selection methods to estimate design floods, while accounting for the observation error that is typically associated with annual maximum flow data. Model selection refers to methods where a single distribution function is chosen based on prior knowledge or by means of selection criteria. Model averaging refers to methods where the results of multiple distribution functions are combined. Numerical experiments were carried out by generating synthetic data using the Wakeby distribution function as the parent distribution. For this study, comparisons were made in terms of relative error and root mean square error (RMSE) referring to the 1-in-100 year flood. The experiments show that model averaging and model selection methods lead to similar results, especially when short samples are drawn from a highly asymmetric parent. Also, taking an arithmetic average of all design flood estimates gives estimated variances similar to those obtained with more complex weighted model averaging.     

复杂河道洪水预报系统研究——以淮河王家坝至小柳巷区间流域为例

随着社会经济的快速发展,洪水灾害造成的损失日益严重.洪水预报作为一项重要的防洪非工程措施,对防洪、抗洪工作起着至关重要的作用.淮河洪水危害的严重性和洪水演进过程的复杂性使得淮河洪水预报系统的研究长期以来受到高度重视.本文以王家坝至小柳巷区间流域为例,以河道洪水演算为主线,采用新安江三水源模型进行子流域降雨径流预报,概化具有行蓄洪区的干流河道,进行支流与干流、行蓄洪区与干流的洪水汇流耦合计算,采用实时更新的基于多元回归的方法确定水位流量关系,并以上游站点降雨径流预报模型提供的流量作为上边界条件、以下游站点的水位流量关系作为下边界条件,结合行蓄洪调度模型,建立具有行蓄洪区的河道洪水预报系统,再与基于K-最近邻(KNN)的非参数实时校正模型耦合,建立淮河中游河道洪水预报系统.采用多年资料模拟取得了较好的预报效果,并以2003和2007年大洪水为例进行检验,模拟结果精度较高,也证明了所建预报系统的合理性和适用性.     

Satellite-based rainfall estimation for river flow forecasting in Africa. I: Rainfall estimates and hydrological forecasts

Abstract

Reliable, real-time river flow forecasting in Africa on a time scale of days can provide enormous humanitarian and economic benefits. This study investigates the feasibility of using daily rainfall estimates based on cold cloud duration (CCD) derived from Meteosat thermal infrared imagery as input to a simple rainfall—runoff model and also whether such estimates can be improved by the inclusion of information from numerical weather prediction (NWP) analysis models. The Bakoye catchment in Mali, West Africa has been used as a test area. The data available for the study covered the main months of the rainy season for three years. The rainfall estimates were initially validated against gauge data. Improvements in quality were observed when information relating to African Easterly Wave phase and storm type was included in a multiple linear regression (MR) algorithm. The estimates were also tested by using them as input to a rainfall—runoff model. When contemporaneous calibrations from raingauges were available for calibration, both CCD-only and MR rainfall estimates gave more accurate river flow forecasts than when using raingauge data alone. In the absence of contemporaneous calibrations, the performance was reduced but the MR did better than the CCDonly input in all years. The use of satellite-derived vegetation index did not improve the quality of the river flow forecasts.     

Temporal and spatial variability of extreme river flow quantiles in the Upper Vistula River basin,Poland

Temporal and spatial variability in extreme quantile anomalies of seasonal and annual maximum river flows was studied for 41 gauging stations at rivers in the Upper Vistula River basin, Poland. Using the quantile perturbation method, the temporal variability in anomalies was analysed. Interdecadal oscillating components were extracted from the series of anomalies using the Hilbert‐Huang transform method. Period length, part of variance of each component, and part of unexplained variance were assessed. Results show an oscillating pattern in the temporal occurrence of extreme flow quantiles with clusters of high values in the 1960–1970s and since the late 1990s and of low values in the 1980s and at the beginning of the 1990s. The anomalies show a high variability on the right bank of the Upper Vistula River basin during the summer season with the highest values in catchments located in the western and south‐western parts of the basin. River flow extreme quantiles were found to be associated with large‐scale climatic variables from the regions of the North Atlantic Ocean, Scandinavia, Eastern Europe, Asia, and, to a lesser extent, the Pacific Ocean. Similarities between temporal variability of river flows and climatic factors were revealed. Results of the study are important for flood frequency analysis because a long observation period is necessary to capture clusters of high and low river flows.