River Flow Forecasting Research Articles

A genetic programming approach to river flow modeling

This paper proposes the application of genetic programming (GP) to forecast monthly river flow. The river flow models were improved by the monthly rainfall and flow data from three stations for Kizilirmak River, Turkey. The coefficient of determination (R2) and root mean square error (RMSE) values were used for evaluating the accuracy of the developed models. The most appropriate GP model was determined as model having monthly flow data of Yamula and Bulakbasi stations according to the model performance criteria for testing data set. The models obtained using the GP were compared with multiple linear regression (MLR) techniques in river flow forecasting. The comparison results revealed that the suggested GP model performs quite well compared to MLR models. It was shown that the suggested GP model with R2 = 0.96 and RMSE = 8.02 m3/s for testing period could be used in planning and management of water resources.

Climate risk management for water in semi–arid regions

Background: New sources of hydroclimate information based on forecast models and observational data have the potential to greatly improve the management of water resources in semi-arid regions prone to drought. Better management of climate-related risks and opportunities requires both new methods to develop forecasts of drought indicators and river flow, as well as better strategies to incorporate these forecasts into drought, river or reservoir management systems. In each case the existing institutional and policy context is key, making a collaborative approach involving stakeholders essential. Methods: This paper describes work done at the IRI over the past decade to develop statistical hydrologic forecast and water allocation models for the semi arid regions of NE Brazil (the “Nordeste”) and central northern Chile based on seasonal climate forecasts. Results: In both locations, downscaled precipitation forecasts based on lagged SST predictors or GCM precipitation forecasts exhibit quite high skill. Spring-summer melt flow in Chile is shown to be highly predictable based on estimates of previous winter precipitation, and moderately predictable up to 6 months in advance using climate forecasts. Retrospective streamflow forecasts here are quite effective in predicting reductions in water rights during dry years. For the multi-use Oros reservoir in NE Brazil, streamflow forecasts have the most potential to optimize water allocations during multi-year low-flow periods, while the potential is higher for smaller reservoirs, relative to demand. Conclusions: This work demonstrates the potential value of seasonal climate forecasting as an integral part of drought early warning and for water allocation decision support systems in semi-arid regions. As human demands for water rise over time this potential is certain to rise in the future.

한강유역 지천홍수예보모형 개발 및 매개변수 최적화

Abstract In this study, comprehensive flood forecasting model for main tributaries in the Han river was developed using the model struc-tures of comprehensive 4 rivers flood forecasting model newly coded in 2012. The main tributaries in the Han river is Joong-RangCheon, Wang-Seok Cheon, Tan Cheon, and An-Yang Cheon. The optimized parameters in the tributaries were proposed as well.In the tributaries, because of the short forecasting time, the model parameters tend to be determined in each basin. Therefore, theoptimized parameters in the newly developed flood forecasting model were proposed. Storage function method is also utilized inthe comprehensive flood forecasting model with the combination of the subsurface flow model which is for calculating infil-tration, percolation, and groundwater flow. Therefore, the newly developed flood forecasting is applicable in the long-term riverflow forecasting. The 2-stage optimization was proposed. In the first stage, parameters in the subsurface flow model were opti-mized using a long-term rainfall-runoff data, and the parameters for the storage function method were optimized in the secondstage using the rainfall events. A range of the optimized parameters was presented, from which the sensitivities and the param-eters could be estimated. However, the understandable interpretation about the ranged parameters was not possible. Therefore, therepresentative parameters were proposed using weighted average, middle value, and representative rainfall events. As a result, theparameter value which provide the smallest error was selected as the final proposed parameter. It could improve the efficiency ofthe runoff forecasting in the practical application.Key words : 4 river comprehensive flood forecasting model, flood forecasting model in main tributaries, storage function method,subsurface flow model, parameter optimization

A comparative study of artificial neural network, adaptive neuro fuzzy inference system and support vector machine for forecasting river flow in the semiarid mountain region

Data driven models are very useful for river flow forecasting when the underlying physical relationships are not fully understand, but it is not clear whether these data driven models still have a good performance in the small river basin of semiarid mountain regions where have complicated topography. In this study, the potential of three different data driven methods, artificial neural network (ANN), adaptive neuro fuzzy inference system (ANFIS) and support vector machine (SVM) were used for forecasting river flow in the semiarid mountain region, northwestern China. The models analyzed different combinations of antecedent river flow values and the appropriate input vector has been selected based on the analysis of residuals. The performance of the ANN, ANFIS and SVM models in training and validation sets are compared with the observed data. The model which consists of three antecedent values of flow has been selected as the best fit model for river flow forecasting. To get more accurate evaluation of the results of ANN, ANFIS and SVM models, the four quantitative standard statistical performance evaluation measures, the coefficient of correlation (R), root mean squared error (RMSE), Nash-Sutcliffe efficiency coefficient (NS) and mean absolute relative error (MARE), were employed to evaluate the performances of various models developed. The results indicate that the performance obtained by ANN, ANFIS and SVM in terms of different evaluation criteria during the training and validation period does not vary substantially; the performance of the ANN, ANFIS and SVM models in river flow forecasting was satisfactory. A detailed comparison of the overall performance indicated that the SVM model performed better than ANN and ANFIS in river flow forecasting for the validation data sets. The results also suggest that ANN, ANFIS and SVM method can be successfully applied to establish river flow with complicated topography forecasting models in the semiarid mountain regions. (C) 2013 Elsevier B.V. All rights reserved.

Improving real time flood forecasting using fuzzy inference system

In order to improve the real time forecasting of foods, this paper proposes a modified Takagi Sugeno (T–S) fuzzy inference system termed as threshold subtractive clustering based Takagi Sugeno (TSC-T–S) fuzzy inference system by introducing the concept of rare and frequent hydrological situations in fuzzy modeling system. The proposed modified fuzzy inference systems provide an option of analyzing and computing cluster centers and membership functions for two different hydrological situations, i.e. low to medium flows (frequent events) as well as high to very high flows (rare events) generally encountered in real time flood forecasting. The methodology has been applied for flood forecasting using the hourly rainfall and river flow data of upper Narmada basin, Central India. The available rainfall–runoff data has been classified in frequent and rare events and suitable TSC-T–S fuzzy model structures have been suggested for better forecasting of river flows. The performance of the model during calibration and validation is evaluated by performance indices such as root mean square error (RMSE), model efficiency and coefficient of correlation (R). In flood forecasting, it is very important to know the performance of flow forecasting model in predicting higher magnitude flows. The above described performance criteria do not express the prediction ability of the model precisely from higher to low flow region. Therefore, a new model performance criterion termed as peak percent threshold statistics (PPTS) is proposed to evaluate the performance of a flood forecasting model. The developed model has been tested for different lead periods using hourly rainfall and discharge data. Further, the proposed fuzzy model results have been compared with artificial neural networks (ANN), ANN models for different classes identified by Self Organizing Map (SOM) and subtractive clustering based Takagi Sugeno fuzzy model (SC-T–S fuzzy model). It has been concluded from the study that the TSC-T–S fuzzy model provide reasonably accurate forecast with sufficient lead-time.

Impact of multi-resolution analysis of artificial intelligence models inputs on multi-step ahead river flow forecasting

Summary In this paper an attempt is made to show that the performance of daily river flow forecasting is improved when data-preprocessing techniques are used with computational intelligence methods. Especially for forecasting longer lead-times, one of the inherent problems in all forecasting methods is that the reliability of forecasting decreases with increasing the lead-time. Therefore wavelet multi-resolution analysis is coupled with artificial neural networks (ANN) and adaptive neuro-fuzzy interface system (ANFIS) which are two promising methods for forecasting nonlinear and non-stationary time series. Different models with a combination of the different input data sets are developed for 1, 2, 3, 4 and 5 days ahead forecasting in Harvey River, Western Australia. Daubechies and Symlet wavelets are used to decompose river flow time series to different levels. Hybrid wavelet neural networks (WNN) models are trained using Levenberg–Marquart (LM) algorithm and the wavelet neuro-fuzzy (WNF) models with subtractive clustering method to find the optimum number of fuzzy rules. Comparing the results with those of the original ANN and ANFIS models indicates that the hybrid models produce significantly better results, especially for the peak values and longer lead-times.

Merging weather radar observations with ground-based measurements of rainfall using an adaptive multiquadric surface fitting algorithm

A system for real-time forecasting of river flow is an essential tool in operational water management. Such systems require well calibrated hydrological and river flow models which can make use of spatially distributed and real-time rainfall observations. Weather radar products provide spatial data on rainfall. However, weather radars sense at certain altitude above ground level and are subject to a large range of error sources. Therefore, these observations often do not correspond to measurements at the ground. Through merging ground-based raingauge observations with the radar rainfall product, often referred to as data merging (or recalibration or rescaling of the radar image), one may force the radar observations to better correspond to the ground-based measurements, without losing the spatial information. In this paper, a methodology is presented to combine radar observations with ground-based measurements of precipitation in near real-time, based on an adaptive multiquadric surface fitting algorithm.

A hybrid artificial intelligence model for river flow forecasting

A hybrid hydrologic estimation model is presented with the aim of performing accurate river flow forecasts without the need of using prior knowledge from the experts in the field. The problem of predicting stream flows is a non-trivial task because the various physical mechanisms governing the river flow dynamics act on a wide range of temporal and spatial scales and almost all the mechanisms involved in the river flow process present some degree of nonlinearity. The proposed system incorporates both statistical and artificial intelligence techniques used at different stages of the reasoning cycle in order to calculate the mean daily water volume forecast of the Salvajina reservoir inflow located at the Department of Cauca, Colombia. The accuracy of the proposed model is compared against other well-known artificial intelligence techniques and several statistical tools previously applied in time series forecasting. The results obtained from the experiments carried out using real data from years 1950 to 2006 demonstrate the superiority of the hybrid system.

Development of a river flow forecast model by coupling a snow accumulation/melt model with a distributed run‐off model

AbstractTo consider the contribution of snowmelt in flood forecast, this paper presents an integrated model, which combines a rigorous snow model with a distributed run‐off model, based on unstructured triangular meshes. Using precipitation, air temperature and wind speed as input data, the integrated model can continuously estimate temporal‐spatial variation of both snow depth and amount of snowmelt and subsequently forecast river flow caused by snowmelt. The result of a long‐term simulation of past floods showed a good match with observations. It is evinced that consideration of snowmelt in a flood forecast can help avoid incorrect prediction. This includes overestimation in the snow accumulation season and/or underestimation in the snowmelt season caused by using precipitation directly as the input data of the run‐off model. A real‐time flood forecasting system has been established for the Hii River basin, Japan, by using the model. This system now operates uninterruptedly to make a forecast 6 h ahead, with an interval of 10 min.

Time Series Forecasting Using Wavelet-Least Squares Support Vector Machines and Wavelet Regression Models for Monthly Stream Flow Data

This study explores the least square support vector and wavelet technique (WLSSVM) in the monthly stream flow forecasting. This is a new hybrid technique. The 30 days periodic predicting statistics used in this study are derived from the subjection of this model to the river flow data of the Jhelum and Chenab rivers. The root mean square error (RMSE), mean absolute error (RME) and correlation (R) statistics are used for evaluating the accuracy of the WLSSVM and WR models. The accuracy of the WLSSVM model is compared with LSSVM, WR and LR models. The two rivers surveyed are in the Republic of Pakistan and cover an area encompassing 39,200 km2 for the Jhelum River and 67,515 km2 for the Chenab River. Using discrete wavelets, the observed data has been decomposed into sub-series. These have then appropriately been used as inputs in the least square support vector machines for forecasting the hydrological variables. The resultant observation from this comparison indicates the WLSSVM is more accurate than the LSSVM, WR and LR models in river flow forecasting.

Forecast of the Maximal Flow for the Caucasus Black Sea Coast Rivers

Methods of the maximum river flow forecasting of the Mzymta, Sochi, Zapadnyi Dagomys, Kuapse, Tuapse and Vulan rivers have been developed in order to prevent the flood risk on the rivers of the Black Sea coast. Approach of sufficiently accurate and efficient forecasting of maximum discharges and water levels with a lead time of one day has been developed on the basis of a hydrological model of snowmelt and rainfall runoff formation. A scheme for the computation of daily critical precipitation amount at the meteorological stations causing the exceeding with a given probability of critical flow rates and water levels during the expected day was proposed. Techniques of dangerous flooding probability computation during the next day and next five days depending on the initial hydro/meteorological data available on the forecast issue date of hydro/meteorological information have been developed. The proposed methods can be implemented in the operational automated flood forecasting system and used for warning of dangerous floods on the rivers of the Black Sea coast.

Water Flow Forecasting and River Simulation for Flood Risk Analysis

The paper presents the state-of-the-art in flood forecasting and simulation applied to a river flood analysis and risk prediction. Different water flow forecasting and river simulation models are analysed. An advanced river flood forecasting and modelling approach developed within the ongoing project INFROM is described. It provides an integrated procedure for river flow forecasting and simulation advanced by integration of different models for improving predictions of the river flood risk outputs. A case study on river flow modelling and simulation for river flood risk analysis is given

Evaporation modeling with multiple linear regression techniques– a review

Evaporation is influenced by number of agro-meteorological parameters and one of the integral components of the hydrological cycle and. Usually, estimates of evaporation are needed in a wide array of problems in agriculture, hydrology, agronomy, forestry and land resources planning, such as water balance computation, irrigation management, crop yield forecasting model, river flow forecasting, ecosystem modeling. Irrigation can substantially increase crop yields, but again the scheduling of the water application is usually based on evaporation estimates. Numerous investigators developed models for estimation of evaporation. The interrelated meteorological factors having a major influence on evaporation have been incorporated into various formulae for estimating evaporation. Unfortunately, reliable estimates of evaporation are extremely difficult to obtain because of complex interactions between the components of the land-plant-atmosphere system. In hot climate, the loss of water by evaporation from rivers, canals and open-water bodies is a vital factor as evaporation takes a significant portion of all water supplies. Even in humid areas, evaporation loss is significant, although the cumulative precipitation tends to mask it due to which it is ordinarily not recognized except during rainless period. Therefore, the need for reliable models for quantifying evaporation losses from increasingly scarce water resources is greater than ever before. Accurate estimation of evaporation is fundamental for effective management of water resources. The evaporation models using MLR techniques is discussed her in details.

Hydrological ensemble prediction systems

Hydrological ensemble prediction systems

Monthly river flow forecasting using artificial neural network and support vector regression models coupled with wavelet transform

Reliable and accurate forecasts of river flow is needed in many water resources planning, design development, operation and maintenance activities. In this study, the relative accuracy of artificial neural network (ANN) and support vector regression (SVR) models coupled with wavelet transform in monthly river flow forecasting is investigated, and compared to regular ANN and SVR models, respectively. The relative performance of regular ANN and SVR models is also compared to each other. For this, monthly river flow data of Kharjegil and Ponel stations in Northern Iran are used. The comparison of the results reveals that both ANN and SVR models coupled with wavelet transform, are able to provide more accurate forecasting results than the regular ANN and SVR models. However, it is found that SVR models coupled with wavelet transform provide better forecasting results than ANN models coupled with wavelet transform. The results also indicate that regular SVR models perform slightly better than regular ANN models.

A hybrid model of self organizing maps and least square support vector machine for river flow forecasting

Abstract. Successful river flow forecasting is a major goal and an essential procedure that is necessary in water resource planning and management. There are many forecasting techniques used for river flow forecasting. This study proposed a hybrid model based on a combination of two methods: Self Organizing Map (SOM) and Least Squares Support Vector Machine (LSSVM) model, referred to as the SOM-LSSVM model for river flow forecasting. The hybrid model uses the SOM algorithm to cluster the entire dataset into several disjointed clusters, where the monthly river flows data with similar input pattern are grouped together from a high dimensional input space onto a low dimensional output layer. By doing this, the data with similar input patterns will be mapped to neighbouring neurons in the SOM's output layer. After the dataset has been decomposed into several disjointed clusters, an individual LSSVM is applied to forecast the river flow. The feasibility of this proposed model is evaluated with respect to the actual river flow data from the Bernam River located in Selangor, Malaysia. The performance of the SOM-LSSVM was compared with other single models such as ARIMA, ANN and LSSVM. The performance of these models was then evaluated using various performance indicators. The experimental results show that the SOM-LSSVM model outperforms the other models and performs better than ANN, LSSVM as well as ARIMA for river flow forecasting. It also indicates that the proposed model can forecast more precisely, and provides a promising alternative technique for river flow forecasting.

Improving reliability of river flow forecasting using neural networks, wavelets and self-organising maps

Neural network (NN) models have gained much attention for river flow forecasting because of their ability to map complex non-linearities. However, the selection of appropriate length of training datasets is crucial and the uncertainty in predictions of the trained NNs with new datasets is a crucial problem. In this study, self-organising maps (SOM) are used to classify the datasets homogeneously and the performance of four types of NN models developed for daily discharge predictions – namely traditional NN, wavelet-based NN (WNN), bootstrap-based NN (BNN) and wavelet-bootstrap-based NN (WBNN) – is analysed for their applicability cluster-wise. SOM classified the training datasets into three clusters (i.e. cluster I, II and III) and the trained SOM is then used to assign testing datasets into these three clusters. Simulation studies show that the WBNN model performs better for the entire testing dataset as well as for values in clusters I and III; for cluster II the performance of BNN model is better compared with others for a 1-day lead time forecasting. Overall, it is found that the proposed methodology can enhance the accuracy and reliability of river flow forecasting.

Towards improved post‐processing of hydrologic forecast ensembles

AbstractForecast ensembles of hydrological and hydrometeorologial variables are prone to various uncertainties arising from climatology, model structure and parameters, and initial conditions at the forecast date. Post‐processing methods are usually applied to adjust the mean and variance of the ensemble without any knowledge about the uncertainty sources. This study initially addresses the drawbacks of a commonly used statistical technique, quantile mapping (QM), in bias correction of hydrologic forecasts. Then, an auxiliary variable, the failure index (γ), is proposed to estimate the ineffectiveness of the post‐processing method based on the agreement of adjusted forecasts with corresponding observations during an analysis period prior to the forecast date. An alternative post‐processor based on copula functions is then introduced such that marginal distributions of observations and model simulations are combined to create a multivariate joint distribution. A set of 2500 hypothetical forecast ensembles with parametric marginal distributions of simulated and observed variables are post‐processed with both QM and the proposed multivariate post‐processor. Deterministic forecast skills show that the proposed copula‐based post‐processing is more effective than the QM method in improving the forecasts. It is found that the performance of QM is highly correlated with the failure index, unlike the multivariate post‐processor. In probabilistic metrics, the proposed multivariate post‐processor generally outperforms QM. Further evaluation of techniques is conducted for river flow forecast of Sprague River basin in southern Oregon. Results show that the multivariate post‐processor performs better than the QM technique; it reduces the ensemble spread and is a more reliable approach for improving the forecast. Copyright © 2012 John Wiley & Sons, Ltd.

Forecasting with prediction intervals for periodic autoregressive moving average models

Periodic autoregressive moving average (PARMA) models are indicated for time series whose mean, variance and covariance function vary with the season. In this study, we develop and implement forecasting procedures for PARMA models. Forecasts are developed using the innovations algorithm, along with an idea of Ansley. A formula for the asymptotic error variance is provided, so that Gaussian prediction intervals can be computed. Finally, an application to monthly river flow forecasting is given, to illustrate the method.

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

AbstractFor 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.