Diagnostic study and modeling of the annual positive water temperature onset

A data-driven model is designed using artificial neural networks (ANN) to predict the average onset for the annual water temperature cycle of North-American streams. The data base is composed of daily water temperature time series recorded at 48 hydrometric stations in Québec (Canada) and northern US, as well as the geographic and physiographic variables extracted from the 48 associated drainage basins. The impact of individual and combined drainage area characteristics on the stream annual temperature cycle starting date is investigated by testing different combinations of input variables. The best model allows to predict the average temperature onset for a site, given its geographical coordinates and vegetation and lake coverage characteristics, with a root mean square error (RMSE) of 5.6 days. The best ANN model was compared favourably with parametric approaches.     

Daily pan evaporation modelling using multi‐layer perceptrons and radial basis neural networks

This paper reports on investigations of the abilities of three different artificial neural network (ANN) techniques, multi‐layer perceptrons (MLP), radial basis neural networks (RBNN) and generalized regression neural networks (GRNN) to estimate daily pan evaporation. Different MLP models comprising various combinations of daily climatic variables, that is, air temperature, solar radiation, wind speed, pressure and humidity were developed to evaluate the effect of each of these variables on pan evaporation. The MLP estimates are compared with those of the RBNN and GRNN techniques. The Stephens‐Stewart (SS) method is also considered for the comparison. The performances of the models are evaluated using root mean square errors (RMSE), mean absolute error (MAE) and determination coefficient (R²) statistics. Based on the comparisons, it was found that the MLP and RBNN computing techniques could be employed successfully to model the evaporation process using the available climatic data. The GRNN was found to perform better than the SS method. Copyright © 2008 John Wiley & Sons, Ltd.     

Choice between competitive pairs of frequency models for use in hydrology: a review and some new results

Abstract

The potential is investigated of the generalized regression neural networks (GRNN) technique in modelling of reference evapotranspiration (ET₀) obtained using the FAO Penman-Monteith (PM) equation. Various combinations of daily climatic data, namely solar radiation, air temperature, relative humidity and wind speed, are used as inputs to the ANN so as to evaluate the degree of effect of each of these variables on ET₀. In the first part of the study, a comparison is made between the estimates provided by the GRNN and those obtained by the Penman, Hargreaves and Ritchie methods as implemented by the California Irrigation Management System (CIMIS). The empirical models were calibrated using the standard FAO PM ET₀ values. The GRNN estimates are also compared with those of the calibrated models. Mean square error, mean absolute error and determination coefficient statistics are used as comparison criteria for the evaluation of the model performances. The GRNN technique (GRNN 1) whose inputs are solar radiation, air temperature, relative humidity and wind speed, gave mean square errors of 0.058 and 0.032 mm² day^?2, mean absolute errors of 0.184 and 0.127 mm day^?1, and determination coefficients of 0.985 and 0.986 for the Pomona and Santa Monica stations (Los Angeles, USA), respectively. Based on the comparisons, it was found that the GRNN 1 model could be employed successfully in modelling the ET₀ process. The second part of the study investigates the potential of the GRNN and the empirical methods in ET₀ estimation using the nearby station data. Among the models, the calibrated Hargreaves was found to perform better than the others.     

The potential of different ANN techniques in evapotranspiration modelling

The potential of three different artificial neural network (ANN) techniques, the multi‐layer perceptrons (MLPs), radial basis neural networks (RBNNs) and generalized regression neural networks (GRNNs), in modelling of reference evapotranspiration (ET₀) is investigated in this paper. Various daily climatic data, that is, solar radiation, air temperature, relative humidity and wind speed from two stations, Pomona and Santa Monica, in Los Angeles, USA, are used as inputs to the ANN techniques so as to estimate ET₀ obtained using the FAO‐56 Penman–Monteith (PM) equation. In the first part of the study, a comparison is made between the estimates provided by the MLP, RBNN and GRNN and those of the following empirical models: The California Irrigation Management Information System (CIMIS) Penman (1985), Hargreaves (1985) and Ritchie (1990). In this part of the study, the empirical models are calibrated using the standard FAO‐56 PM ET₀ values. The estimates of the ANN techniques are also compared with those of the calibrated empirical models. Mean square errors, mean absolute errors and determination coefficient statistics are used as comparing criteria for the evaluation of the models' performances. Based on the comparisons, it is found that the MLP and RBNN techniques could be employed successfully in modelling the ET₀ process. In the second part of the study, the potential of ANN techniques and the empirical methods in ET₀ estimation using nearby station data is investigated. Among the models, the calibrated Hargreaves model is found to perform better than the others. Copyright © 2007 John Wiley & Sons, Ltd.     

Committee neural networks for porosity and permeability prediction from well logs

Neural computing has moved beyond simple demonstration to more significant applications. Encouraged by recent developments in artificial neural network (ANN) modelling techniques, we have developed committee machine (CM) networks for converting well logs to porosity and permeability, and have applied the networks to real well data from the North Sea. Simple three‐layer back‐propagation ANNs constitute the blocks of a modular system where the porosity ANN uses sonic, density and resistivity logs for input. The permeability ANN is slightly more complex, with four inputs (density, gamma ray, neutron porosity and sonic). The optimum size of the hidden layer, the number of training data required, and alternative training techniques have been investigated using synthetic logs. For both networks an optimal number of neurons in the hidden layer is in the range 8–10. With a lower number of hidden units the network fails to represent the problem, and for higher complexity overfitting becomes a problem when data are noisy. A sufficient number of training samples for the porosity ANN is around 150, while the permeability ANN requires twice as many in order to keep network errors well below the errors in core data. For the porosity ANN the overtraining strategy is the suitable technique for bias reduction and an unconstrained optimal linear combination (OLC) is the best method of combining the CM output. For permeability, on the other hand, the combination of overtraining and OLC does not work. Error reduction by validation, simple averaging combined with range‐splitting provides the required accuracy. The accuracy of the resulting CM is restricted only by the accuracy of the real data. The ANN approach is shown to be superior to multiple linear regression techniques even with minor non‐linearity in the background model.     

Evaluation of gravity data by using artificial neural networks case study: Seferihisar geothermal area (Western Turkey)

Artificial neural networks (ANN) have been used in a variety of problems in the fields of science and engineering. Applications of ANN to the geophysical problems have increased within the last decade. In particular, it has been used to solve such inversion problems as seismic, electromagnetic, resistivity. There are also some other applications such as parameter estimation, prediction, and classification. In this study, multilayer perceptron neural networks (MLPNN) and radial basis function neural networks (RBFNN) were applied to synthetic gravity data and Seferihisar gravity data to investigate the applicability and performance of these networks for the method of gravity. Additionally performance of MLPNN and RBFNN were tested by adding random noise to the same synthetic test data. The structure parameters, such as the depths, the density contrasts, and the locations of the structures were obtained closely for different signal-to-noise ratios (S/N). Bouguer data of Seferihisar area were analyzed by MLPNN and RBFNN to estimate depth, density contrast, and location of the structure. The results of MLPNN, RBFNN, and classical inversion method were compared for real data obtained from Seferihisar Geothermal area and similar structure parameters were obtained. The experiments show that in general RBFNN not only increases the speed of the training stage enormously, but also provides slightly better performance.     

Multi-layer perceptrons with Levenberg-Marquardt training algorithm for suspended sediment concentration prediction and estimation / Prévision et estimation de la concentration en matières en suspension avec des perceptrons multi-couches et l’algorithme d’apprentissage de Levenberg-Marquardt

Abstract

Abstract The prediction and estimation of suspended sediment concentration are investigated by using multi-layer perceptrons (MLP). The fastest MLP training algorithm, that is the Levenberg-Marquardt algorithm, is used for optimization of the network weights for data from two stations on the Tongue River in Montana, USA. The first part of the study deals with prediction and estimation of upstream and down-stream station sediment data, separately, and the second part focuses on the estimation of downstream suspended sediment data by using data from both stations. In each case, the MLP test results are compared to those of generalized regression neural networks (GRNN), radial basis function (RBF) and multi-linear regression (MLR) for the best-input combinations. Based on the comparisons, it was found that the MLP generally gives better suspended sediment concentration estimates than the other neural network techniques and the conventional statistical method (MLR). However, for the estimation of maximum sediment peak, the RBF was mostly found to be better than the MLP and the other techniques. The results also indicate that the RBF and GRNN may provide better performance than the MLP in the estimation of the total sediment load.     

Modelling evapotranspiration using discrete wavelet transform and neural networks

This study combines wavelet transforms and feed‐forward neural network methods for reference evapotranspiration estimation. The climatic data (air temperature, solar radiation, wind speed, relative humidity) from two stations in the United States was evaluated for estimating models. For wavelet and neural network (WNN) model, the input data was decomposed into wavelet sub‐time series by wavelet transformation. Later, the new series (reconstructed series) are produced by adding the available wavelet components and these reconstructed series are used as the input of the WNN model. This phase is pre‐processing of raw data and the main different of the WNN model. The performance of the WNN model was compared with classical neural networks approach [artificial neural network (ANN)], multi‐linear regression and Hargreaves empirical method. This study shows that the wavelet transforms and neural network methods could be applied successfully for evapotranspiration modelling from climatic data. Copyright © 2009 John Wiley & Sons, Ltd.     

Adaptive Training of Neural Networks for Automatic Seismic Phase Identification

—?A neural network module has been implemented in the Prototype International Data Centre (PIDC) for automated identification of the initial phase type of seismic detections. Initial training of the neural networks for stations of the International Monitoring System (IMS) requires considerable effort. While there are many seismic phases in the analyst-reviewed database that can be assumed as the ground-truth resource of the initial phase type of Teleseism (T), Regional P (P), and Regional S (S), no ground-truth database of noise (N) is available. To reduce analyst effort required in building a ground-truth database, an “Adaptive Training Approach” is proposed in this paper. This approach automatically selects training patterns to take advantage of the learning ability of neural networks and information on the accumulated observation database. Using this approach, neural networks were trained on the data provided by station STKA, Australia. The performance of automated phase identification has been improved significantly by the retrained neural networks. This approach is also validated by comparison with the performance using the ground-truth noise database.     

Spatial variability of precipitation regimes over Turkey

Abstract

This study aims to predict the daily precipitation from meteorological data from Turkey using the wavelet—neural network method, which combines two methods: discrete wavelet transform (DWT) and artificial neural networks (ANN). The wavelet—ANN model provides a good fit with the observed data, in particular for zero precipitation in the summer months, and for the peaks in the testing period. The results indicate that wavelet—ANN model estimations are significantly superior to those obtained by either a conventional ANN model or a multi linear regression model. In particular, the improvement provided by the new approach in estimating the peak values had a noticeably high positive effect on the performance evaluation criteria. Inclusion of the summed sub-series in the ANN input layer brings a new perspective to the discussions related to the physics involved in the ANN structure.     

Wave data assimilation using a hybrid approach in the Persian Gulf

The main goal of this study is to develop an efficient approach for the assimilation of the hindcasted wave parameters in the Persian Gulf. Hence, the third generation SWAN model was employed for wave modeling forced by the 6-h ECMWF wind data with a resolution of 0.5°. In situ wave measurements at two stations were utilized to evaluate the assimilation approaches. It was found that since the model errors are not the same for wave height and period, adaptation of model parameter does not result in simultaneous and comprehensive improvement of them. Therefore, an approach based on the error prediction and updating of output variables was employed to modify wave height and period. In this approach, artificial neural networks (ANNs) were used to estimate the deviations between the simulated and measured wave parameters. The results showed that updating of output variables leads to significant improvement in a wide range of the predicted wave characteristics. It was revealed that the best input parameters for error prediction networks are mean wind speed, mean wind direction, wind duration, and the wave parameters. In addition, combination of the ANN estimated error with numerically modeled wave parameters leads to further improvement in the predicted wave parameters in contrast to direct estimation of the parameters by ANN.     

An improved neural network approach to the determination of aquifer parameters

In this paper, an artificial neural network (ANN) approach to the determination of aquifer parameters is developed. The approach is based on the combination of an ANN and the Theis solution. The proposed ANN approach has advantages over the existing ANN approach. It avoids inappropriate setting of a trained range. It also determines the aquifer parameters more accurately and needs less required training time. Testing the existing and the proposed ANN approaches by 1000 sets of synthetic data also demonstrates these advantages. As to the comparison between the proposed ANN approach and the type-curve graphical method, an application to actual time-drawdown data shows that the proposed ANN approach determines the aquifer parameters more precisely. The proposed ANN approach is recommended as an alternative to the type-curve graphical method and the existing ANN approach.     

Artificial neural network application to estimate kinematic soil pile interaction response parameters

Six artificial neural network (ANN) models are developed to predict various response parameters of kinematic soil pile interaction. These responses include (1) kinematic response factors for free and fixed head piles in homogenous soil layer to derive foundation input motion (2) normalized bending moment at fixed head of pile in homogenous soil layer (3) normalized kinematic pile moment at the interface of two soil layers of sharply different soil stiffnesses. These ANN models represent simple solutions that can be implemented in a simple calculator capable of matrix operation and bypass the site response analysis and the complex wave diffraction analysis. The data required for ANN training is generated using beam on dynamic Winkler formulation (BDWF). Fifty percent of the data is used to train the ANN models while remaining 50% is used to test the ANN models. The trained ANN models show good agreement with BDWF results.     

On the importance of training methods and ensemble aggregation for runoff prediction by means of artificial neural networks

ABSTRACT

Artificial neural networks (ANNs) become widely used for runoff forecasting in numerous studies. Usually classical gradient-based methods are applied in ANN training and a single ANN model is used. To improve the modelling performance, in some papers ensemble aggregation approaches are used whilst in others, novel training methods are proposed. In this study, the usefulness of both concepts is analysed. First, the applicability of a large number of population-based metaheuristics to ANN training for runoff forecasting is tested on data collected from four catchments, namely upper Annapolis (Nova Scotia, Canada), Biala Tarnowska (Poland), upper Allier (France) and Axe Creek (Victoria, Australia). Then, the importance of the search for novel training methods is compared with the importance of the use of a very simple ANN ensemble aggregation approach. It is shown that although some metaheuristics may slightly outperform the classical gradient-based Levenberg-Marquardt algorithm for a specific catchment, none performs better for the majority of the tested ones. One may also point out a few metaheuristics that do not suit ANN training at all. On the other hand, application of even the simplest ensemble aggregation approach clearly improves the results when the ensemble members are trained by any suitable algorithms.

EDITOR D. Koutsoyiannis; ASSOCIATE EDITOR E. Toth     

Using Artificial Neural Networks to Predict the Presence of Overpressured Zones in the Anadarko Basin,Oklahoma

Many sedimentary basins throughout the world exhibit areas with abnormal pore-fluid pressures (higher or lower than normal or hydrostatic pressure). Predicting pore pressure and other parameters (depth, extension, magnitude, etc.) in such areas are challenging tasks. The compressional acoustic (sonic) log (DT) is often used as a predictor because it responds to changes in porosity or compaction produced by abnormal pore-fluid pressures. Unfortunately, the sonic log is not commonly recorded in most oil and/or gas wells. We propose using an artificial neural network to synthesize sonic logs by identifying the mathematical dependency between DT and the commonly available logs, such as normalized gamma ray (GR) and deep resistivity logs (REID). The artificial neural network process can be divided into three steps: (1) Supervised training of the neural network; (2) confirmation and validation of the model by blind-testing the results in wells that contain both the predictor (GR, REID) and the target values (DT) used in the supervised training; and 3) applying the predictive model to all wells containing the required predictor data and verifying the accuracy of the synthetic DT data by comparing the back-predicted synthetic predictor curves (GRNN, REIDNN) to the recorded predictor curves used in training (GR, REID). Artificial neural networks offer significant advantages over traditional deterministic methods. They do not require a precise mathematical model equation that describes the dependency between the predictor values and the target values and, unlike linear regression techniques, neural network methods do not overpredict mean values and thereby preserve original data variability. One of their most important advantages is that their predictions can be validated and confirmed through back-prediction of the input data. This procedure was applied to predict the presence of overpressured zones in the Anadarko Basin, Oklahoma. The results are promising and encouraging.     

基于主组分分析的概率神经网络损伤定位研究

概率神经网络(PNN)以贝叶斯概率的方法描述测量数据,因而PNN在有噪声条件下的结构损伤检测方面具有巨大潜力。与此同时,PNN的网络规模随着训练样本的增加而增大,这极大地降低了网络运行速度。基于此,本文提出了基于主组分分析(PCA)的PNN损伤定位方法,分别用传统PNN(TPNN)、主组分分析PNN(PCAPNN)和自适应PNN(APNN)三种模型进行了悬索桥的损伤定位研究。研究发现,APNN的识别精度最好,PCAPNN次之,TPNN最差。但APNN需要很长的训练时间,网络规模较大;其他两个网络几乎不需要训练时间,且PCAPNN网络规模较其他两个网络减少了1／3～1／4。在低噪声情况下,PCAPNN的识别效果基本上等同于APNN。     

A hybrid approach of artificial neural network and multiple regression to forecast typhoon rainfall and groundwater-level change

ABSTRACT

A forecasting model is developed using a hybrid approach of artificial neural network (ANN) and multiple regression analysis (MRA) to predict the total typhoon rainfall and groundwater-level change in the Zhuoshui River basin. We used information from the raingauge stations in eastern Taiwan and open source typhoon data to build the ANN model for forecasting the total rainfall and the groundwater level during a typhoon event; then we revised the predictive values using MRA. As a result, the average accuracy improved up to 80% when the hybrid model of ANN and MRA was applied, even where insufficient data were available for model training. The outcome of this research can be applied to forecasts of total rainfall and groundwater-level change before a typhoon event reaches the Zhuoshui River basin once the typhoon has made landfall on the east coast of Taiwan.     

Modelling daily suspended sediment of rivers in Turkey using several data-driven techniques / Modélisation de la charge journalière en matières en suspension dans des rivières turques à l'aide de plusieurs techniques empiriques

Abstract

The transport of sediment load in rivers is important with respect to pollution, channel navigability, reservoir filling, longevity of hydroelectric equipment, fish habitat, river aesthetics and scientific interest. However, conventional sediment rating curves cannot estimate sediment load accurately. An adaptive neuro-fuzzy technique is investigated for its ability to improve the accuracy of the streamflow—suspended sediment rating curve for daily suspended sediment estimation. The daily streamflow and suspended sediment data for four stations in the Black Sea region of Turkey are used as case studies. A comparison is made between the estimates provided by the neuro-fuzzy model and those of the following models: radial basis neural network (RBNN), feed-forward neural network (FFNN), generalized regression neural network (GRNN), multi-linear regression (MLR) and sediment rating curve (SRC). Comparison of results reveals that the neuro-fuzzy model, in general, gives better estimates than the other techniques. Among the neural network techniques, the RBNN is found to perform better than the FFNN and GRNN.     

Predicting and forecasting flow discharge at sites receiving significant lateral inflow

Two models, one linear and one non‐linear, were employed for the prediction of flow discharge hydrographs at sites receiving significant lateral inflow. The linear model is based on a rating curve and permits a quick estimation of flow at a downstream site. The non‐linear model is based on a multilayer feed‐forward back propagation (FFBP) artificial neural network (ANN) and uses flow‐stage data measured at the upstream and downstream stations. ANN predicted the real‐time storm hydrographs satisfactorily and better than did the linear model. The results of sensitivity analysis indicated that when the lateral inflow contribution to the channel reach was insignificant, ANN, using only the flow‐stage data at the upstream station, satisfactorily predicted the hydrograph at the downstream station. The prediction error of ANN increases exponentially with the difference between the peak discharge used in training and that used in testing. ANN was also employed for flood forecasting and was compared with the modified Muskingum model (MMM). For a 4‐h lead time, MMM forecasts the floods reliably but could not be applied to reaches for lead times greater than the wave travel time. Although ANN and MMM had comparable performances for an 8‐h lead time, ANN is capable of forecasting floods with lead times longer than the wave travel time. Copyright © 2007 John Wiley & Sons, Ltd.     

Estimating anisotropic aquifer parameters by artificial neural networks

In recent years, many approaches have been developed using the artificial neural networks (ANN) model incorporated with the Theis analytical solution to estimate the effective hydrological parameters for homogeneous and isotropic porous media, such as the Lin and Chen approach (ANN approach) and the principal component analysis (PCA)‐ANN approach. The above methods assume a full superimposition of the type curve and the observed drawdown and try to use the first time‐drawdown data as a match point to make a fine approximation of the effective parameters. However, using first time‐drawdown data or early time‐drawdown data does not always allow for an accurate estimation of the hydrological parameters, especially for heterogeneous and anisotropic aquifers. Therefore, this article corrects the concept of the superimposed plot by modifying the ANN approach and the PCA‐ANN approach, as well as incorporating the Papadopoulos analytical solution, to estimate the transmissivities and storage coefficient for anisotropic, homogeneous aquifers. The ANN model is trained with 4000 training sets of the well function, and tested with 1000 sets and 300 sets of synthetic time‐drawdown generated from the homogeneous and heterogeneous parameters, respectively. In situ observation data from the time‐drawdown at station Shi‐Chou on the Choushui River alluvial fan, Taiwan, is further adopted to test the applicability and reliability of the proposed methods, as well as provide a basis for comparison with the Straight‐line method and the Type‐curve method. Results suggest that both of the modified methods perform better than the original ones, and using late time‐drawdown to optimize the effective parameters is shown to be better than using early time‐drawdown. Additionally, results indicate that the modified ANN approach is better than the modified PCA‐ANN approach in terms of precision, while the efficiency of the modified PCA‐ANN approach is approximately three times better than that of the modified ANN approach. Copyright © 2010 John Wiley & Sons, Ltd.