Approximating SWAT Model Using Artificial Neural Network and Support Vector Machine1

Abstract: With the popularity of complex, physically based hydrologic models, the time consumed for running these models is increasing substantially. Using surrogate models to approximate the computationally intensive models is a promising method to save huge amounts of time for parameter estimation. In this study, two learning machines [Artificial Neural Network (ANN) and support vector machine (SVM)] were evaluated and compared for approximating the Soil and Water Assessment Tool (SWAT) model. These two learning machines were tested in two watersheds (Little River Experimental Watershed in Georgia and Mahatango Creek Experimental Watershed in Pennsylvania). The results show that SVM in general exhibited better generalization ability than ANN. In order to effectively and efficiently apply SVM to approximate SWAT, the effect of cross‐validation schemes, parameter dimensions, and training sample sizes on the performance of SVM was evaluated and discussed. It is suggested that 3‐fold cross‐validation is adequate for training the SVM model, and reducing the parameter dimension through determining the parameter values from field data and the sensitivity analysis is an effective means of improving the performance of SVM. As far as the training sample size, it is difficult to determine the appropriate number of samples for training SVM based on the test results obtained in this study. Simple examples were used to illustrate the potential applicability of combining the SVM model with uncertainty analysis algorithm to save efforts for parameter uncertainty of SWAT. In the future, evaluating the applicability of SVM for approximating SWAT in other watersheds and combining SVM with different parameter uncertainty analysis algorithms and evolutionary optimization algorithms deserve further research.     

滇池水污染物浓度预测的人工神经网络模型

本文针对滇池日益严重的水污染现状,根据云南昆明西苑隧道断面2004年-2010年的监测资料,建立了基于BP神经网络的主要污染指标预测模型,并对其进行训练检验,研究结果表明：独立样本中pH、溶解氧、氨氮、高锰酸盐浓度的预测值与监测值的线性相关系数分别为0．952、0．967、0．945、0．936。结果证明该模型预测精度满足要求,通过准确地预测湖泊水污染物可以为治理湖泊营养化和综合利用水资源、规划管理、决策提供重要的科学依据。     

Modeling Flow and Sediment Transport in a River System Using an Artificial Neural Network

A river system is a network of intertwining channels and tributaries, where interacting flow and sediment transport processes are complex and floods may frequently occur. In water resources management of a complex system of rivers, it is important that instream discharges and sediments being carried by streamflow are correctly predicted. In this study, a model for predicting flow and sediment transport in a river system is developed by incorporating flow and sediment mass conservation equations into an artificial neural network (ANN), using actual river network to design the ANN architecture, and expanding hydrological applications of the ANN modeling technique to sediment yield predictions. The ANN river system model is applied to modeling daily discharges and annual sediment discharges in the Jingjiang reach of the Yangtze River and Dongting Lake, China. By the comparison of calculated and observed data, it is demonstrated that the ANN technique is a powerful tool for real-time prediction of flow and sediment transport in a complex network of rivers. A significant advantage of applying the ANN technique to model flow and sediment phenomena is the minimum data requirements for topographical and morphometric information without significant loss of model accuracy. The methodology and results presented show that it is possible to integrate fundamental physical principles into a data-driven modeling technique and to use a natural system for ANN construction. This approach may increase model performance and interpretability while at the same time making the model more understandable to the engineering community.     

多元回归和BP人工神经网络模型对混合厌氧消化产气量的预测研究

在中温且控制pH值条件下,对脂肪类单基质和城市污水厂剩余污泥进行混合厌氧消化试验。基于多元回归原理和BP人工神经网络原理,对其建立产气量预测模型。由实验数据计算得出:两个阶段多元回归模型的预测平均准确率分别为75.69%和79.29%;BP神经网络模型的预测平均准确率为79.05%。通过对比两种模型的预测结果可知,两种模型都有较高的预测准确率,但BP模型的预测准确率更高,更适用于混合厌氧消化产气量预测。     

ANN模型对污水处理系统性能的评价

作者基于B－P算法的人工神经网络，对污水处理系统的性能作了预测和评价，并通过所建立的一个生物转盘处理系统的人工神经网络预测模型实例，验证了人工神经网络用于污水处理系统性能评价方面的可行性与准确性。     

Spatial Prediction of Ground Subsidence Susceptibility Using an Artificial Neural Network

Ground subsidence in abandoned underground coal mine areas can result in loss of life and property. We analyzed ground subsidence susceptibility (GSS) around abandoned coal mines in Jeong-am, Gangwon-do, South Korea, using artificial neural network (ANN) and geographic information system approaches. Spatial data of subsidence area, topography, and geology, as well as various ground-engineering data, were collected and used to create a raster database of relevant factors for a GSS map. Eight major factors causing ground subsidence were extracted from the existing ground subsidence area: slope, depth of coal mine, distance from pit, groundwater depth, rock-mass rating, distance from fault, geology, and land use. Areas of ground subsidence were randomly divided into a training set to analyze GSS using the ANN and a test set to validate the predicted GSS map. Weights of each factor’s relative importance were determined by the back-propagation training algorithms and applied to the input factor. The GSS was then calculated using the weights, and GSS maps were created. The process was repeated ten times to check the stability of analysis model using a different training data set. The map was validated using area-under-the-curve analysis with the ground subsidence areas that had not been used to train the model. The validation showed prediction accuracies between 94.84 and 95.98%, representing overall satisfactory agreement. Among the input factors, “distance from fault” had the highest average weight (i.e., 1.5477), indicating that this factor was most important. The generated maps can be used to estimate hazards to people, property, and existing infrastructure, such as the transportation network, and as part of land-use and infrastructure planning.     

太湖生态模拟系统构建与应用

湖泊生态系统模拟在湖泊富营养化研究中发挥着越来越重要的作用,是湖泊生态生态系统管理的重要手段。湖泊生态系统模型及其相关软件的发展经历了从简单的回归模型、单一的营养盐平衡模型到目前复杂的生态系统动力学模型。本文详细的介绍了湖泊生态模拟的原理、结构框架设计、功能、运行环境及参数等特性,借助太湖有关资料建立了太湖生态系统模拟模型,并对该模型进行了验证分析,验证结果表明该模型在太湖有很好的适应性。     

Predicting Carbon Monoxide Concentrations in the Air of Pardis City,Iran, Using an Artificial Neural Network

To date, several methods have been proposed to explain the complex process of air pollution prediction. One of these methods uses neural networks. Artificial neural networks (ANN) are a branch of artificial intelligence, and because of their nonlinear mathematical structures and ability to provide acceptable forecasts, they have gained popularity among researchers. The goal of our study as documented in this article was to compare the abilities of two different ANNs, the multilayer perceptron (MLP) and radial basis function (RBF) neural networks, to predict carbon monoxide (CO) concentrations in the air of Pardis City, Iran. For the study, we used data collected hourly on temperature, wind speed, and humidity as inputs to train the networks. The MLP neural network had two hidden layers that contained 13 neurons in the first layer and 25 neurons in the second layer and reached a mean bias error (MBE) of 0.06. The coefficient of determination (R²), index of agreement (IA), and the Nash–Scutcliffe efficiency (E) between the observed and predicted data using the MLP neural network were 0.96, 0.9057, and 0.957, respectively. The RBF neural network with a hidden layer containing 130 neurons reached an MBE of 0.04. The R², IA, and E between the observed and predicted data using the RBF neural network were 0.981, 0.954, and 0.979, respectively. The results provided by the RBF neural network had greater acceptable accuracy than was the case with the MLP neural network. Finally, the results of a sensitivity analysis using the MLP neural network indicated that temperature is the primary factor in the prediction of CO concentrations and that wind speed and humidity are factors of second and third importance when forecasting CO levels.     

基于因子分析的BP神经网络对成都市需水量的预测

在对四川省成都市的水资源进行综合规划与评价的过程中,通过主成分因子分析可知人口、GDP、给排水管道长度等因素对需水量预测有较大影响,如何建立这些因素与需水量之间的数学关系是预测工作的重点。本文将介绍通过MATLAB数学分析软件建立BP神经网络预测模型,并对模型的预测结果进行评价。     

神经网络在空气污染预报中的应用研究

空气污染预报是一项复杂的系统工程,是当今环境科学研究的热点,国内外已有将神经网络法应用于大气污染预报的研究。本论文以PM2．5为例,采用伦敦市PM2．5的小时平均浓度数据,使用传统的BP神经网络建立预报模型,定量预测伦敦市PM2.5的小时平均浓度,探讨了大气污染预报网络的建模过程中,扩大样本集、去除样本集数据噪声和在输入向量中加入气象变量等因素对建模所产生的影响。最后得出结论,适当的选择样本集、气象变量,有利于提高所建立网络模型的预测精度。     

基于人工神经网络模型预测2017~2018年成都市危险废物增量研究

危险废物对环境或者人体健康会造成有害影响,有效地预测其产量是优化管理和合理处置的重要依据。以2008~2016年成都市危险废物产生量为基础,通过数据带入和整合及综合各参数因子的影响,利用人工神经网络模型预测方法客观反映并预测成都市危废产量的变化趋势。结果表明该模型预测2017~2018年成都市危险废物年产量分别达到24.46万t和26.88万t,模拟精度偏差低。因此,人工神经网络模型可以作为一种预测危险废物产生量的工具,其预测结果可以为职能部门提供决策参考。     

基于灰色神经网络的能源消费组合预测模型

组合预测对于信息不完备的复杂经济系统具有一定的实用性。鉴于能源消费系统的复杂性和非线性特征，利用我国能源消费的历史数据，采用灰色预测的GM（1，1）、无偏GM（1，1）和pGM（1，1）3种模型与人工神经网络进行优化组合，建立了灰色神经网络的能源消费组合预测模型，实证分析结果获得了更为精确的预测效果，可以作为能源消费预测的有效工具。同时，能源消费的预测结果也表明今后必须以节能为主导思想，努力建设资源节约型社会和环境友好型社会。     

南四湖湿地建设的综合效益

南四湖作为南水北调东线工程的主要输水干线和调蓄水库,目前人工湿地建设已初见成效。从环境效益、生态效益、社会效益、经济效益、科研效益等方面分析了南四湖人工湿地在改善南四湖流域生态环境、稳定南水北调调水水质方面发挥的重要作用。     

人工神经网络用于大气环境质量评价与排序

本文应用于人工神经网络Ｂ－Ｐ算法，建立了大气环境质量Ｂ－Ｐ网络评价模型，该模型应用于实例评价和环境质量排序结果与灰色综合评价法评价结果相比较，表明Ｂ－Ｐ网络法用于大气环境质量评价合理，客观，并具有可比性。     

Assessing Linkages in Stream Habitat,Geomorphic Condition,and Biological Integrity Using a Generalized Regression Neural Network

Watershed managers often use physical geomorphic and habitat assessments in making decisions about the biological integrity of a stream, and to reduce the cost and time for identifying stream stressors and developing mitigation strategies. Such analysis is difficult since the complex linkages between reach‐scale geomorphic and habitat conditions, and biological integrity are not fully understood. We evaluate the effectiveness of a generalized regression neural network (GRNN) to predict biological integrity using physical (i.e., geomorphic and habitat) stream‐reach assessment data. The method is first tested using geomorphic assessments to predict habitat condition for 1,292 stream reaches from the Vermont Agency of Natural Resources. The GRNN methodology outperforms linear regression (69% vs. 40% classified correctly) and improves slightly (70% correct) with additional data on channel evolution. Analysis of a subset of the reaches where physical assessments are used to predict biological integrity shows no significant linear correlation, however the GRNN predicted 48% of the fish health data and 23% of macroinvertebrate health. Although the GRNN is superior to linear regression, these results show linking physical and biological health remains challenging. Reasons for lack of agreement, including spatial and temporal scale differences, are discussed. We show the GRNN to be a data‐driven tool that can assist watershed managers with large quantities of complex, nonlinear data.     

Predicting Debris Yield From Burned Watersheds: Comparison of Statistical and Artificial Neural Network Models1

Abstract: Alluvial fans in southern California are continuously being developed for residential, industrial, commercial, and agricultural purposes. Development and alteration of alluvial fans often require consideration of mud and debris flows from burned mountain watersheds. Accurate prediction of sediment (hyper‐concentrated sediment or debris) yield is essential for the design, operation, and maintenance of debris basins to safeguard properly the general population. This paper presents results based on a statistical model and Artificial Neural Network (ANN) models. The models predict sediment yield caused by storms following wildfire events in burned mountainous watersheds. Both sediment yield prediction models have been developed for use in relatively small watersheds (50‐800 ha) in the greater Los Angeles area. The statistical model was developed using multiple regression analysis on sediment yield data collected from 1938 to 1983. Following the multiple regression analysis, a method for multi‐sequence sediment yield prediction under burned watershed conditions was developed. The statistical model was then calibrated based on 17 years of sediment yield, fire, and precipitation data collected between 1984 and 2000. The present study also evaluated ANN models created to predict the sediment yields. The training of the ANN models utilized single storm event data generated for the 17‐year period between 1984 and 2000 as the training input data. Training patterns and neural network architectures were varied to further study the ANN performance. Results from these models were compared with the available field data obtained from several debris basins within Los Angeles County. Both predictive models were then applied for hind‐casting the sediment prediction of several post 2000 events. Both the statistical and ANN models yield remarkably consistent results when compared with the measured field data. The results show that these models are very useful tools for predicting sediment yield sequences. The results can be used for scheduling cleanout operation of debris basins. It can be of great help in the planning of emergency response for burned areas to minimize the damage to properties and lives.     

基于人工神经网络的绥化市县域经济发展水平分析

选取8个经济指标，运用人工神经网络（ANN）的理论和方法，构建应用最为广泛的BP网络模型，对2004年绥化市10个县市的经济发展水平进行了评价。结果表明，绥化市县域经济发展水平差异十分显著，其中肇东等3县域属于高水平类型，海伦等4个县域为中等类型，明水等3个县域属于落后类型。     

模糊神经网络模型在拟建项目环境质量评价中的应用

由于评价因子与环境质量标准之间的模糊关系,导致传统的评价模型结果的不确定性。本文将模糊理论和神经网络模型相结合,利用模糊理论的隶属度反映各因子的质量相对状态,从而更精确地识别模糊性。通过模糊神经网络模型对环境质量标准样本的学习和测试,得出评价等级,并与模糊综合评价方法进行对比,实验结果表明,模糊神经网络模型用于环境质量评价是可行的,且比模糊综合评价方法更为客观、合理。     

神经网络模型在北海市空气日报污染指数计算中的应用

根据空气质量日报的实际需要,引入了Excel软件神经网络技术建立空气质量日报污染指数计算模型,采用LM算法提高了计算精度,并将模型应用于北海市空气日报。结果表明:此法较之实际公式计算法更加快捷方便,并且计算结果相当吻合。