随机噪声干扰下水质模型参数的鲁棒估计方法

本文采用Dobbins-BOD-DO水质模型，用仿真试验模拟了随机噪声干扰对河流水质模型参数估计的影响。试验表明，最小二乘(LS)估计方法关于有色噪声和较高水平白噪声干扰不具鲁棒性，噪声干扰使估计参数漂离系统真实参数。为了克服随机噪声对河流水质模型参数估计的干扰，提出了一种水质模型参数的鲁棒估计方法，即基于M-估计的信赖域算法。通过对比试验和计算表明，M-估计对于有色噪声和白噪声干扰具有鲁棒性，无论是没有扰动情况还是有各种水平各种噪声类型的干扰情况，该方法能稳健可靠地搜索到真实值，且在估值精度、收敛性、抗噪性和鲁棒性方面均优于最小二乘(LS)估计方法。     

Regionalization of the Modified Bartlett–Lewis rectangular pulse stochastic rainfall model across the Korean Peninsula

The six parameters of the Modified Bartlett–Lewis Rectangular Pulse (MBLRP) model were regionalized across the Korean Peninsula for all 12 calendar months. The parameters of the MBLRP model were estimated at each of the 59 rain gauges and they were spatially interpolated using the Ordinary Kriging method in order to produce maps. The parameter search space used in the parameter estimation process was repetitively narrowed through cross-validation in order to remove the impact of the multi-modality of the MBLRP model. The synthetic rainfall time series generated based on the parameter maps successfully reproduced the various statistical properties of the observed rainfall, such as mean, variance, lag-1 autocorrelation, and probability of zero rainfall at a wide range of time accumulation levels (e.g. hourly through daily). The maps representing the general rainfall characteristics, such as the average rainfall depth per rain storm, the average rain storm duration, the average number of rain cells per rain storm, and the average rain cell duration were also produced based on the estimated parameters. Lastly, some helpful tips in regionalizing the parameters of the Poisson cluster rainfall models are discussed.     

Identification of virus transport parameters in groundwater: Data errors and bias

The present study discusses the identification of virus transport parameters in groundwater from the virus concentration data using inverse procedure. The parameters are estimated by minimizing the deviations between the model predicted and experimentally observed virus concentrations. Model parameters are estimated from hypothetically generated virus concentration data by numerical inversion of the governing virus transport equation employing Levenberg-Marquardt optimization algorithm. The bias induced by the objective function on the parameter estimates is studied in detail by adding Gaussian noise to the hypothetically generated virus concentration data. Statistical analysis is performed for quantifying the bias in terms of sample mean and confidence intervals. The parameter estimation results indicate that while estimating two or more parameters simultaneously, the objective function induces an undue bias in the parameter estimates. It is also found that the induced bias is quite significant in case of inactivation parameters even at lower noise levels. The optimization algorithm is also applied to estimate the transport parameters from the virus concentration data of a column experiment.     

地下河天窗水位变化分析及预测

以贵州省普定县后寨地下河流域内平山天窗水位为研究对象,分析了水位变化规律,并采用偏最小二乘回归、人工神经网络及两种方法的结合建立了高水位阶段水位半日预测模型,对各模型进行了对比分析。结果表明：天窗水位变化划分为枯水位、开采-恢复、高水位3个阶段;高水位阶段天窗水位主要受前半日内的降雨及前半日时的水位影响;偏最小二乘回归是进行天窗水位预测的合适方法。该研究有助于岩溶地区水循环规律的认识及地下河水资源的开发利用。     

Modeling river water quality parameters using modified adaptive neuro fuzzy inference system

Water quality is always one of the most important factors in human health. Artificial intelligence models are respected methods for modeling water quality. The evolutionary algorithm(EA) is a new technique for improving the performance of artificial intelligence models such as the adaptive neuro fuzzy inference system(ANFIS) and artificial neural networks(ANN). Attempts have been made to make the models more suitable and accurate with the replacement of other training methods that do not suffer from some shortcomings, including a tendency to being trapped in local optima or voluminous computations. This study investigated the applicability of ANFIS with particle swarm optimization(PSO)and ant colony optimization for continuous domains(ACO_R) in estimating water quality parameters at three stations along the Zayandehrood River, in Iran. The ANFIS-PSO and ANFIS-ACO_R methods were also compared with the classic ANFIS method, which uses least squares and gradient descent as training algorithms. The estimated water quality parameters in this study were electrical conductivity(EC), total dissolved solids(TDS), the sodium adsorption ratio(SAR), carbonate hardness(CH), and total hardness(TH). Correlation analysis was performed using SPSS software to determine the optimal inputs to the models. The analysis showed that ANFIS-PSO was the better model compared with ANFIS-ACO_R. It is noteworthy that EA models can improve ANFIS' performance at all three stations for different water quality parameters.     

果蝇优化算法在区域高程拟合中的应用

针对最小二乘支持向量机拟合法难以选择最优参数的问题,将果蝇优化算法引入最小二乘支持向量机中,构建区域GPS高程拟合模型的方法,利用果蝇优化算法全局寻优能力强、过程简洁、参数少等优点,解决最小二乘支持向量机的参数寻优问题,并通过最小二乘支持向量机来构建高程拟合模型。结果表明,与BP神经网络拟合方法相比,引入果蝇优化算法的最小二乘支持向量机拟合方法具有更高的稳定性,内符合精度比标准最小二乘支持向量机提高了26%。     

韩江流域横山站蒸发量季节周期预报模型

采用季节周期SARIMA模型预报横山水文站2008年1月至12月蒸发量。根据AIC准则优选模型阶数,采用非线性最小二乘法估计模型参数,残差序列经x2检验为白噪声序列,模型较合理。对横山站2008年月蒸发量预报结果表明,SARIMA模型预报精度较高,预报误差低于20%的月份占全年91.7%,相对误差低于10%的月份占全年58.3%。     

Application of a simulated annealing optimization to a physically based erosion model

A major risk concerning the calibration of physically based erosion models has been partly attributable to the lack of robust optimization tools. This paper presents the essential concepts and application to optimize the erosion parameters of an erosion model using data collected in an experimental basin, with a global optimization method known as simulated annealing (SA) which is suitable for solving optimization problems of large scales. The physically based erosion model that was chosen to be optimized here is the Watershed Erosion Simulation Program (WESP), which was developed for small basins to generate the hydrograph and the respective sedigraph. The field data were collected in an experimental basin located in a semiarid region of Brazil. On the basis of these results, the following erosion parameters were optimized: the soil moisture-tension parameter (N(s)) that depends also on the initial moisture content, the channel erosion parameter (a), the soil detachability factor (K(R)), and the sediment entrainment parameter by rainfall impact (K(I)), whose values could serve as initial estimates for semiarid regions within northeastern Brazil.     

Non-Stationary Rainfall Intensity-Duration-Frequency Relationship: a Comparison between Annual Maximum and Partial Duration Series

The rainfall Intensity-Duration-Frequency (IDF) relationship is the primary input for storm water management and other engineering design applications across the world and it is developed by fitting an appropriate theoretical probability distribution to annual maximum (AM) series or partial duration series (PDS) of rainfall. The existing IDF relationship developing methods consider the extreme rainfall series as a stationary series. There exist few studies that compared AM and PDS datasets for developing rainfall IDF relationship in a stationary condition. However, during the last few decades, the intensity and frequency of extreme rainfall events are increasing due to global climate change and creating a non-stationary component in the extreme rainfall series. Therefore, the rainfall IDF relationship developed with the stationary assumption is no longer tenable in a changing climate. Hence, it is inevitable to develop non-stationary rainfall IDF relationship and to understand the differences in non-stationary rainfall IDF relationships derived using AM and PDS datasets. Consequently, the objectives of this study are: (1) to develop non-stationary rainfall IDF relationships using both AM and PDS datasets; (2) to compare them in terms of return level estimation. In particular, the non-linear trend in different durations’ PDS and AM datasets of Hyderabad city (India) rainfall is modeled using Multi-objective Genetic Algorithm (MGA) generated Time based covariate. In this study, the PDS datasets are modeled by the Generalized Pareto Distribution (GPD) while the AM datasets are modeled by the Generalized Extreme Value Distribution (GEVD). The time-varying component is introduced in the scale parameter of the GPD and the location parameter of the GEVD by linking the MGA generated covariate. In addition, the complexity of each non-stationary model is identified using the corrected Akaike Information Criteria (AICc) and the statistical significance of trend parameter in the non-stationary models is estimated using the Likelihood Ratio (LR) test. Upon detecting significant superiority of non-stationary models, the return levels of extreme rainfall event for 2-, 5-, 10- and 25-year return periods are calculated using non-stationary models. From the results, it is observed that the non-stationary return levels estimated with PDS datasets are higher than those estimated with AM datasets for short durations and smaller return periods while the non-stationary return levels estimated with AM datasets are higher than those estimated with PDS datasets for long durations and higher return periods.     

Self-adaptive Green-Ampt infiltration parameters obtained from measured moisture processes

The Green-Ampt(G-A) infiltration model(i.e., the G-A model) is often used to characterize the infiltration process in hydrology. The parameters of the G-A model are critical in applications for the prediction of infiltration and associated rainfall-runoff processes. Previous approaches to determining the G-A parameters have depended on pedotransfer functions(PTFs) or estimates from experimental results, usually without providing optimum values. In this study, rainfall simulators with soil moisture measurements were used to generate rainfall in various experimental plots. Observed runoff data and soil moisture dynamic data were jointly used to yield the infiltration processes, and an improved self-adaptive method was used to optimize the G-A parameters for various types of soil under different rainfall conditions. The two G-A parameters, i.e., the effective hydraulic conductivity and the effective capillary drive at the wetting front, were determined simultaneously to describe the relationships between rainfall, runoff, and infiltration processes. Through a designed experiment, the method for determining the GA parameters was proved to be reliable in reflecting the effects of pedologic background in G-A type infiltration cases and deriving the optimum G-A parameters. Unlike PTF methods, this approach estimates the G-A parameters directly from infiltration curves obtained from rainfall simulation experiments so that it can be used to determine site-specific parameters. This study provides a self-adaptive method of optimizing the G-A parameters through designed field experiments. The parameters derived from field-measured rainfall-infiltration processes are more reliable and applicable to hydrological models.     

Calibration of three resuspension/sedimentation models

Three resuspension and sedimentation models (Blom, Lick and Partheniades and Krone) are calibrated and evaluated on data from flume experiments with sediments from Lake Ketel and in situ suspended solids measurements. We applied a formal parameter estimation technique in combination with a statistical evaluation of the model fit and parameter estimates. All three models produce a reasonable reconstruction of the data from the flume experiment and the in situ observations. The differences in the model fit of the three models are small, except for the in situ observations. Here the sum of squared residuals for Partheniades and Krone's is about twice the sum for Blom's and Lick's model. The correlation between parameters in resuspension/sedimentation models can be very high, leading to an uncertainty in parameter estimates of 25-50. The parameter estimations based on the flume data are up to orders of magnitude higher than those estimated from field observations.     

Kinetic parameters estimation in an anaerobic digestion process using successive quadratic programming.

In this work, an optimization method is implemented in an anaerobic digestion model to estimate its kinetic parameters and yield coefficients. This method combines the use of advanced state estimation schemes and powerful nonlinear programming techniques to yield fast and accurate estimates of the aforementioned parameters. In this method, we first implement an asymptotic observer to provide estimates of the non-measured variables (such as biomass concentration) and good guesses for the initial conditions of the parameter estimation algorithm. These results are then used by the successive quadratic programming (SQP) technique to calculate the kinetic parameters and yield coefficients of the anaerobic digestion process. The model, provided with the estimated parameters, is tested with experimental data from a pilot-scale fixed bed reactor treating raw industrial wine distillery wastewater. It is shown that SQP reaches a fast and accurate estimation of the kinetic parameters despite highly noise corrupted experimental data and time varying inputs variables. A statistical analysis is also performed to validate the combined estimation method. Finally, a comparison between the proposed method and the traditional Marquardt technique shows that both yield similar results; however, the calculation time of the traditional technique is considerable higher than that of the proposed method.     

Forecasting Hourly Intermittent Rainfall by Deep Belief Networks with Simple Exponential Smoothing

Accurate rainfall forecasting is essential in planning and managing water resource systems efficiently. However, intermittent rainfall patterns increase the difficulty of accurately forecasting rainfall values. Deep learning techniques have recently been popular and powerful in forecasting. Thus, this study employed deep belief networks with a simple exponential smoothing procedure (DBNSES) to forecast hourly intermittent rainfall values in Taiwan. Weather factors were used as independent variables to forecast rainfall volume. The simple exponential smoothing data preprocessing procedure was used to deal with the intermittent data patterns. The other three forecasting models, namely the least squares support vector regression (LSSVR), the generalized regression neural network (GRNN), and the backpropagation neural network (BPNN), were employed to forecast rainfall using the same data sets. In addition, genetic algorithms were utilized to determine the parameters of four forecasting models. The empirical results indicate that the developed DBNSES models are superior to the other forecasting models in terms of forecasting accuracy. In addition, the DBNSES can obtain smaller values of RMSE than those in the previous studies. Therefore, the DBNSES model is a suitable and effective way of forecasting rainfall with intermittent data patterns.

     

Application of isotopic information for estimating parameters in Philip infiltration model

Minimizing parameter uncertainty is crucial in the application of hydrologic models.Isotopic information in various hydrologic components of the water cycle can expand our knowledge of the dynamics of water flow in the system,provide additional information for parameter estimation,and improve parameter identifiability.This study combined the Philip infiltration model with an isotopic mixing model using an isotopic mass balance approach for estimating parameters in the Philip infiltration model.Two approaches to parameter estimation were compared:(a) using isotopic information to determine the soil water transmission and then hydrologic information to estimate the soil sorptivity,and(b) using hydrologic information to determine the soil water transmission and the soil sorptivity.Results of parameter estimation were verified through a rainfall infiltration experiment in a laboratory under rainfall with constant isotopic compositions and uniform initial soil water content conditions.Experimental results showed that approach(a),using isotopic and hydrologic information,estimated the soil water transmission in the Philip infiltration model in a manner that matched measured values well.The results of parameter estimation of approach(a) were better than those of approach(b).It was also found that the analytical precision of hydrogen and oxygen stable isotopes had a significant effect on parameter estimation using isotopic information.     

基于总体最小二乘法的系统响应修正方法

为了提高实时洪水预报的预报精度,提出了基于总体最小二乘平差理论的系统响应方法。传统的系统响应方法基于最小二乘法,只能考虑观测值的误差,因而传统的系统响应方法不能考虑动态系统响应矩阵存在的误差。对传统方法进行了分析并引入了总体最小二乘平差理论,改进方法同时考虑了动态系统响应矩阵和观测值的误差。针对动态系统响应矩阵的病态情形,通过总体最小二乘的岭估计解法给出稳定解。通过改进方法对新安江模型中的土壤含水量进行修正,应用于七里街流域,并与传统的系统响应方法进行比较。结果表明,两种方法都能提高模拟精度,改进方法相比于传统方法精度明显提升;改进方法的修正效果要优于传统方法,更加稳定。     

Prediction of hydraulic load for urban storm control of a municipal WWT plant

Three different methodologies are assessed which provide predictions of the hydraulic load to the treatment plant one hour ahead. The three models represent three different levels of complexity ranging from a simple regression model over an adaptive grey-box model to a complex hydrological and full dynamical wave model (Chow et al., 1988). The simple regression model is estimated as a transfer function model of rainfall intensity to influent flow. It also provides a model for the base flow. The grey-box model is a state space model which incorporates adaptation to the dry weather flow as well as the rainfall runoff. The full dynamical flow model is a distributed deterministic model with many parameters, which has been calibrated based on extensive measurement campaigns in the sewer system. The three models are compared by the ability to predict the hydraulic load one hour ahead. Five rain events in a test period are used for evaluating the three different methods. The predictions are compared to the actual measured flow at the plant one hour later. The results show that the simple regression model and the adaptive grey-box model which are identified and estimated on measured data perform significantly better than the hydrological and full dynamical flow model which is not identifiable and needs calibration by hand. For frontal rains no significant difference in the prediction performance between the simple regression model and the adaptive grey-box model is observed. This is due to a rather uniform distribution of frontal rains. A single convective rain justifies the adaptivity of the grey-box model for non-uniformly distributed rain, i.e. the predictions of the grey-box model were significantly better than the predictions of the simple regression model for this rain event. In general, models for model-based predictive control should be kept simple and identifiable from measured data.     

Converting Daily Rainfall Data to Sub-daily—Introducing the MIMD Method

One of the most important analysis in many hydrological and agricultural studies is to convert the daily rainfall data into sub-daily (hourly) because in many rainfall stations, only the daily rainfall data are available and for a comprehensive rainfall analysis, these data should be converted to sub-daily. Many experimental and analytical methods are available for this conversion but one of the simplest yet accurate ones has been proposed by the Indian Meteorological Department (IMD). Since the IMD method has shown low accuracy in some regions, in this study, the IMD method is modified to a single parameter equation, called Modified Indian Meteorological Department (MIMD) in order to improve the accuracy of the conversion. For this reason, the parameter is calibrated so that the maximum correlation between observed and estimated values is achieved. Five stations in different regions with different climatic conditions were selected so that the daily and sub-daily rainfall data were available in each of them. Then, the parameter of the MIMD method was derived for each station. The results were compared with both observed data and IMD method and it was shown that the mean correlation coefficient of MIMD and IMD methods were 0.9 and 0.73 respectively for 12-h rainfall depth which indicated that the accuracy of the MIMD method in estimation of sub-daily rainfall depths was significantly increased. Moreover, the results showed that the accuracy of the MIMD method decreases as rainfall duration decreases.

     

PREDICTING ECOLOGICAL FLOW REGIME AT UNGAGED SITES: A COMPARISON OF METHODS

Nineteen ecologically relevant streamflow characteristics were estimated using published rainfall–runoff and regional regression models for six sites with observed daily streamflow records in Kentucky. The regional regression model produced median estimates closer to the observed median for all but two characteristics. The variability of predictions from both models was generally less than the observed variability. The variability of the predictions from the rainfall–runoff model was greater than that from the regional regression model for all but three characteristics. Eight characteristics predicted by the rainfall–runoff model display positive or negative bias across all six sites; biases are not as pronounced for the regional regression model. Results suggest that a rainfall–runoff model calibrated on a single characteristic is less likely to perform well as a predictor of a range of other characteristics (flow regime) when compared with a regional regression model calibrated individually on multiple characteristics used to represent the flow regime. Poor model performance may misrepresent hydrologic conditions, potentially distorting the perceived risk of ecological degradation. Without prior selection of streamflow characteristics, targeted calibration, and error quantification, the widespread application of general hydrologic models to ecological flow studies is problematic. Published 2012. This article is a U.S. Government work and is in the public domain in the USA.     

Permeable Asphalt Hydraulic Conductivity and Particulate Matter Separation With XRT

Permeable asphalt (PA) is a composite material with an open graded mix design that provides a pore structure facilitating stormwater infiltration. PA is often constructed as a wearing course for permeable pavements and on impervious pavements to reduce aquaplaning and noise. The pore structure of PA functions as a filter promoting particulate matter (PM) separation. The infiltrating flow characteristics are predominately dependent on pore diameter and pore interconnectivity. X-Ray microTomography (XRT) has successfully estimated these parameters that are otherwise difficult to obtain through conventional gravimetric methods. Pore structure parameters allow modeling of hydraulic conductivity (k) and filtration mechanisms; required to examine the material behavior for infiltration and PM separation. In this study, pore structure parameters were determined through XTR for three PA mixture designs. Additionally, the Kozeny-Kovàv model was implemented to estimate k. PM separation was evaluated using a pore-to-PM diameter categorical model. This filtration mechanism model was validated with data from a rainfall simulator. The filtration model provided a good correlation between measured and modeled data. The identification of filtration mechanisms and k facilitate the design and evaluation of permeable pavement systems as a best management practice (BMP) for runoff volume and peak flow as well as PM and PM-partitioned chemical separation.

     

基于模型筛选法的卡尔曼滤波法在大坝变形分析中的应用

以模型筛选法为基础，将筛选出的变形误差最小的模型的模型参数作为状态向量，用卡尔曼滤波法进行大坝变形分析，实例计算表明，这种方法能够提高模型的拟合及预报精度。