Ensemble Streamflow Forecast: A GLUE‐Based Neural Network Approach1

Abstract: While training a Neural Network to model a rainfall‐runoff process, generally two aspects are considered: its capability to be able to describe the complex nature of the processes being modeled and the ability to generalize so that novel samples could be mapped correctly. The general conclusion is that, the smallest size network capable of representing the sample distribution is the best choice, as far as generalization is concerned. Oftentimes input variables are selected a priori in what is called an explanatory data analysis stage and are not part of the actual network training and testing procedures. When they are, the final model will have only a “fixed” type of inputs, lag‐space, and/or network structure. If one of these constituents was to change, one would obtain another equally “optimal” Neural Network. Following Beven and others' generalized likelihood uncertainty estimate approach, a methodology is introduced here that accounts for uncertainties in network structures, types of inputs, and their lag‐space relationships by looking at a population of Neural Networks rather than target in getting a single “optimal” network. It is shown that there is a wide array of networks that provide “similar” results, as seen by a likelihood measure, for different types of inputs, lag‐space, and network size combinations. These equally optimal networks expose the range of uncertainty in streamflow predictions and their expected value results in a better performance than any of the single network predictions.     

SUPPLEMENTATION OF MISSING VALUES IN WATER QUALITY DATA1

ABSTRACT The problem of estimating missing values in water quality data using linear interpolation and harmonic analysis is studied to see which one of these two methods yields better estimates for the missing values. The data used in this study consisted of midnight values of dissolved oxygen from the Ohio River collected over a period of one year at Stratton station. Various hypothetical cases of missing data are considered and the two methods of supplementing missing values are evaluated using statistical tests. The results indicate that when the percentage of missed data points exceeded ten percent of the total number in the original sample, harmonic analysis usually yielded better estimates for both the regularly and irregularly missed cases. For data that exhibit cyclic variation, examples of which are dissolved oxygen concentration and water temperature, harmonic analysis as a data generation technique appears to be superior to linear interpolation.     

COMPARING SOME METHODS OF ESTIMATING MEAN AREAL RAINFALL1

ABSTRACT: A comparison of 13 different methods of estimating mean areal rainfall was made on two areas in New Mexico, U.S.A., and one area in Great Britain. Daily, monthly and yearly rainfall data were utilized. All methods, in general, yielded comparable estimates, especially for yearly values. This suggested that a simpler method would be preferable for estimating mean areal rainfall in these areas.     

A COMPARISON OF METHODS FOR CALCULATION OF RADAR-RAINFALL HOURLY ACCUMULATIONS1

ABSTRACT: We compared two interpolation schemes for calculation of hourly accumulation of radar-rainfall. The schemes are: (1) the Advection Method, and (2) the Space-Time Kriging Method. The performance of the methods is investigated using numerical simulation experiments. Space-time evolution of rainfall fields is generated from a stochastic model. The generated fields are sampled following typical radar scanning strategies, and the investigated schemes are applied to obtain accumulated rainfall patterns. The statistical results and a visual analysis of the graphical images suggest that it is advisable to use an interpolation scheme for radar observations even when storm velocity is not high. The Space-Time Kriging Method provides the best results for low wind velocity. The Advection Method has the smallest standard deviation and mean absolute error, and preserves well the true rainfall pattern for high wind velocity.     

Critical Steps for Continuing Advancement of Satellite Rainfall Applications for Surface Hydrology in the Nile River Basin1

Gebremichael, Mekonnen, Emmanouil N. Anagnostou, and Menberu M. Bitew, 2010. Critical Steps for Continuing Advancement of Satellite Rainfall Applications for Surface Hydrology in the Nile River Basin. Journal of the American Water Resources Association (JAWRA) 46(2):361-366. DOI: 10.1111/j.1752-1688.2010.00428.x. Abstract: Given the increasingly higher resolution and data accessibility, satellite precipitation products could be useful for hydrological application in the Nile River Basin, which is characterized by lack of reasonably dense hydrological in situ sensors and lack of access to the existing dataset. However, in the absence of both extreme caution and research results for the Nile basin, the satellite rainfall (SR) products may not be used, or may even be used erroneously. We identify two steps that are critical to enhance the value of SR products for hydrological applications in the Nile basin. The first step is to establish representative validation sites in the Nile basin. The validation site will help to quantify the errors in the different kinds of SR products, which will be used to select the best products for the Nile basin, include the errors in decision making, and design strategies to minimize the errors. Using rainfall measurements collected from the unprecedented high-density rain gauge network over a small region within the Nile basin, we indicate that SR estimates could be subject to significant errors, and quantification of estimation errors by way of establishing validation sites is critically important in order to use the SR products. The second step is to identify the degree of hydrologic model complexity required to obtain more accurate hydrologic simulation results for the Nile basin when using SR products as input. The level of model complexity may depend on basin size and SR algorithm, and further research is needed to spell out this dependence for the Nile basin.     

基于BP模型对城市交通噪声的数据处理和预测

城市交通噪声的预测和评价技术是城市交通可持续发展的重要研究内容,直接为城市交通规划中环境容量分析和环境影响评价服务。本文通过实测的大量数据,运用神经网络中的BP模型及其算法建立车辆数、道路宽度和交通噪声之间的关系,对城市道路交通噪声数据进行处理和预测。     

SPACE-TIME VALIDATION OF A THUNDERSTORM RAINFALL MODEL1

A probability model for predicting the occurrence and magnitude of thunderstorm rainfall developed in the southwestern United States was tested in the metropolitan Chicago area with reasonable success, especially for the moderate to the extreme runoff-producing events. The model requires the estimation of two parameters, the mean number of events per year and the conditional probability of rain given that an event has occurred. To tie in the data from more than one gage in an area, an event can be defined in several ways, such as the areal mean rainfall exceeding 0.50 inch and at least one gage receiving more than 1.0 inch. This type of definition allows both of the model parameters to be obtained from daily warm-season rainfall records. Regardless of the definition used a Poisson distribution adequately described the number of events per season. A negative binomial distribution was derived as representing the frequency density function for rainfall where several gages are employed in defining a storm. Chicago data fit both distributions very well at events with relatively high return periods. The results indicate the possibility of using the model on a regional basis where limited amount of data may be used to estimate parameters for extensive areas.     

MODELING HYDROLOGIC TIME SERIES FROM THE ARCTIC1

The general intervention model is applied to hydrologic and meteorologjc time series from the Canadian Arctic. The authors show how the model is able to account for environmental interventions, missing observations in the data, changes in data collection procedures, the effects of external inputs, as well as seasonality and autocorrelation. Methods for identifying transfer functions by making use of a physical understanding of the processes involved are demonstrated and sample applications of the general intervention model to Arctic data are shown.     

PREDICTING MEASUREMENT ERROR OF AREAL MEAN PRECIPITATION DURING EXTREME EVENTS1

ABSTRACT: The areal mean precipitation (AMP) over a catchment is normally calculated using point measurements at rainfall gages. Error in AMP estimates occurs when an insufficient number of gages are used to sample precipitation which is highly variable in space. AMP error is investigated using historic, severe rainfalls with a set of hypothetical catchments and raingage networks. The potential magnitude of error is estimated for typical gage network densities and arrangements. Possible sources of error are evaluated, and a method is proposed for predicting the magnitude of error using data that are commonly available for severe, historic rainfall.     

基于人工神经网络的突发环境事故应急系统

突发环境事故现场具有形势紧迫、信息获取困难等特点,引入人工神经网络技术(Artificial Neural Networks,ANN)设计开发的突发环境事故应急决策支持系统(ANN-EEDSS),能够完成现场应急指挥调度的定量化决策支持。关键技术包括突发环境事故神经网络构建方法,事故信息与应急措施的数字化方法,决策信息的翻译表示,事故影响因素的权重分析。最后阐述了系统现场运行和工作的流程。     

基于Zig Bee技术的压力传感器数据采集系统设计

介绍了一种基于无线传感器网络的高精度压力传感器数据采集系统的设计方法,将无线传感器网络的Zig Bee技术嵌入到系统的上位机和下位机中,实现了系统可在恶劣环境下的无线通信的需求,系统同时扩展了RS-485数字输出接口和4~20 m A模拟输出接口两种形式,增加了系统的适用性。在实际应用中该系统性能稳定可靠,传感器在常温情况下精度达到0.07%。     

Field‐Scale Application of Three Types of Neural Networks to Predict Ground‐Water Levels1

Abstract: In this paper, a field‐scale applicability of three forms of artificial neural network algorithms in forecasting short‐term ground‐water levels at specific control points is presented. These algorithms are the feed‐forward back propagation (FFBP), radial basis networks (RBN), and generalized regression networks (GRN). Ground‐water level predictions from these algorithms are in turn to be used in an Optimized Regional Operations Plan that prescribes scheduled wellfield production for the coming four weeks. These models are up against each other for their accuracy of ground‐water level predictions on lead times ranging from a week to four weeks, ease of implementation, and execution times (mainly training time). In total, 208 networks of each of the three algorithms were developed for the study. It is shown that although learning algorithms have emerged as a viable solution at field scale much larger than previously studied, no single algorithm performs consistently better than others on all the criteria. On average, FFBP networks are 20 and 26%, respectively, more accurate than RBN and GRN in forecasting one week ahead water levels and this advantage drops to 5 and 9% accuracy in forecasting four weeks ahead water levels, whereas GRN posted a training time that is only 5% of the training time taken by that of FFBP networks. This may suggest that in field‐scale applications one may have to trade between the type of algorithm to be used and the degree to which a given objective is honored.