SOIL SALINITY PREDICTION USING ARTIFICIAL NEURAL NETWORKS1

ABSTRACT: This study explores the applicability of Artificial Neural Networks (ANNs) for predicting salt build‐up in the crop root zone. ANN models were developed with salinity data from field lysimeters subirrigated with brackish water. Different ANN architectures were explored by varying the number of processing elements (PEs) (from 1 to 30) for replicate data from a 0.4 m water table, 0.8 m water table, and both 0.4 and 0.8 m water table lysimeters. Different ANN models were developed by using individual replicate treatment values as well as the mean value for each treatment. For replicate data, the models with twenty, seven, and six PEs were found to be the best for the water tables at 0.4 m, 0.8 m and both water tables combined, respectively. The correlation coefficients between observed salinity and ANN predicted salinity of the test data with these models were 0.89, 0.91, and 0.89, respectively. The performance of the ANNs developed using mean salinity values of the replicates was found to be similar to those with replicate data. Not only was there agreement between observed and ANN predicted salinity values, the results clearly indicated the potential use of ANN models for predicting salt build‐up in soil profile at a specific site.     

FLOOD STAGE FORECASTING WITH SUPPORT VECTOR MACHINES1

ABSTRACT: Machine learning techniques are finding more and more applications in the field of forecasting. A novel regression technique, called Support Vector Machine (SVM), based on the statistical learning theory is explored in this study. SVM is based on the principle of Structural Risk Minimization as opposed to the principle of Empirical Risk Minimization espoused by conventional regression techniques. The flood data at Dhaka, Bangladesh, are used in this study to demonstrate the forecasting capabilities of SVM. The result is compared with that of Artificial Neural Network (ANN) based model for one‐lead day to seven‐lead day forecasting. The improvements in maximum predicted water level errors by SVM over ANN for four‐lead day to seven‐lead day are 9.6 cm, 22.6 cm, 4.9 cm and 15.7 cm, respectively. The result shows that the prediction accuracy of SVM is at least as good as and in some cases (particularly at higher lead days) actually better than that of ANN, yet it offers advantages over many of the limitations of ANN, for example in arriving at ANN's optimal network architecture and choosing useful training set. Thus, SVM appears to be a very promising prediction tool.     

智能型地理信息系统（AIGIS）在城市抗震防灾规划中的应用

本文简要介绍地理信息系统GIS的发展概况、主要功能以及在国内外的应用现状.对现有的GIS软件技术加以改进,引进人工神经元网络和模糊综合评判技术,发展了一种智能型的GIS,在我国若干城市的抗震设防区划工作中应用,取得良好的效果。     

WATER QUALITY PREDICTION WITHIN AN INTERBASIN TRANSFER SYSTEM1

ABSTRACT A methodology for predicting the spatial and temporal levels of conservative water quality constituents within a multibasin water resource system is presented. Dissolved solids, sulfates, and chlorides are the constituents used during this investigation; however, any other conservative ion or mineral can be incorporated into the simulation model. The methodology is tested on the proposed Texas Water System. The water quality model, QNET-I, utilizes monthly canal and river flows and reservoir storage levels calculated by the Texas Water Development Board's systems simulation model. Discharge-concentration relationships are developed for each source of water in the system, including significant waste-water discharges. Reservoirs in the system are assumed to be completely mixed with respect to conservative constituents. A mass balance analysis is performed for each node and each month during the simulation period. The output from the water quality simulation is a table of the concentrations of the conservative water quality constituents at each demand point in the system and in each reservoir and canal for every month the system is in operation. The desired quality of the water at the demand locations is used to determine the economic utility of transporting and mixing water from various sources.     

DEVELOPMENT OF A BASIN GEOMORPHIC INFORMATION SYSTEM USING A TIN-DEM DATA STRUCTURE1

ABSTRACT: To make a distributed rainfall-runoff model, it is very important to build a model of topographic surface of a basin which takes account of the direction of water flow. In this paper, a geographic information system in hydrologic modeling, the BGIS (Basin Geomorphic Information Systems) are presented for modeling a river basin using a TIN-DEM (Triangulated Irregular Network - Digital Elevation Model) data structure. The BGIS have two core systems, which are the TIN-DEM generating system and the topographic analysis system. In the TIN-DEM generating system, landscapes are modeled as a set of contiguous non-overlapping terangular facets whose vertices are made up of points on a regular grid DEM and on river segments. These triangular facets are subdivided, if needed, so that each of them has only one side through which water flows out. The TIN-DEM generating system is made up of four modules, (1) a module for generating triangles from a grid DEM, (2) a module for getting rid of pits, (3) a module for joining discontinuous valley segments to a channel network, (4) a module for subdividing triangular facets. In the topographic analysis system, using datasets processed with the TIN-DEM generating system, a watershed source area for any segments in a stream network are delineated automatically, and topographic attributes of slopes, aspects, flow path lengths and upslope contributing areas are computed.     

Application of Decision Analysis to Forest Road Deactivation in Unstable Terrain

Resource managers require objective methodologies to optimize decisions related to forest road deactivation and other aspects of road management, especially in steep terrain, where road-related slope failures inflict extensive environmental damage. Decision analysis represents a systematic framework that clearly identifies real options and critical decision points. This framework links current decisions with expected future outcomes and provides advantages such as a common currency to systematically explore the liability consequences of limited budget expenditures to road deactivation and other road-related activities. Furthermore, the decision framework prevents the analysis from becoming hopelessly entangled by the vast number of possibilities generated by the alternative occurrences, magnitudes, and consequences of landslide/debris flow events and provides the information required for the first step of an adaptive management process. Here, a structured analysis of potential environmental risks for a road deactivation project in coastal British Columbia, Canada is presented. The application of decision analysis generates a ranking of the expected benefits of proposed deactivation activities on various road sections. The ranking distinguishes between road sections that offer high expected benefit from those that offer moderate to low expected benefit. Seventeen of 171, 100–m road segments accounted for 18% of the cumulative cost and 98% of the cumulative expected net benefits from road deactivation. Furthermore, the cost of deactivating a section of road is related to the expected benefit from such deactivation, thus providing the basis for more effective resource allocation and budgeting decisions.     

煤层底板采动导水破坏深度计算的神经网络方法

在综合分析影响煤层底板采动导水破坏深度因素的基础上 ,应用人工神经网络方法 ,建立了底板破坏深度的计算模型。该模型利用现场观测资料作为学习训练样本和测试样本 ,对模型的测算结果、理论计算值和实测值进行了对比分析。结果表明 :用神经网络方法计算底板破坏深度考虑的因素更加全面 ,结果更接近于实际。笔者研究的计算模型和测算方法 ,为承压水上安全采煤决策提供了科学依据。     

人工神经网络方法在资源与环境预测方面的应用

用人工神经网络方法对不同水域、不同环境因子之间非线性和不确定性的复杂关系进行学习训练并预测检验。结果表明：人工神经网络方法在模拟和预测方面 优于传统的统计回归模型,在资源与环境方面的应用是可行的。具有较强的模拟预测能力。与传统的回归模型相比,人工神经网络方法不要求监测数据具有很强的规律性,就可用后的网络模型对其进行预报,燕且预测相对误差均比回归模型预测相对误差要小,具有一定的实用性。两个实例的应用     

DETERMINATION OF SEDIMENT YIELD BY TRANSFERRING RAINFALL DATA1

ABSTRACT: A methodology for obtaining the optimal design value to allow for sediment storage in a reservoir is presented for the situation where no data on sediment loads in the incoming streams are available. Information concerning the amount of sediment delivered to the reservoir over its life-time is obtained by a sediment yield model which uses data on rainfall amount and duration obtained from a nearby experimental watershed. Bayesian Decision Theory is used to obtain the optimal storage requirements in order to consider the natural variation of rainfall and the sampling error due to the short rainfall record available. The normally difficult calculations involved were made tractable by the use of simplifications and approximations valid in the context of the problem. Results show that sediment storage requirements can be calculated in this manner and that consideration of the uncertainties involved leads to a storage requirement substantially larger than that calculated without such consideration.     

A New Approach to Plant Diversity Assessment Combining HPLC Data, Simplex Mixture Design and Discriminant Analysis

The quantitative assessment of plant diversity and its monitoring with time represent a key environmental issue for management and conservation of natural resources. Assessment of plant diversity could be based on chemical analyses of secondary metabolites (e.g. flavonoids, terpenoids), because of the substantial quantitative and qualitative between-individual variability in such compounds. At a geographical scale, the plant populations become widely dispersed, and their monitoring from numerous routine individual analyses could become restricting. To overcome such constraint, this study develops a multivariate calibration model giving the relative frequency of a particular taxon from a simple high-performance liquid chromatography (HPLC) analysis of a plant mixture. The model was built from a complete set of mixtures combining different taxons, according to an experimental design (Scheffé’s matrix). For each mixture, a reference HPLC pattern was simulated by averaging the individual HPLC profiles of the constitutive taxons. The calibration models, based on Bayesian discriminant analysis (BDA), gave statistical relationships between the contributions of each taxon in mixtures and reference HPLC patterns of these mixtures. Finally, these models were validated on new mixtures by using outside plants. This new biodiversity survey approach is illustrated on four chemical taxons (four chemotypes) of Astragalus caprinus (Fabaceae). The more differentiated the taxon, the better predicted its contributions (in mixtures) were by BDA calibration model. This new approach could be very useful for a global routine survey of plant diversity.