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.     

CONSTRUCTING A PROBLEM SOLVING ENVIRONMENT TOOL FOR HYDROLOGIC ASSESSMENT OF LAND USE CHANGE1

ABSTRACT: This paper describes the integration of a comprehensive hydrological model known as the Hydrological Simulation Program Fortran (HSPF) into a problem solving environment (PSE) for watershed management. The original PSE concept was a structure providing web‐based access to a suite of models, including HSPF and other models of in‐stream hydrodynamics, biological impacts and economic effects, for the watershed‐wide assessment of alternative land use scenarios. The present paper describes only the HSPF integration into the PSE program. Example applications to the 148 square kilometer (57 square mile) Back Creek subwatershed in the upper Roanoke River system (1,479 square kilometers or 571 square miles) in southwest Virginia are used to illustrate important concepts and linkages between land development and hydrological change using hypothetical' what if'scenarios. The features of HSPF and its limitations in this context are discussed. The paper as such is a proof‐of‐concept paper and not a completion report. It is intended to describe the PSE tool building process rather than analysis of the many possible simulation outcomes. However, the dominance of raw imperviousness as a contributor to hydrograph response is apparent in all the PSE simulations described in this paper.     

ASSEMBLY OF MAP‐BASED STREAM NARRATIVES TO FACILITATE STAKEHOLDER INVOLVEMENT IN WATERSHED MANAGEMENT1

ABSTRACT: Watershed stewardship activities throughout North America have evolved into a process that requires more involvement in planning and decision making by community stakeholders. Active involvement of all stakeholders in the process of watershed stewardship is dependent on effective exchange of information among participants, and active involvement of a wide range of stakeholders from “communities of place” as well as those from “communities of interest.” We developed a map‐based stream narrative tool as a means to: (a) assemble a wealth of incompletely documented, “traditional” ecological or natural history observations for rivers or streams; and (b) promote a higher level of active involvement by community stakeholders in contributing to information‐based, watershed management. Creation of stream narratives is intended for use as a tool to actively engage local stakeholders in the development of a more comprehensive information system to improve management for multiple stewardship objectives in watersheds. Completion of map‐based stream narrative atlases provides a valuable supplement to other independent efforts to assemble observations and knowledge about land‐based natural resources covering entire watersheds. We are confident that completion of stream narrative projects will make a valuable addition to the information and decision making tools that are currently available to the public and resource agencies interested in advancing the cause of community‐based approaches to watershed and ecosystem management.     

长年关水稻田耕作改制的选择

试验研究表明,当早稻收获时间作晚稻才开始幼穗分化,即适宜早、晚稻出穗开花之间有60多天的时间,种植双杂间作稻就可获得好的收成,效果、效益都很显著。1992年用D 优287间汕优63,在达县永进乡的示范田中个别田块产量已达15t/ha,而盆地丘陵区适宜早、晚稻出穗开花之间的时间多数地区不到70天,种植双杂连作稻有困难。且双杂间作稻在抗灾增产,合理利用农村劳动力,提高稻田复种等方面优于双杂连作稻。因此,四川盆地丘陵区的中稻→冬水田在耕作改制中发展部分双杂间作稻是恰当的,其范围在巴中、盐亭、简阳一线以南海拔400m 以下的广大丘陵区。     

Application of a GIS-based stream buffer generation model to environmental policy evaluation

In this article a GIS method is presented for riparian environmental buffer generation. It integrates a scientifically tested buffer width delineation model into a GIS framework. Using the generally available data sets, it determines buffer widths in terms of local physical conditions and expected effectiveness. Technical burdens of data management, computation, and result presentation are handled by the GIS. The case study in which the method was used to evaluate the stream buffer regulations in a North Carolina county demonstrates its capability as a decision support tool to facilitate environmental policy formulation and evaluation, and environmental dispute resolution.     

粉料无尘装车机工作原理及其在冶金厂矿的应用

介绍集装车与防尘统一于一体的无尘装车技术及具有设备本质安全化特点的无尘装车机在解决粉状物料装车作业时的扬尘问题,保护作业工人健康,维护厂区环境方面的重要优越性。     

浅谈我国逆反射材料现状及其发展趋势展望

随着逆反射材料的广泛应用 ,其带来的安全警示效果、产品质量、生产工艺、检测方法及规范管理等方面的问题被日趋重视 ,笔者通过对逆反射材料的基本构造及反光特性的分析 ,论述了其对人和物的安全防护作用 ,并对我国目前逆反射材料的应用领域、产品质量及检测设备、检测标准等方面的问题进行探讨和评述 ,根据国内外的发展趋势 ,对我国开发及发展逆反射材料产业提出了笔者观点 ,为政府决策提供科学依据     

3S技术在财产保险防洪救灾中的应用

分析了保险业防灾减损的研究现状,阐述了3S的技术特点及其应用领域。和如何将3S技术应用于财产保险防洪救灾领域,建立财产保险防洪救灾决策支持系统。介绍了系统的整体构架及其功能特点,该系统可与保险公司业务系统紧密结合,显示直观。同时包括保险分析、灾害分析和专题图显示等多项功能。实现了多源数据的无缝集成。因而能从理论方法和关键技术上指导保险公司的防灾减损工作。切实提高保险公司的经济效益。以深圳市为示范区,该系统在应用中取得了良好效果。     

Characterization of Dust Storms to Hong Kong in April 1998

In April 1998, two intense dust storms were generated in CentralAsia and transported eastward across East Asia (15 and 19 April). This article presents the chemical characterization ofHong Kong (HK) aerosols during the dust storms. During the 15 Aprildust storm, hourly respiratory suspended particles (RSP)(particle diameter smaller than 10 m) concentrationsmonitored at 7 sites in Hong Kong reached the peak valuessynchronously between 9 and 11 a.m. on 17 April, in which thehighest concentration was 267 g m^-3. Analysis ofthe RSP samples showed that concentrations of crustalelements (Ba, Ca, Cd, Cr, Fe, Mg, K⁺) and anthropogenicspecies (As, Ni, Pb, Zn, NH₄ ⁺, NO₃ ^-,SO₄ ^2- and total carbon) were substantiallyenhanced. Enhancement of these species was more than afactor of 2 to 14 relative to the non dust period. The totalcarbon content was high, at 59 g m^-3 (notincluding carbonate), and the enrichment factors of Asand Pb on 17 April were 122 and 117, respectively. Thisimplied that anthropogenic materials together with mineraldust were transported to HK from Mainland China. Based onmaterial balance calculations, mineral dust contributed41% to the observed RSP mass on 17 April, which was 2 times thatof the nondust sample (22%). From the 5-day backwardtrajectory analysis, this storm was transported directlyfrom Northwest China to HK. However, there was nocorresponding observation for the 19 April dust stormaerosol. Consequently, 15 April storm had stronger impact onHK's atmosphere than 19 April storm. Compared to the HK AirQuality Objective, 15 April dust storm did not cause seriousair pollution in HK.     

Evaluating Nonpoint Pollution Policy Using a Tightly Coupled Spatial Decision Support System

Policy makers often must rely on the cumulative impact of independent actions taken by local landowners to achieve environmental goals. The connection between policy, regulation, and local action, however, is often not well understood and, thus, the impact of proposed policies may be difficult to predict. In this study we evaluate the effectiveness of alternative policy scenarios for agricultural set aside programs (e.g., the conservation reserve program administered by the United States Department of Agriculture) in reducing nonpoint pollution. Two alternative policy scenarios are developed and analyzed; one based on the erodibility index (detachment), the other sediment yield (transport). An estimate of the cumulative impact of associated land use change on nonpoint pollution is made using the AGNPS distributed parameter watershed model. This work is completed within the Cypress Creek watershed in southern Illinois. An analysis of the resulting data suggests that the most efficacious regulatory strategy for achieving nonpoint water pollution goals depends, in part, on place-specific land use patterns. This conclusion provides a solid argument for place-based regulatory strategies.