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.     

A LARGE ELECTRONIC WEIGHING LYSIMETER SYSTEM: DESIGN AND INSTALLATION1

ABSTRACT: To measure crop evapotranspiration, a large double tank, electronic weighting lysimeter system was designed and installed at the Shahid Bahonar University farm, Kerman, Iran. The system was installed in a 50 m² underground building. It includes two tanks of 3.00 m in diameter and 1.75 m deep. The weighing mechanism for each tank is a set of three compression strain gage load cells, which are fixed on 1.20 m height column above the floor. According to the specification of the load cells, the maximum possible weighing error may be about 0.01 percent of total mass, which is equivalent to 0.28 mm of water, but the measured error was equal to 1 kg mass, which is equivalent to 0.14 mm of water. The load cell data from each tank and the on‐site environmental data (temperature, humidity, and wind velocity and direction) are automatically recorded and saved in a personal computer hard disk for further use and analysis.     

Evaluation of the Impacts of Dock Structures and Land Use on Tidal Creek Ecosystems in South Carolina Estuarine Environments

Tidal creeks and their associated salt marshes are the primary link between uplands and estuaries in the southeastern region. They are also critical nursery and feeding grounds. In addition, the uplands surrounding creeks are preferred sites for homebuilding because of their natural beauty and the ability to access the estuary from a personal dock structure. The objective of this study was to evaluate the cumulative impacts of docks on tidal creek nursery habitats for both small and large tidal creeks. The number of docks was associated with the amount of impervious cover in both small and large creeks. The presence of docks had little measurable effect on sediment metal concentrations at the scale of small and large creeks. In small and large creeks, sediment polycyclic aromatic hydrocarbon (PAH) concentrations were related to the human activity in the upland that includes the presence of docks at the scale of small and large creeks. Some impacts on the benthic community were associated with docks and human activity in small creeks but not in large creeks. Suburban development may reduce fish and crustacean abundances, but the dock may potentially mediate the development effect. Individually, the harm to the marine environment resulting from dock shading, chrominated copper arsenate leachates, and PAH contamination was small at the scale of tidal creeks. However, impacts from dock structures could not be separated from anthropogenic watershed-scale effects. These results demonstrate that suburban development with its accompanying dock construction does represent a major source of environmental degradation to tidal creeks and associated salt marsh habitats.     

Methodological Development of an Index of Coastal Water Quality: Application in a Tourist Area

With the aim of obtaining an index of coastal water quality, a methodological procedure based on numerical classification and discriminant analysis is presented. The procedure was applied to nutrient data (ammonia, nitrite, nitrate, and phosphate) analyzed along the coastal waters of a Spanish tourist area. Using numerical classification, three levels of nutrient loading were revealed, characterizing oligotrophic, mesotrophic, and potentially eutrophic waters. Discriminant analysis was shown to be an effective methodological tool in the discrimination between trophic groups. For every group, the discriminant procedure generated the centroids. The centroids representing oligotrophic and potentially eutrophic conditions were used to establish the two extremes of the continuum of mesotrophic conditions in these coastal waters: Standardizing values from -1 to 1, the centroids for oligotrophic and potentially eutrophic waters yielded an interval that defined the range of mesotrophic conditions. This interval is proposed as a water quality index. The ability of the coastal water quality index to successfully predict mesotrophic conditions was proved with random samples.     

Emotional intelligence and individual performance: evidence of direct and moderated effects

We examined the direct and moderated effects of an ability‐based measure of emotional intelligence (MSCEIT© V2.0) on individual performance in a sample of business undergraduates. Controlling for general mental ability and personality, emotional intelligence explained unique incremental variance in performance ratings on only one of two measures of interpersonal effectiveness (public speaking effectiveness). However, the interaction of emotional intelligence with conscientiousness explained unique incremental variance both in public speaking and group behavior effectiveness, as well as academic performance (cumulative GPA). We conclude that the effects of emotional intelligence on performance are more indirect than direct in nature. Individuals must not only have emotional intelligence, but also must be motivated to use it. Copyright © 2006 John Wiley & Sons, Ltd.     

Hydrological Connectivity Between Headwater Streams and Downstream Waters: How Science Can Inform Policy1

Abstract: In January 2001, the U.S. Supreme Court ruled that the U.S. Army Corps of Engineers exceeded its statutory authority by asserting Clean Water Act (CWA) jurisdiction over non‐navigable, isolated, intrastate waters based solely on their use by migratory birds. The Supreme Court’s majority opinion addressed broader issues of CWA jurisdiction by implying that the CWA intended some “connection” to navigability and that isolated waters need a “significant nexus” to navigable waters to be jurisdictional. Subsequent to this decision (SWANCC), there have been many lawsuits challenging CWA jurisdiction, many of which are focused on headwater, intermittent, and ephemeral streams. To inform the legal and policy debate surrounding this issue, we present information on the geographic distribution of headwater streams and intermittent and ephemeral streams throughout the U.S., summarize major findings from the scientific literature in considering hydrological connectivity between headwater streams and downstream waters, and relate the scientific information presented to policy issues surrounding the scope of waters protected under the CWA. Headwater streams comprise approximately 53% (2,900,000 km) of the total stream length in the U.S., excluding Alaska, and intermittent and ephemeral streams comprise approximately 59% (3,200,000 km) of the total stream length and approximately 50% (1,460,000 km) of the headwater stream length in the U.S., excluding Alaska. Hillslopes, headwater streams, and downstream waters are best described as individual elements of integrated hydrological systems. Hydrological connectivity allows for the exchange of mass, momentum, energy, and organisms longitudinally, laterally, vertically, and temporally between headwater streams and downstream waters. Via hydrological connectivity, headwater, intermittent and ephemeral streams cumulatively contribute to the functional integrity of downstream waters; hydrologically and ecologically, they are a part of the tributary system. As this debate continues, scientific input from multiple fields will be important for policymaking at the federal, state, and local levels and to inform water resource management regardless of the level at which those decisions are being made. Strengthening the interface between science, policy, and public participation is critical if we are going to achieve effective water resource management.     

Modeling the long-term effects of fuelwood harvests on Pinyon-Juniper Woodlands

In the Southwestern United States, increasing demand for firewood has quickly promoted pinyon-juniper woodlands to commercial status. Slow recovery rates for pinyon and juniper and inadequate mensuration data present significant obstacles to predictive management efforts. Many National Forest districts have witnessed continued fuel use for at least the past 100 years. To explore the need for long-term analysis at the district level, we have developed FORMAN I, a computer simulation written in FORTRAN IV that models prolonged fuel harvesting and its impact on pinyon-juniper woodlands. The technique is well-suited for historical analyses and we comply with an initial application that involves the suggested impact of prehistoric peoples on a marginal woodland in Chaco Canyon, New Mexico. To accentuate the significance of the simulation, we have deliberately overestimated woodland parameters while maintaining conservative annual rates for firewood procurement. A low-density woodland (less than 14.8 cords/ha) is completely depleted within 200 years when subjected to tenth-through-twelfth century estimates of human demography for the canyon. Interesting conclusions are drawn from the lack of pinyon-juniper recovery at Chaco over the past millennium. Traditional assumptions, such as the pristine state of Southwestern vegetation prior to historic settlement and subsequent invasion of marginal grasslands by pinyon and juniper, are questioned.     

Macrobenthic surveys as a valuable tool for assessing freshwater quality in the Iberian Peninsula

Relationships between natural macrobenthic communities and aquatic environmental conditions have been recently used to perform the Biological Monitoring Water Quality (BMWQ) score system for assessing freshwater quality in the Iberian Peninsula. In this paper, the second version of the BMWQ score system is applied to two selected field studies (organic and nonorganic pollution) in order to illustrate the suitable functional character of this method. The total BMWQ (t-BMWQ) score and the average BMWQ (a-BMWQ) score are calculated, and their relationships with diversity indices are also examined. The BMWQ method appears to be very sensitive to changes in freshwater quality generated by simple organic pollution, but this method appears to be less sensitive to changes generated by complex nonorganic pollution. In addition, both scores are more sensitive measures to assess macroinvertebrate response to organic pollution than total density and diversity indices, exhibiting higher positive coefficients of linear correlation with Camargo's diversity index than with Shannon's and MacArthur's diversity indices. However, the a-BMWQ score is more sensitive to changes in family composition than the t-BMWQ score, whereas this latter score is more sensitive to family richness and sample size. In this sense, the use of the a-BMWQ score is recommended for the biological assessment of freshwater quality degradation in the Iberian Peninsula. More information is obtained for less effort when the a-BMWQ score is used.