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.     

Applying value of information methods to prioritize elements for water quality management with an example of linear alkylbenzene sulfonate in the Yodo River,Japan

Narrowing the decision space is crucial in water quality management at the meso-scale for developing countries, where a lack of data and financial budgets prevent the development of appropriate management plans and result in serious water quality degradation in many rivers. In this study, a framework for handling this task is proposed, comprising a lumped water quality model, with sensitivity and uncertainty analyses, and a management domain, including loss estimation and value of information analysis. Through a case study with linear alkylbenzene sulfonate (LAS) in the Yodo River, it is found that non-point sources and flow rate are factors that influence LAS concentration at the hot spot location. By considering the entire process of water quality management planning, we identify that the definition of the cost function of LAS treatment determines the appropriate estimation for the expected loss in reducing LAS under uncertain water quality conditions. The value of information analysis with “expected value of including uncertainty” and “expected value of perfect information” further helps estimate the benefit of including uncertainty in decision-making and the financial cost for obtaining more information regarding inputs that have been previously prioritized.     

Predicting Particulate Matter (PM2.5) Concentrations in the Air of Shahr‐e Ray City,Iran, by Using an Artificial Neural Network

Particulate matter (PM), along with other air pollutants, pose serious hazards to human health. The Artificial Neural Network (ANN) is a branch of artificial intelligence that has an ability to make accurate predictions. In this article, the authors describe such methods and how historical data on air quality, moisture, wind velocity, and temperature in Shahr‐e Ray City, located at the southern tip of Tehran, was used to train an ANN to provide accurate predictions of PM concentrations. The availability of such predictions can offer significant assistance to those who are working to reduce air pollution.     

Approximating SWAT Model Using Artificial Neural Network and Support Vector Machine1

Abstract: With the popularity of complex, physically based hydrologic models, the time consumed for running these models is increasing substantially. Using surrogate models to approximate the computationally intensive models is a promising method to save huge amounts of time for parameter estimation. In this study, two learning machines [Artificial Neural Network (ANN) and support vector machine (SVM)] were evaluated and compared for approximating the Soil and Water Assessment Tool (SWAT) model. These two learning machines were tested in two watersheds (Little River Experimental Watershed in Georgia and Mahatango Creek Experimental Watershed in Pennsylvania). The results show that SVM in general exhibited better generalization ability than ANN. In order to effectively and efficiently apply SVM to approximate SWAT, the effect of cross‐validation schemes, parameter dimensions, and training sample sizes on the performance of SVM was evaluated and discussed. It is suggested that 3‐fold cross‐validation is adequate for training the SVM model, and reducing the parameter dimension through determining the parameter values from field data and the sensitivity analysis is an effective means of improving the performance of SVM. As far as the training sample size, it is difficult to determine the appropriate number of samples for training SVM based on the test results obtained in this study. Simple examples were used to illustrate the potential applicability of combining the SVM model with uncertainty analysis algorithm to save efforts for parameter uncertainty of SWAT. In the future, evaluating the applicability of SVM for approximating SWAT in other watersheds and combining SVM with different parameter uncertainty analysis algorithms and evolutionary optimization algorithms deserve further research.     

SOIL MOISTURE PREDICTION USING SUPPORT VECTOR MACHINES1

ABSTRACT: Herein, a recently developed methodology, Support Vector Machines (SVMs), is presented and applied to the challenge of soil moisture prediction. Support Vector Machines are derived from statistical learning theory and can be used to predict a quantity forward in time based on training that uses past data, hence providing a statistically sound approach to solving inverse problems. The principal strength of SVMs lies in the fact that they employ Structural Risk Minimization (SRM) instead of Empirical Risk Minimization (ERM). The SVMs formulate a quadratic optimization problem that ensures a global optimum, which makes them superior to traditional learning algorithms such as Artificial Neural Networks (ANNs). The resulting model is sparse and not characterized by the “curse of dimensionality.” Soil moisture distribution and variation is helpful in predicting and understanding various hydrologic processes, including weather changes, energy and moisture fluxes, drought, irrigation scheduling, and rainfall/runoff generation. Soil moisture and meteorological data are used to generate SVM predictions for four and seven days ahead. Predictions show good agreement with actual soil moisture measurements. Results from the SVM modeling are compared with predictions obtained from ANN models and show that SVM models performed better for soil moisture forecasting than ANN models.     

OPTIMIZING WELL PLACEMENT IN A COASTAL AQUIFER: OUTER CAPE COD,MASSACHUSETTS1

ABSTRACT: An inverse‐simulation approach is used to determine optimal strategies for developing public water‐supply systems in a shallow, coastal aquifer on the outermost arm of the Cape Cod peninsula in Massachusetts. Typically a forward simulation (or “trial and error”) approach is used to find best pumping strategies, but the chances of finding success with this tact diminish as the number of potential options grows large. Well locations and pumping rates are optimized with respect to: (1) providing sufficient water to areas of water‐quality impairment, (2) minimizing impacts to nearby surface waters, (3) preventing saltwater contamination due to overpumping, and (4) minimizing financial cost of well development. Potential well sites and water‐supply scenarios are separated into “politically‐based” and “resource‐based” categories to gain insight into the degree that pre‐existing political boundaries hinder best management practices. The approach provides a promising tool in transboundary water‐resources settings because it allows stakeholders to find solutions that best meet everyone's goals, as opposed to pursuing options that will create conflict, or are less than optimal.     

Optimal Reorganization of NASA Earth Science Data for Enhanced Accessibility and Usability for the Hydrology Community

A long‐standing “Digital Divide” in data representation exists between the preferred way of data access by the hydrology community and the common way of data archival by earth science data centers. Typically, in hydrology, earth surface features are expressed as discrete spatial objects (e.g., watersheds), and time‐varying data are contained in associated time series. Data in earth science archives, although stored as discrete values (of satellite swath pixels or geographical grids), represent continuous spatial fields, one file per time step. This Divide has been an obstacle, specifically, between the Consortium of Universities for the Advancement of Hydrologic Science, Inc. and NASA earth science data systems. In essence, the way data are archived is conceptually orthogonal to the desired method of access. Our recent work has shown an optimal method of bridging the Divide, by enabling operational access to long‐time series (e.g., 36 years of hourly data) of selected NASA datasets. These time series, which we have termed “data rods,” are pre‐generated or generated on‐the‐fly. This optimal solution was arrived at after extensive investigations of various approaches, including one based on “data curtains.” The on‐the‐fly generation of data rods uses “data cubes,” NASA Giovanni, and parallel processing. The optimal reorganization of NASA earth science data has significantly enhanced the access to and use of the data for the hydrology user community.     

Growth of the Decision Tree: Advances in Bottom‐Up Climate Change Risk Management

There has recently been a return in climate change risk management practice to bottom‐up, robustness‐based planning paradigms introduced 40 years ago. The World Bank's decision tree framework (DTF) for “confronting climate uncertainty” is one incarnation of those paradigms. In order to better represent the state of the art in climate change risk assessment and evaluation techniques, this paper proposes: (1) an update to the DTF, replacing its “climate change stress test” with a multidimensional stress test; and (2) the addition of a Bayesian network framework that represents joint probabilistic behavior of uncertain parameters as sensitivity factors to aid in the weighting of scenarios of concern (the combination of conditions under which a water system fails to meet its performance targets). Using the updated DTF, water system planners and project managers would be better able to understand the relative magnitudes of the varied risks they face, and target investments in adaptation measures to best reduce their vulnerabilities to change. Next steps for the DTF include enhancements in: modeling of extreme event risks; coupling of human‐hydrologic systems; integration of surface water and groundwater systems; the generation of tradeoffs between economic, social, and ecological factors; incorporation of water quality considerations; and interactive data visualization.     

Uncertainty Reduction of the Flood Stage Forecasting Using Neural Networks Model1

Abstract: The Elman Discrete Recurrent Neural Networks Model (EDRNNM), which is one of the special types of neural networks model, is developed and applied for the flood stage forecasting at the Musung station (No. 1) of the Wi‐stream catchment, which is one of the International Hydrological Program representative basins, Korea. A total of 135 different training patterns, which involve hidden nodes, standardization process, data length, and lead‐time, are selected for the minimization of the architectural uncertainty. The model parameters, such as optimal connection weights and biases, are estimated during the training performance of the EDRNNM, and we apply them to evaluate the validation performance of the EDRNNM. Sensitivity analysis is used to reduce the uncertainty of input data information of the EDRNNM. As the results of sensitivity analysis, the Improved EDRNNM consists of four input nodes resulting from the exclusion of Dongkok station (No.5) in initial five input nodes group of the EDRNNM. The accuracy of flood stage forecasting during the training and validation performances of the Improved EDRNNM remains the same as that of the EDRNNM. The Improved EDRNNM, therefore, gives highly reliable flood stage forecasting. The best optimal EDRNNM, so called the Improved EDRNNM, is determined by elimination of the uncertainties of architectural and input data information in this study. Consequently, we can avoid unnecessary data collection and operate the flood stage forecasting system economically.     

Applications of Explicitly Incorporated/Post‐Processing Measurement Uncertainty in Watershed Modeling

In the field of watershed modeling, the impact of measurement uncertainty (MU) on calibration results indicates the potential issue of inaccurate model predictions. It is important to note that MU refers to the uncertainty in measured data such as flow and nutrient values that are used to evaluate model outputs. The calculation of error statistics assuming measured data are deterministic may not be appropriate as has been frequently stated in literature. Although MU can affect model calibration results, it is rarely incorporated in modeling practice. MU can be incorporated in two schemes: explicitly incorporated (MU‐EI) during model calibration and post‐processed (MU‐PP) after calibration is completed. In this study, both schemes are implemented in a case study of the Arroyo Colorado Watershed, Texas. Unexpectedly, no substantial differences were observed between each scheme for flow predictions. Although MU did not cause dramatic differences in most sediment and NH₄‐N predictions, error statistics were affected in cases with MU greater than 50%, especially for sediment and NH₄‐N. Therefore, it is concluded that MU may not exert a significant impact on model predictions until certain threshold is reached. This study demonstrates that high levels of uncertainty in measured calibration/validation data significantly affect parameter estimation, especially in the auto‐calibration process.     

Critical Evaluation of How the Rosgen Classification and Associated “Natural Channel Design” Methods Fail to Integrate and Quantify Fluvial Processes and Channel Response1

Abstract: Over the past 10 years the Rosgen classification system and its associated methods of “natural channel design” have become synonymous to some with the term “stream restoration” and the science of fluvial geomorphology. Since the mid 1990s, this classification approach has become widely adopted by governmental agencies, particularly those funding restoration projects. The purposes of this article are to present a critical review, highlight inconsistencies and identify technical problems of Rosgen’s “natural channel design” approach to stream restoration. This paper’s primary thesis is that alluvial streams are open systems that adjust to altered inputs of energy and materials, and that a form‐based system largely ignores this critical component. Problems with the use of the classification are encountered with identifying bankfull dimensions, particularly in incising channels and with the mixing of bed and bank sediment into a single population. Its use for engineering design and restoration may be flawed by ignoring some processes governed by force and resistance, and the imbalance between sediment supply and transporting power in unstable systems. An example of how C5 channels composed of different bank sediments adjust differently and to different equilibrium morphologies in response to an identical disturbance is shown. This contradicts the fundamental underpinning of “natural channel design” and the “reference‐reach approach.” The Rosgen classification is probably best applied as a communication tool to describe channel form but, in combination with “natural channel design” techniques, are not diagnostic of how to mitigate channel instability or predict equilibrium morphologies. For this, physically based, mechanistic approaches that rely on quantifying the driving and resisting forces that control active processes and ultimate channel morphology are better suited as the physics of erosion, transport, and deposition are the same regardless of the hydro‐physiographic province or stream type because of the uniformity of physical laws.     

区域绿地概念形成脉络与深层发展研究

《城市绿地分类标准》(CJJ/T 85-2017)中区域绿地概念是构建区域生态安全格局的重要"生态源"与"锚固点",目前被视为提升人类福祉的重要环节。总结国内外相关研究进展,本文从宏观、中观、微观三个尺度,梳理国内外具有区域绿地类型特征的生态用地起源与形成脉络。根据我国区域绿地所处的发展阶段,分析今后我国区域绿地的发展方向,从三个方面探讨区域绿地深层发展方向:①依据区域绿地类型划分保护的优先级,建立完善的保护政策;②深化研究区域绿地空间形态与功能的耦合协调机制,促进区域绿地最优效能发挥;③统筹多类型的区域绿地,维系区域一体化可持续发展。     

DIAGNOSTIC APPROACH TO STREAM CHANNEL ASSESSMENT AND MONITORING1

ABSTRACT: We suggest that a diagnostic procedure, not unlike that followed in medical practice, provides a logical basis for stream channel assessment and monitoring. Our argument is based on the observation that a particular indicator or measurement of stream channel condition can mean different things depending upon the local geomorphic context and history of the channel in question. This paper offers a conceptual framework for diagnosing channel condition, evaluating channel response, and developing channel monitoring programs. The proposed diagnostic framework assesses reach‐level channel conditions as a function of location in the channel network, regional and local biogeomorphic context, controlling influences such as sediment supply and transport capacity, riparian vegetation, the supply of in‐channel flow obstructions, and disturbance history. Field assessments of key valley bottom and active channel characteristics are needed to formulate an accurate diagnosis of channel conditions. A similar approach and level of understanding is needed to design effective monitoring programs, as stream type and channel state greatly affect the type and magnitude of channel response to changes in discharge and sediment loads. General predictions are made for five channel types with respect to the response of various stream characteristics to an increase in coarse sediment inputs, fine sediment inputs, and the size and frequency of peak flows, respectively. These predictions provide general hypotheses and guidance for channel assessment and monitoring. However, the formulation of specific diagnostic criteria and monitoring protocols must be tailored to specific geographic areas because of the variability in the controls on channel condition within river basins and between regions. The diagnostic approach to channel assessment and monitoring requires a relatively high level of training and experience, but proper application should result in useful interpretation of channel conditions and response potential.     

Repeatability,Sensitivity, and Uncertainty Analyses of the BANCS Model Developed to Predict Annual Streambank Erosion Rates

Accelerated streambank erosion caused by channel instability can be the leading cause of sediment impairment of streams. Obtaining accurate streambank erosion rates for sediment budgeting and prioritizing mitigation efforts can be difficult and costly. One approach to quantifying streambank erosion rates is through the development and implementation of an empirically derived “Bank Assessment for Non‐point Source Consequences of Sediment” (BANCS) model. This study aims to improve the BANCS model application by evaluating repeatability between users and identifying sensitive and/or uncertain model inputs. Statistical analysis of streambank evaluations conducted by 10 different individuals suggests the implementation of the BANCS model may not be repeatable. This finding may be due to sensitive model inputs, such as streambank height and near‐bank stress level prediction method selection, and/or uncertain model inputs, such as bank material identification and the associated adjustment of erosion potential. Furthermore, it was found assessing streambanks as a group by obtaining a measure of central tendency from individual evaluations, as well as obtaining a higher level of training, may improve model implementation precision. Application of these suggestions may result in improved prediction of streambank erosion rates utilizing the BANCS model methodology.     

Twenty Years of New Zealand’s National Rivers Water Quality Network: Benefits of Careful Design and Consistent Operation1

Davies‐Colley, Robert J., David G. Smith, Robert C. Ward, Graham G. Bryers, Graham B. McBride, John M. Quinn, and Mike R. Scarsbrook, 2011. Twenty Years of New Zealand’s National Rivers Water Quality Network: Benefits of Careful Design and Consistent Operation. Journal of the American Water Resources Association (JAWRA) 47(4):750‐771. DOI: 10.1111/j.1752‐1688.2011.00554.x Abstract: This paper reviews New Zealand’s National Rivers Water Quality Network (NRWQN), which is now in its third decade of monitoring. The NRWQN is noteworthy for being operationally stable throughout its history, and the resulting consistency is increasingly valuable for detecting water quality trends and for “anchoring” temporary special purpose monitoring campaigns. The NRWQN was carefully designed following considerable effort to learn from monitoring experiences elsewhere. Monthly visits are made to 77 sites (all near hydrometric stations) on 35 river systems that cumulatively drain about one half of the national landscape. “Core” (routinely measured) variables are: conductivity, pH, temperature, dissolved oxygen, visual clarity, turbidity, colored dissolved organic matter, fecal indicator bacteria, and different forms of nitrogen and phosphorus (italics indicate field measurements). Associated benthic biological monitoring comprises monthly visual assessment of periphyton and annual sampling for macro‐invertebrates. We overview the conception, design, initiation, and operational history of the NRWQN, and highlight the diverse applications of its datasets including numerous scientific applications, national‐scale modeling of material fluxes, and state‐of‐environment reporting and practical water management at both regional and national scale. The qualified success of the NRWQN can probably be attributed to careful (and parsimonious) design and consistent operation.     

MODELING TRANSIENT pH DEPRESSIONS IN COASTAL STREAMS OF BRITISH COLUMBIA USING NEURAL NETWORKS1

ABSTRACT: Transient events in water chemistry in small coastal watersheds, particularly pH depressions, are largely driven by inputs of precipitation. While the response of each watershed depends upon both the nature of the precipitation event and the season of the year, how the response changes over time can provide insight into landscape changes. Neural network models for an urban watershed and a rural‐suburban watershed were developed in an attempt to detect changes in system response resulting from changes in the landscape. Separate models for describing pH depressions for wet season and dry season conditions were developed for a seven year period at each watershed. The neural network models allowed separation of the effects of precipitation variations and changes in watershed response. The ability to detect trends in pH depression magnitudes was improved by analyzing neural network residuals rather than the raw data. Examination of sensitivity plots of the models indicated how the neural networks were affected by different inputs. There were large differences in effects between seasons in the rural‐suburban watershed whereas effects in the urban watershed were consistent between seasons. During the study period, the urban watershed showed no change in pH depression response, while the rural‐suburban watershed showed a significant increase in the magnitude of pH depressions, likely the result of increased urbanization.     

Challenges and Opportunities for Creating Intelligent Hazard Alerts: The “FloodHippo” Prototype

Disasters evolving from hazards are a persistent and deadly occurrence in the United States. Despite this, hazard alerts have remained spatially vague, temporally imprecise, and lack actionable information. These deficiencies indicate a divide between the status quo and what is possible given modern environmental models, geographic information systems (GIS), and smartphone capabilities. This work describes an alternative, prototype system, “FloodHippo,” which integrates operational model outputs, cloud‐based GIS, and expanded communication channels to provide personal and interactive disaster alerts for floods. The precepts and methods underpinning FloodHippo apply equally to other disasters that evolve over space and time, presenting the opportunity for a more intelligent disaster response system. The development of such a system would not only minimize current shortcomings in disaster alerts but also improve resilience through individual action, along with community, academic, and federal cooperation.     

Estimating tree bole volume using artificial neural network models for four species in Turkey

Tree bole volumes of 89 Scots pine (Pinus sylvestris L.), 96 Brutian pine (Pinus brutia Ten.), 107 Cilicica fir (Abies cilicica Carr.) and 67 Cedar of Lebanon (Cedrus libani A. Rich.) trees were estimated using Artificial Neural Network (ANN) models. Neural networks offer a number of advantages including the ability to implicitly detect complex nonlinear relationships between input and output variables, which is very helpful in tree volume modeling. Two different neural network architectures were used and produced the Back propagation (BPANN) and the Cascade Correlation (CCANN) Artificial Neural Network models. In addition, tree bole volume estimates were compared to other established tree bole volume estimation techniques including the centroid method, taper equations, and existing standard volume tables. An overview of the features of ANNs and traditional methods is presented and the advantages and limitations of each one of them are discussed. For validation purposes, actual volumes were determined by aggregating the volumes of measured short sections (average 1 meter) of the tree bole using Smalian's formula. The results reported in this research suggest that the selected cascade correlation artificial neural network (CCANN) models are reliable for estimating the tree bole volume of the four examined tree species since they gave unbiased results and were superior to almost all methods in terms of error (%) expressed as the mean of the percentage errors.     

PREDICTING IMPACTS FROM WATER CONSERVATION AND ENERGY DEVELOPMENT ON THE SALTON SEA,CALIFORNIA1

ABSTRACT: An input-output model was developed to predict changes in Salton Sea salinity and water level until the year 2000 due to proposed water conservation efforts and geothermal and solar pond energy developments. The model SALINP provided good agreement with the observed salinities for 1960–80. While SALINP was not overly sensitive to one-year changes in any of the major inputs, a change in the historical means of the Imperial Valley runoff and evaporative loss inputs produced a significant effect on future predictions. The proposed water conservation measures caused the predicted Salton Sea salinity for 2000 to greatly exceed 40,000 ppm, the level at which adverse effects to wildlife are believed to occur. The possible geothermal development also produced predicted salinities considerably above 40,000 ppm. The salinity predictions for solar ponds by themselves and in conjunction with geothermal development were below 45,000 ppm for 2000. The solar pond and geothermal combination also resulted in a predicted lowering of the “natural” water level by 5 to 7 feet by 2000.