Disaster investigation and analysis of Typhoon Morakot

Typhoon Morakot first struck Taiwan on 6 August 2009, which led to a much more serious disaster than the notorious ‘August 7th Flood’ which occurred 50 years previously; it broke records for almost all meteorological observing data with an estimated rainfall in five days surpassing Taiwan's annual average rainfall of 2500?mm. Such a huge amount of rainfall caused more serious damage than the 921 earthquake. The typhoon left 620 people dead, 80 people missing, and over NT$90 billion in total direct property losses in 2009. Therefore, within one week after the disaster, the National Science Council requested the National Science and Technology Center for Disaster Reduction to investigate the disasters caused by Typhoon Morakot and to carry out a comprehensive investigation to better understand the type, numbers, and scale of the disasters. Meanwhile, this investigation also aims to better understand the causes and to spot potentially dangerous areas. Our key focuses include: flooding, water facilities, landslides, debris flows, bridges, traffic, and agricultural losses. Finally, based on an intensive investigation, analysis, and review results, we hope to develop both short-term and long-term countermeasures for future disaster prevention reference.     

A modified entropy-based dynamic gravity model for the evacuation trip distribution problem during typhoons

Typhoon Morakot, which formed on 2 August 2009, was the deadliest typhoon in Taiwan’s history, responsible for over 700 deaths on the island. During the typhoon evacuation process, one critical issue is how to efficiently distribute the evacuation trips to a limited number of shelters based on both spatial and temporal considerations. This paper proposes a modified entropy-based dynamic gravity model to reflect the spatial and temporal distribution of the evacuees and the shelters. A unique feature of the proposed model is that the entropy is explicitly incorporated within the travel cost constraints. The spatial and temporal relationships between evacuees and shelters can be reflected through the impedance functions and the discretized time intervals with better performance than the traditional model. A simulation-assignment model is applied to generate the zone-to-zone travel time. A calibration analysis based on the solution procedure is conducted for the Jiasian network, in Kaohsiung city, which was heavily affected by the Typhoon Morakot. The calibration results show that the modified entropy-based dynamic gravity model leads to better convergence patterns in the entropy values, higher travel cost coefficients, and lower average generalized trip costs than the traditional model, and is suitable for use with the evacuation plan during typhoons.     

Typhoon Morakot meteorological analyses

Typhoon Morakot yielded record-breaking precipitation and caused severe disasters in Southern Taiwan in early August 2009. This event revealed the desperate need for extreme rainfall and disaster prediction. The multiscale analysis on circulations demonstrates that Typhoon Morakot not only developed in a monsoon trough, which is a large-scale system in favor of the development of convection systems, it was also involved with complicated features of different scales that included typhoon circulation itself, the circulation of northwestward-propagating 10–30-day oscillation, and the circulation of northward-propagating 40–50-day oscillation. The other remarkable features of Typhoon Morakot are: the asymmetric structure of convection, less damage occurred near the typhoon center than at the fringes, and the record-breaking torrential rain in the mountain areas of south Taiwan induced by the flows on the south fringe of the typhoon. The causes of disastrous precipitation from Typhoon Morakot can be stated as: (1) a large-scale environment in favor of convection development and a plentiful vapor supply provided by strong southwesterly flow, (2) the interaction between topography, typhoon circulation, and large-scale circulation resulting in the heavy rainfall in the mountains of Southern Taiwan, and (3) the weakening of steering flow slowed down Morakot's translation speed before it made landfall and extended the duration of heavy precipitation.     

Comparison of regression models,grey models,and supervised learning models for forecasting flood stage caused by typhoon events

Abstract

In this paper, four simple dynamic prediction methods and two supervised learning techniques including a linear regression model, a quadratic regression model, an original grey prediction model, a modified grey prediction model, a back‐propagation neural network model, and an epsilon‐SVM regression model were investigated for the forecasting of flood stage one hour ahead for early warning of flooding hazards. Quantitative evaluations were carried out between the predicted values by using the six forecasting models and the measured values obtained in the field. The comparisons confirm the ability of the simple grey prediction model to forecast flood stage by using only three observations of water stage with reasonable accuracy for the study cases, especially for study areas with scanty hydrological data.     

基于动态集成的遥感图像分类

鉴于多分类器集成能够获得比单个分类器更好的性能,但是对于支持向量机(support vector ma-chine,SVM),一般的集成方法很难达到效果.特提出了基于局部精度(local accuracy,LA)的动态集成算法.首先,通过多种方法产生个体分类器;其次,利用验证数据集来定义LA,LA用来衡量各个体分类器的权重以及判断是否挑选该个体分类器的标准;最后,在某研究区的遥感图像数据集上进行实验.实验结果表明,动态集成的效果要优于静态集成,特别是异类动态集成效果最好.静态集成只考虑了分类器在训练样本中的表现而没有考虑测试样本的特征,对于动态集成,可以根据测试样本在验证集上的表现来选择个体分类器,因此它展现出更好的性能.     

A study on damage assessment of the scoured bridges

Abstract

Throughout the past decade, scoured bridges in Taiwan have frequently collapsed as a result of heavy rainfall during typhoon seasons. To mitigate the bridge damage caused by scouring, an improved warning system is required. Three fundamental parameters including the water level, the flow velocity, and the scouring depth (SD) are significant for the structural safety of a scoured bridge. Nowadays, these parameters can be successfully detected by advanced monitoring instruments and serve as a database for a damage assessment system to carry out the function of early warning. Therefore, this article aims at developing the damage assessment system for scoured bridges as the core of a warning system.

We have proposed an analytical process as an assessment system. Through finite element analysis with respect to water level and flow velocity, the critical scouring depth with a specific degree of safety factor for a bridge foundation can be determined by choosing among different failure modes. Accordingly, the relationships between various possible sets of water levels, flow velocities and critical SD are established to present the surfaces of structural safety corresponding to different levels of safety factor. The surfaces obtained can then be applied to the warning system.

Based on the proposed assessment process, the collapse of the Shuang-Yuan Bridge caused by Typhoon Morakot in 2009 is discussed in detail. The consequence of bridge collapse is reasonably explained by the analytical surfaces of structural safety based on the reported parameters at the bridge site. In addition, the Dia-Jia-Hsi Bridge, which has implemented advanced monitoring instruments serves as the second case study. The surfaces of structural safety obtained have been used in conjunction with real-time observations of sensitive parameters to display the situation of structural safety at any time. The results obtained in this article benefit bridge engineers, giving a safety-management platform to speed up the decision making process during an emergency.     

Exploring the spatial patterns of disaster impacts caused by Typhoon Morakot

Geographic context is recognized as an important factor in explaining the potential vulnerability and damage from disasters. However, the spatial dimension of disaster damages is not always taken into account in traditional damage assessment. This article aims to provide a comprehensive methodology that incorporates multivariable analysis with a spatial statistical autocorrelation analysis model in order to assess damages caused by Typhoon Morakot. Three watersheds are selected as the case study site with the consideration of complexity among these divergent affected areas. Spatial autocorrelation analysis was utilized to measure localized hotspots of damage. The grouping maps of affected areas were constructed using cluster analysis. The geographic characteristics of these damages were further compared among the three watersheds. Finally, we discuss the underlying factors that result in the variability of spatial patterns associated with damage, and explain how the findings can make a contribution to policy making and recovery plan implementation regarding typhoon risk reduction.     

Inland and coastal flooding: developments in prediction and prevention

We review the scientific and engineering understanding of various types of inland and coastal flooding by considering the different causes and dynamic processes involved, especially in extreme events. Clear progress has been made in the accuracy of numerical modelling of meteorological causes of floods, hydraulics of flood water movement and coastal wind-wave-surge. Probabilistic estimates from ensemble predictions and the simultaneous use of several models are recent techniques in meteorological prediction that could be considered for hydraulic and oceanographic modelling. The contribution of remotely sensed data from aircraft and satellites is also considered. The need to compare and combine statistical and computational modelling methodologies for long range forecasts and extreme events is emphasized, because this has become possible with the aid of kilometre scale computations and network grid facilities to simulate and analyse time-series and extreme events. It is noted that despite the adverse effects of climatic trends on flooding, appropriate planning of rapidly growing urban areas could mitigate some of the worst effects. However, resources for flood prevention, including research, have to be considered in relation to those for other natural disasters. Policies have to be relevant to the differing geology, meteorology and cultures of the countries affected.     

长江流域气象-水文暴雨洪涝预警技术系统研究

对长江流域暴雨洪涝预警研究和业务的4个方面,即暴雨中尺度野外观测科学试验、暴雨中尺度系统形成机理研究、暴雨数值模式预报技术研发以及水文气象耦合模式发展等进行了回顾,分析了制约长江流域暴雨预报和洪涝灾害预警能力和水平提高的主要因素,提出了亟待解决的主要科学技术问题及其应对策略。     

Diagnosing the variance shifts signal in multivariate process control using ensemble classifiers

Statistical process control charts have been successfully used to monitor process stability in various industries. The need to simultaneously monitor two or more quality characteristics has led to the prevalent adoption of multivariate control charts. However, out-of-control signals in multivariate control charts may be caused by one or more variables, or a set of variables. Therefore, effective quality control requires not only the rapid detection of process fluctuations, but also the correct identification of the variable(s) responsible for those changes. This study approaches the diagnosis of out-of-control signals as a classification task and proposes a support vector machine (SVM)-based ensemble classification model focused on variance shifts in multivariate processes. We address the issues of data diversity and ensemble method in constructing an ensemble model. Simulation results demonstrate the effectiveness of the proposed ensemble classification model in identifying the source of variance change. The proposed method clearly outperforms single classifiers as well as other comparable models including bagging and boosting. The results also reveal that the use of extracted features as input vectors for SVM provides better classification performance than the use of raw data. The proposed SVM-based ensemble classification system provides a reliable tool for the interpretation of out-of-control signals in multivariate process control.     

Real-Time and Intelligent Flood Forecasting Using UAV-Assisted Wireless Sensor Network

The Wireless Sensor Network (WSN) is a promising technology that could be used to monitor rivers’ water levels for early warning flood detection in the 5G context. However, during a flood, sensor nodes may be washed up or become faulty, which seriously affects network connectivity. To address this issue, Unmanned Aerial Vehicles (UAVs) could be integrated with WSN as routers or data mules to provide reliable data collection and flood prediction. In light of this, we propose a fault-tolerant multi-level framework comprised of a WSN and a UAV to monitor river levels. The framework is capable to provide seamless data collection by handling the disconnections caused by the failed nodes during a flood. Besides, an algorithm hybridized with Group Method Data Handling (GMDH) and Particle Swarm Optimization (PSO) is proposed to predict forthcoming floods in an intelligent collaborative environment. The proposed water-level prediction model is trained based on the real dataset obtained from the Selangor River in Malaysia. The performance of the work in comparison with other models has been also evaluated and numerical results based on different metrics such as coefficient of determination (), correlation coefficient (), Root Mean Square Error (), Mean Absolute Percentage Error (), and are provided.     

An Optimized Ensemble Model for Prediction the Bandwidth of Metamaterial Antenna

Metamaterial Antenna is a special class of antennas that uses metamaterial to enhance their performance. Antenna size affects the quality factor and the radiation loss of the antenna. Metamaterial antennas can overcome the limitation of bandwidth for small antennas. Machine learning (ML) model is recently applied to predict antenna parameters. ML can be used as an alternative approach to the trial-and-error process of finding proper parameters of the simulated antenna. The accuracy of the prediction depends mainly on the selected model. Ensemble models combine two or more base models to produce a better-enhanced model. In this paper, a weighted average ensemble model is proposed to predict the bandwidth of the Metamaterial Antenna. Two base models are used namely: Multilayer Perceptron (MLP) and Support Vector Machines (SVM). To calculate the weights for each model, an optimization algorithm is used to find the optimal weights of the ensemble. Dynamic Group-Based Cooperative Optimizer (DGCO) is employed to search for optimal weight for the base models. The proposed model is compared with three based models and the average ensemble model. The results show that the proposed model is better than other models and can predict antenna bandwidth efficiently.     

Applicability of multilayer feed-forward neural networks to model the onset of river breakup

Predicting the onset of breakup is an essential component of any ice jam flood forecasting system, yet it presents a difficult challenge due to the complex nature of the relationship between meteorological conditions, streamflow hydraulics and ice mechanics. For this research, data extracted from historical hydrometric and meteorological records were used to develop and assess a three-layer feed-forward artificial neural network (ANN) model for predicting the onset of breakup, using the Hay River in northern Canada as the demonstration site. The calibration results illustrate the potential of the ANN model for successful forecasting of the onset of river ice breakup, i.e. the first transverse cracking of the ice cover. However, rigorous validation also indicates that the accuracy of such ANN models can be optimistically overestimated by their performance during the calibration phase. The possible reasons for this poor predictive capability of the ANN model are also discussed. Despite this caveat, the proposed model shows improved performance as compared to the more conventional multiple linear regression (MLR) techniques typically applied to this problem.     

Development of machine learning‐based real time scheduling systems: using ensemble based on wrapper feature selection approach

There are two items that significantly enhance the generalisation ability (i.e. classification accuracy) of machine learning‐based classifiers: feature selection (including parameter optimisation) and an ensemble of the classifiers. Accordingly, the objective in this study is to develop an ensemble of classifiers based on a genetic algorithm (GA) wrapper feature selection approach for real time scheduling (RTS). The proposed approach can better enhance the generalisation ability of the RTS knowledge base (i.e. classifier) in comparison with three classical machine learning‐based classifier RTS systems, including the GA‐based wrapper feature selection mechanism, in terms of the prediction accuracy of 10‐fold cross validation as measured according to all the performance criteria. The proposed ensemble classifier RTS also provides better system performance than the three machine learning‐based RTS systems, including the GA‐based wrapper feature selection mechanism and heuristic dispatching rules, under all the performance criteria, over a long period in a flexible manufacturing system (FMS) case study.     

Empirical comparison of tree ensemble variable importance measures

Tree ensembles are becoming well-established as popular and powerful data modelling techniques. Tree ensemble models are essentially black box models, although their individual members may not be, and with their growing popularity, interest in the interpretation of tree ensemble models has also grown. This study presents variable importance measures associated with random forests, conditional inference forests and boosted trees, and employs a number of simulated data sets to compare these methods. Overall, variable importance indicators based on bagged conditional inference forests appear to strike a good balance between identification of significant variables and avoiding unnecessary flagging of correlated variables. Data preprocessing and interpretation by experts knowledgeable with a specific data set remain vital.     

Snow Cover Mapping for Mountainous Areas by Fusion of MODIS L1B and Geographic Data Based on Stacked Denoising Auto-Encoders

Snow cover plays an important role in meteorological and hydrological researches. However, the accuracies of currently available snow cover products are significantly lower in mountainous areas than in plains, due to the serious snow/cloud confusion problem caused by high altitude and complex topography. Aiming at this problem, an improved snow cover mapping approach for mountainous areas was proposed and applied in Qinghai-Tibetan Plateau. In this work, a deep learning framework named Stacked Denoising Auto-Encoders (SDAE) was employed to fuse the MODIS multispectral images and various geographic datasets, which are then classified into three categories: Snow, cloud and snow-free land. Moreover, two independent SDAE models were trained for snow mapping in snow and snow-free seasons respectively in response to the seasonal variations of meteorological conditions. The proposed approach was verified using in-situ snow depth records, and compared to the most widely used snow products MOD10A1 and MYD10A1. The comparison results show that our method got the best performance: Overall accuracy of 98.95% and F-measure of 73.84%. The results indicated that our method can effectively improve the snow recognition accuracy, and it can be further extended to other multi-source remote sensing image classification issues.     

基于差分进化的中文情感分类集成算法研究

情感分类是一种从文本中提取情感倾向的文本分类任务。集成学习通过结合几个分类器,在情感分类任务上能够获得比个体分类器更好的分类效果。但是,由于个体分类器在数据集上的表现不同,个体分类器在集成方法中的权重难以确定。针对集成学习中个体分类器的权重优化问题,提出一种基于差分进化优化个体分类器权重的集成分类方法,并将其应用于中文情感分类。以分类准确率为适应度值,通过差分进化算法优化5种个体分类器的权重组合,在3个领域的评论语料集上进行实验。实验结果表明,与一般的集成方法相比,该方法在中文情感分类上有更好的分类效果。     

Optimal scheduling of hydro-thermal power systems considering the flood risk of cascade reservoirs

The increased flood risk linked to global warming affects the safety of cascade reservoirs, with direct effects on the stable operation of power systems. In this article, an optimal scheduling framework for hydro-thermal power systems considering the flood risk of cascade reservoirs is presented. First, the extreme value theory-based peak over threshold model is adopted to build a generalized Pareto distribution of extreme flood inflow for a single reservoir. Then, the Copula function is used to build a joint probability distribution function of extreme inflow for cascade reservoirs during a flood period. Based on the superior performance of the conditional value at risk (CVaR) in characterizing the tail risk of the cascade reservoir spillway safety margin, a CVaR constraint for cascade reservoir flood prevention is proposed, and a scheduling model for hydro-thermal power systems considering the flood prevention risk of head-dependent cascade reservoirs is presented. Secondly, Rockafeller and Uryasey reformulation and sample average approximation are employed to transform the proposed model with a CVaR constraint into a convex solvable optimization model. Finally, a modified IEEE 14-node system is used to verify the better performance of the proposed model than that of the models with the chance constraint and with the independent normal distribution of extreme flood inflow. The impacts of flood prevention confidence level and Monte Carlo sample size on the optimal scheduling results are analysed quantitatively.     

Joint multiple adaptive wavelet regression ensembles

Multiple adaptive discrete wavelet transforms were applied during a multiple regression of spectroscopic data for the purpose of investigating the hypothesis — does the use of different wavelets, at different points, within a spectrum, elucidate predictive capability. The model investigated was a constrained stacking regression ensemble with individual regression models chosen initially by a Bayes Metropolis search. The ensemble approach provided the ability to combine different regression models that used different types of wavelets. Models were applied to a publically available dataset, pertaining to biscuit dough, of near infrared spectra, that were measured by a FOSS 5000, and laboratory measurements of the fat, flour, sugar and moisture content.The resultant model, which is referred to as a joint multiple adaptive wavelet regression ensemble (JMAWRE), was found to be the superior predictive model when compared to models that used standard wavelets as part of the regression ensembles. The JMAWRE was also superior when compared to other models from literature that used the same publicly available NIR dataset.