Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 263-273(2004)
The most pressing global problem nowadays is to identify a sustainable vision of human activities under the conditions of limited natural resources and limited environmental capacity of the earth. Also the divide between industrialized society and developing regions is expanding and getting much more difficult to overcome. For understanding and finding a solution to diversified and complicated problem, statistical handling of numbers of issues by using quantifiable parameters and by meaningful treatment are essential to clearly define policy targets. The present paper shows what types of approaches are currently adopted in this area and invites statistical scientists involve themselves in this crucial approach.
Key words: Sustainability, sustainable society, indicator, genuine progress indicator, environment, Millenium Declaration Goal.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 275-279(2004)
With the development of various information technologies in measurement, communication and computation, prediction and knowledge discovery based on massive data sets have become crutial problems. However, "globalization" has increased the uncertainty of society, and decisions based on proper evaluation of risk has become important. To solve these two important problems, statistical modeling is indispensable, and it also reveals some challenges in research on statistical science.
Key words: State space models, information criterion.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 281-295(2004)
This paper overviews the possible future development of statistical data on environmental issues collected and summarized by the Ministry of the Environment in Japan. Considering the necessity for various statistical environmental data in the analysis and formulation of environment preservation policy planning, the Ministry of the Environment has published a comprehensive statistical data compendium as an annual report since 2002. The compilation of environmental state data (e.g. monitoring data of pollutants in ambient air and public water) should be enhanced in order to make it easier to use these data sets in policy analysis. Furthermore, increasing environmental consciousness has lead to the formulation and implementation of voluntary initiatives for reducing environmental pressures by industries. The environmental reports of industries and the Pollutant Release and Transfer Register System have enabled the public to access statistical data on pollutants discharged from the whole business entity and from individual industries. These data could be fully utilised in considering new formulation of policy planning to combat environmental degradation. This paper also discusses a statistical trial on the setting of Environment Quality Standard for sediment polluted by dioxins.
Key words: Environment, monitoring data, emission, greenhouse gas, dioxins, GIS.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 297-307(2004)
Polycyclic Aromatic Hydrocarbons (PAHs) released into the road environment from car exhaust gas are known to have a potential carcinogenic risk. In this study, we carried out a statistical assessment of PAH distribution in the road environment by using Azalea leaves as samples for air monitoring. The Azalea leaves were collected from 13 roadside points from September to December, 2003 and examined for PAHs by gas chromatography/mass spectrometry (GC/MS). The results showed that the PAH content correlated with traffic density. It was found in particular that the contents of Phenantherene, Fluoranthene and Pyrene, which seemed to be derived from car exhaust gas, were higher and had stronger correlations with traffic density. From this date, we inferred that the content of PAHs in Azalea leaves is in proportion to traffic density and that PAH pollution in Azalea leaves is caused by car exhaust gas. In order to prove this hypothesis, we classified the PAH profile in Azalea leaves using a Cluster Analysis. It was revealed that this profile classification corresponds with the traffic density classification. Therefore, the PAH profile in Azalea leaves is directly correlated to the traffic density.
Key words: Road environment, Polycyclic Aromatic Hydrocarbons, Cluster Analysis.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 309-327(2004)
This paper discusses the relationships between environmental factors and polycyclic aromatic hydrocarbons (PAH) contained in the exhaust gases on roads. We collected road sediments at seven stations contaminated by car exhaust gases in Okayama prefecture. As the amounts of rain and ultraviolet rays were not observed in the specific area, we estimated them by Tiesen and latitude adjusted methods. In addition, the areas and observed periods of the seven stations differed, so that data were normalized for joint analysis. We considered a multivariate linear regression model to predict PAH by means of exploratory variables for environmental factors like road run-off due to rain and irradiation of ultraviolet rays. Furthermore, since few data were collected at each station, we detected a cluster of road stations with similar linear structures and the influence the environmental factors was quantitatively evaluated in the cluster.
Key words: Road sediment, contamination, polycyclic aromatic hydrocarbons, environmental factors, irradiation of ultraviolet rays, multivariate linear regression model.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 329-342(2004)
A simple method for finding target refractory chemical-degrading bacteria statistically from a mixed culture system was developed by combination of quinone profiling and DGGE banding pattern. Techniques for identifying degrading bacteria species by applying two kinds of biomarker (16S rRNA and microbial quinone) are still limited. To statistically evaluate the change of DGGE band pattern, a matrix based on DGGE band intensities has been made by digitalizing DGGE band pattern. Refractory chemical-degrading bacteria species were identified by cluster analysis with repeatedly measured data such as specific degradation rate, mole fraction of quinone species and DGGE band intensity. This approach has been applied to analysis of microbial acclimation to dimethylformamide (DMF) in biological wastewater treatment experiments.
Key words: Cluster analysis, DGGE, dimethylformamide, microbial community structure, quinone profile, submerged biofilter reactor.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 343-352(2004)
Activated sludge processes are the most widely used method for treating wastewater. However, only limited information is available on microbial community activities in activated sludges. In recent years, use of a technique using quinone profiles has increased as a simple and useful tool for analysis of microbial population dynamics in mixed cultures. Microbial respiratory quinones are components of the bacterial respiratory chain and play an important role in electron transfer during microbial respiration. Quinones exist in almost all bacteria, and in general, one species or genus of bacteria has only one dominant type of quinone. Thus, the quinone profile, which is usually represented as the mole fraction of each quinone type, should be specific for a microbial community. Changes in a microbial community in a mixed culture of microbes could effectively be quantified using quinone profiles. In this study, the technique of quinone profiles was applied to clarify seasonal change in the microbial community in an activated sludge process treating wastewater. We tried to analyze the data of quinone profiles by descriptive multivariable analysis. Hierarchical cluster analysis and a clustering method based on homogeneity are applied in analyzing seasonal change of quinone profiles. In this study, clusters of quinones that have seasonal change could not be specified. However, it is suggested that the quality of the data needs to be determined.
Key words: Activated sludge, quinone profile, seasonal change, descriptive multivariate analysis, clustering method based on homogeneity.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 353-365(2004)
The detection limit for the Ames Salmonella mutagenicity assay, which is very widely used all over the world to evaluate the genetoxicity of environmental samples, was determined in the present study. Salmonella typhimurium TA100 strain was used and exogenous activation S9 was not used in the study, because these test conditions are sensitive to detection of mutagen in chlorinated water samples. To determine the detection limit, 100 accumulated results from the assay conducted by two experimenters A and B were taken as examples. The total number of base-agar plates used in the assay was up to 10,702. The distribution function was considered to be a normal distribution by the chi-squared test, so that the two sample t -test was employed to determine it. The mean and variance of the negative control test conducted by using 45 base-agar plates (45 recurrences) were 127 rev. plate-1 and 117 rev. plate-1, respectively. The detection limit was determined at 1.7 as the MR value (the number of revertant colonies in the sample test divided by those in the negative control test) at a level of significance of 0.05 when duplicate plates were used in the negative control test. However, it decreased to 1.4 as MR level when quadruple plates were used in the negative control test. Therefore, it was found that the sensitivity of the Ames Salmonella mutagenicity assay was improved very easily by increasing the number of plates in the negative control test from two to four. The application of the conventional two-fold rule to the data obtained with the strain TA100 was considered too conservative. It was proved by comparing the data of the two students that the detection limit determined in the present study was acceptable to well trained students. However, the accurate detection limit should be determined for each experimenter.
Key words: Two sample t -test, mutagenicity, Ames Salmonella assay, TA100 strain, environmental sample.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 367-380(2004)
A discrete distribution converging on a certain continuous distribution is not necessarily unique. For example, a discrete distribution should be essentially assumed to be a population distribution from the viewpoint of description of data even if the inverse Gaussian distribution is assumed as a population distribution for convenience and data analysis is conducted. As assumed for discrete distribution, the generalized Poisson distribution and the generalized negative hypergeometric distribution can serve as a candidate in this case. A generalized negative hypergeometric distribution is newly proposed. The generalized beta distribution and the generalized power inverse Gaussian distribution are obtained as limiting distributions in this distribution.
Key words: Generalized negative hypergeometric distribution, generalized power inverse Gaussian distribution, coefficient of variation.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 381-391(2004)
Recently, Genetic Algorithm (GA) is recognized as a useful optimization tool because of its high performance. However, a major drawback of GA is its heavy computational load. This paper deals with the application of GA to grid computing. It discusses several features of grid computing, and then introduces an implementation of a GA, Genetic algorithm with Search area Adaptation (GSA), on a computational grid. It is shown through numerical experiments that the performance of GSA is proportional to the number of server computers and it is superior to a parallel simulated annealing algorithm.
Key words: Genetic algorithm, grid, distributed computation, parallel computation, optimization.
Proceedings of the Institute of Statistical Mathematics Vol. 52, No. 2, 393-405(2004)
Belief propagation (BP) is a universal method of stochastic reasoning. It gives exact inference for stochastic models with tree interactions, and works well even if the models have loopy interactions. Its performance has been analyzed separately in many fields, such as, AI, statistical physics, information theory, and information geometry. The present paper provides a unified framework for understanding BP. The stability of BP is analyzed from this framework, and its approximation accuracy is investigated.
Key words: Belief propagation, information geometry, graphical model.