取代芳烃的生物降解性与结构相关性研究

采用量子化学MOPAC－AM1法计算了42种取代芳烃的生成热Hf、分子最高占有轨道能EHOMO、分子量MW、分子总表面积TSA及偶极矩μ。分别采用线性回归分析法和人工神经网络法对所研究化合物的生物降解性参数BOD进行QSBR研究。对训练组而言，线性方法和神经网络法的平均预测误差分别为15．9％和11．4％；而测试组化合物的平均百分误差分别为14．5％和13．0％。无论对于测试组还是训练组，神经网络法的预测都更精确。     

Accounting for geophysical information in geostatistical characterization of unexploded ordnance (UXO) sites

Efficient and reliable unexploded ordnance (UXO) site characterization is needed for decisions regarding future land use. There are several types of data available at UXO sites and geophysical signal maps are one of the most valuable sources of information. Incorporation of such information into site characterization requires a flexible and reliable methodology. Geostatistics allows one to account for exhaustive secondary information (i.e.,, known at every location within the field) in many different ways. Kriging and logistic regression were combined to map the probability of occurrence of at least one geophysical anomaly of interest, such as UXO, from a limited number of indicator data. Logistic regression is used to derive the trend from a geophysical signal map, and kriged residuals are added to the trend to estimate the probabilities of the presence of UXO at unsampled locations (simple kriging with varying local means or SKlm). Each location is identified for further remedial action if the estimated probability is greater than a given threshold. The technique is illustrated using a hypothetical UXO site generated by a UXO simulator, and a corresponding geophysical signal map. Indicator data are collected along two transects located within the site. Classification performances are then assessed by computing proportions of correct classification, false positive, false negative, and Kappa statistics. Two common approaches, one of which does not take any secondary information into account (ordinary indicator kriging) and a variant of common cokriging (collocated cokriging), were used for comparison purposes. Results indicate that accounting for exhaustive secondary information improves the overall characterization of UXO sites if an appropriate methodology, SKlm in this case, is used.     

Source apportionment and source/sink relationships of PAHs in the coastal atmosphere of Chicago and Lake Michigan

Multivariate statistical techniques were used to investigate source apportionment and source/sink relationships for polycyclic aromatic hydrocarbons (PAHs) in the urban and adjacent coastal atmosphere of Chicago/Lake Michigan in 1994–1995. The PAH signatures for the atmospheric particle phase, surface water particle phase and sediments indicate that atmospheric deposition is the major source of PAHs to the sediments and water column particulate phase of Lake Michigan. The PAH signature for the atmospheric gas phase and water dissolved phase indicate an intimate linkage between the lake and its overlying atmosphere. A modified factor analysis-multiple regression model was successfully applied to the source apportionment of atmospheric PAHs (gas+particle). Coal combustion accounted for 48±5% of the ΣPAH concentration in both the urban and adjacent coastal atmosphere, natural gas combustion accounted for 26±2%, coke ovens accounted for 14±3%, and vehicle emissions (gas+diesel) accounted for 9±4%. Each is an identified source category for the region. These results are consistent with the mix of fossil fuel combustion sources and ratios of indicator PAHs.     

Multivariate Statistics of the Pyrolysis Products of High Molecular Components in Pulping Wastewaters to Explain Their Toxicity (6 pp)

Background, Aim and Scope At present, large-scale paper manufacture involves delignification and bleaching by elemental chlorine free (ECF) or totally chlorine free (TCF) processes. The wastewater is purified by secondary treatment (mechanical, chemical and biological) which removes most of the toxic substances from the discharge. However, we found residual toxicity in the high molecular (> 1000 D) matter (HMWM) of the discharge by test of the RET (reverse electronic transfer) inhibition. This fraction consists mainly of polydisperse lignin (LIG) and carbohydrate (CH) macromolecules. Structural units in these molecules are studied by pyrolysis gas chromatography / mass spectrometry (Pyr-GC/MS). In the present work, our aim was to find out those structural units which could explain the RET toxicity of LIG or CH molecules. We compared statistically RET toxicity values of the HMWM samples from treated wastewaters of pilot pulping experiments and intensity variation of the pyrolysis product gas chromatograms of these samples. This application is a novel study procedure. Methods Pyrolysis products (Py-GC/MS results) and inhibition of RET (reverse electronic transport toxicity) as TU50 and TU20 of HMWM (High Molecular Weight Material; Mw > 1000 D) were compared by multivariate statistics. The samples were from laboratory pilot stages of TCF (Totally Chlorine Free) and ECF (Elemental Chlorine Free) manufacture of softwood pulp. Py-GC/MS was done without and with addition of TMAH (Tetra Methyl Ammonium Hydroxide). The name and structure of each abundant fragment compound was identified from its retention time and mass spectrum compared to authentic reference compounds or literature. Four sets of Toxicity Units (TUs) and GC peak areas of the pyrolysis fragments were obtained. The data were normalized by division with LIG (lignin content of each sample). TU values were dependent and the fragment values independent (explanatory) variables in statistical treatments by SPSS system. Separate analyses of correlations, principal components (PCA) and stepwise multiple linear regression (SMLR) were performed from the four sample sets TCF and ECF with and without TMAH. Results and Discussion From the CH fragments, 2-furfural in TCF, and from the LIG fragments, styrene in ECF showed the highest probabilities to originate from source structures of toxicity. Other possible compounds in concern were indicated to be CH fragment 2-methyl-2-cyclopenten-1-one in ECF and LIG fragments 2-methoxy-4-methylphenol, 4,5-dimethoxy-2-methylphenol and 2-methylphenol in TCF.     

Natural forest regrowth as a proxy variable for agricultural land abandonment in the Swiss mountains: a spatial statistical model based on geophysical and socio-economic variables

In many European mountain regions, natural forest regrowth on abandoned agricultural land and the related consequences for the environment are issues of increasing concern. We developed a spatial statistical model based on multiple geophysical and socio-economic variables to investigate the pattern of natural forest regrowth in the Swiss mountain area between the 1980s and 1990s. Results show that forest regrowth occurred primarily in areas with low temperature sum, intermediate steepness and soil stoniness as well as close to forest edges and relatively close to roads. Model results suggest that regions with weak labor markets are favored in terms of land abandonment and forest regrowth. We could not find an effect of population change on land abandonment and forest regrowth. Therefore, we conclude that decision makers should consider non-linearities in the pattern of forest regrowth and the fact that labor markets have an effect on land abandonment and forest regrowth when designing measures to prevent agricultural land abandonment and natural forest regrowth in the Swiss mountains.     

Exploratory Analysis and a Stochastic Model for Humusdisintegration

Ulrich (1981) supposes in the hypothesis of humusdisintegrationthat the balance between polymerisation and breakdown of organicmaterial may be disturbed in chemically well buffered Europeanforest soils. This new aspect of aluminium toxicity may causenitrogen exceedance in forest ecosystems and subsequent seasonalnitrate output (Eichhorn and Hütterman, 1999).In a research program the substances in the seepage water aremonitored in a small woodland in central Germany. We explorethese multivariate data for examining possible influences on theprocess of humusdisintegration and its temporal evolution. As aresult, a regression model for carbon is developed, whichincludes covariables, i.e., other substances, as well as spatialand temporal terms describing systematic variability. Especiallyiron and aluminium turn out to be very influential in the model.So far our work is a basic step for monitoring the seepage waterdata by means of stochastic modelling.     

Fitting Data to Linear Combinations of Covariograms

We pursue a regression model for spatially-indexed data whose spatial correlation is determined by a linear combination of simple covariograms. The main interest lies in the estimation of the spatial parameters. As several common techniques appear ineffective for this setting, an algorithm is proposed to obtain parameter estimates and is assessed through simulation. It is found to provide greater stability than other methods of estimation. We discuss the influence of parametrization and site location on the efficacy of the estimation algorithms, and develop some guidelines as to the placement of sampling sites to improve the algorithm's performance. This revised version was published online in July 2006 with corrections to the Cover Date.     

校准曲线回归计算方法的比较

本文对目前国内所使用和已提出讨论的校准曲线回归计算方法与国际标准校准曲线回归计算方法进行了比较和讨论,表明无论采用何种回归方法进行计算,对测定结果的精密度和准确度均无显著影响。但为了使校准曲线的回归计算方法国际化、标准化,建议采用国际标准回归法。     

中国青年女性血沉参考值与地理因素的关系

为制定中国青年女性血沉参考值的统一标准提供科学依据,本文收集了中国各地用温氏法测定的健康青年女性血沉参考值,并对其与地理因素的关系进行了研究,发现海拔高度是影响青年女性血沉参考值最主要的因素,随着海拔高度的逐渐增大,青年女性血沉参考值在逐渐的减小,相关性很显著。用逐步回归分析的方法推导出了一个回归方程。如果知道了中国某地的地理因素,就可以用回归方程估算这个地区的青年女性血沉参考值。依据青年女性血沉参考值与地理因素的依赖关系把中国分为青藏区,西南区,西北区,东南区,华北区,东北区等六个区     

CART-分类与回归树方法介绍

分类与回归树（CART－ClassificationandRegressionTrees）是一种非常有趣并且十分有效的非参数分类和回归方法．它通过构建二叉树达到预测目的。该方法是四位美国统计学家耗时十多年辛勤劳动的成果。在他们所著的“ClassificationandRegressionTrees（1984）”一书中有该方法的详细说明。笔者认为该方法在古生物化石分类和鉴定以及矿产资源预测等方面颇有用武之地。本文只对该方法作简单介绍,希望能对从事预测工作的读者有所启发。美国加利福尼亚统计软件公司已歼发了CART方法软件。