基于随机森林评价的兰州市主城区校园地表灰尘重金属污染

将2018年1~12月兰州市主城区校园地表灰尘重金属元素含量计算得到的综合污染指数（P_N）和潜在生态风险指数（RI）作为训练集,使用11个影响地表灰尘重金属污染和积累的特征参数,利用随机森林算法对信息采样点的P_N、RI进行估算,分析了地表灰尘重金属污染的时空变化特征,并对传统算法插值结果和随机森林插值结果进行了比较.结果表明,研究区地表灰尘重金属各元素浓度均高于本地背景值;研究区P_N排序为城关区 > 西固区 > 安宁区 > 七里河区,RI排序为城关区 > 西固区 > 七里河区 > 安宁区,P_N和RI在空间分布特征上很相似,都位于交通枢纽或市中心;P_N在冬季和夏季出现高值,RI高值则出现在冬季,冬季高值主要源于采暖燃煤源的增加;空间插值结果对比表明随机森林插值结果优于传统算法插值结果.     

Uncertainty propagation in vegetation distribution models based on ensemble classifiers

Ensemble learning techniques are increasingly applied for species and vegetation distribution modelling, often resulting in more accurate predictions. At the same time, uncertainty assessment of distribution models is gaining attention. In this study, Random Forests, an ensemble learning technique, is selected for vegetation distribution modelling based on environmental variables. The impact of two important sources of uncertainty, that is the uncertainty on spatial interpolation of environmental variables and the uncertainty on species clustering into vegetation types, is quantified based on sequential Gaussian simulation and pseudo-randomization tests, respectively. An empirical assessment of the uncertainty propagation to the distribution modelling results indicated a gradual decrease in performance with increasing input uncertainty. The test set error ranged from 30.83% to 52.63% and from 30.83% to 83.62%, when the uncertainty ranges on spatial interpolation and on vegetation clustering, respectively, were fully covered. Shannon’s entropy, which is proposed as a measure for uncertainty of ensemble predictions, revealed a similar increasing trend in prediction uncertainty. The implications of these results in an empirical distribution modelling framework are further discussed with respect to monitoring setup, spatial interpolation and species clustering.     

基于机器学习分类算法解析EIS数据的有机涂层性能评价方法

目的 基于机器学习分类算法快速评估有机涂层的防腐性能。方法 通过实验室加速试验模拟涂层真实的退化过程,并根据测得的电化学数据,分析不同退化阶段的等效电路元件参数。随后,采用随机抽样方法获取大量数据,用于机器学习模型训练。通过对比支持向量机(SVM)、k最近邻(k-NN)和随机森林(RF)3种不同的机器学习算法,以及多种输入特征集训练的涂层性能分类器模型的准确率,分析最适合用于涂层性能快速评估的机器学习算法和电化学特征。结果 根据不同输入特征训练的k-NN和RF模型均表现出良好的预测效果,而SVM模型的预测效果相对较差。根据不同频率范围训练的分类器模型中,在低频区表现最佳,而在高频区表现较差。结论 基于阻抗虚部、虚部+实部和阻抗模值3种输入特征训练的RF分类器模型的预测效果最准确。不同频率区间内,低频区的阻抗特征更能准确表征涂层性能。     

汽车外部照明装置振动前后配光图比较

根据详细的振动要求,我们测试了某款汽车外部照明装置在振动前后的配光数据变化,根据测试结果得知灯座和灯体一体化结构的灯具可以承受汽车的振动等情况而保持配光性能不变.     

Structurally-consistent estimation of use and nonuse values for landscape-wide environmental change

We address the problem of estimating the use and nonuse value derived from a landscape-wide programme of environmental change. Working in the random utility framework, we develop a structural model that describes both demand for recreational trips to the landscape's quality-differentiated natural areas and preferences over different landscape-wide patterns of environmental quality elicited in a choice experiment. The structural coherence of the model ensures that the parameters of the preference function can be simultaneously estimated from the combination of revealed and stated preference data. We explore the properties of the model in a Monte Carlo experiment and then apply it to a study of preferences for changes in the ecological quality of rivers in northern England. This implementation reveals plausible estimates of the use and nonuse parameters of the model and provides insights into the distance decay in those two different forms of value.     

Differential use of stored sperm during oviposition in the damselfly Calopteryx splendens xanthostoma (Charpentier)

Female Calopteryx splendens xanthostoma (Charpentier) (Odonata), have two discrete sperm storage organs, the bursa copulatrix and the paired spermathecae. We used the Random Amplified Polymorphic DNA(RAPD) profiling technique to determine from which sperm storage organ spermatozoa were used to fertilise eggs during behaviourally distinct oviposition bouts. During oviposition bouts following remating sperm from the bursa copulatrix are largely used to fertilise eggs, but when females avoid remating before an oviposition bout the sperm used to fertilise eggs are derived mainly from the paired spermathecae. Our observations suggest that female C. s. xanthostoma can avoid male paternity assurance mechanisms by a combination of behavioural and anatomical adaptation: the reason(s) they do so is unclear. Received: 28 November 1995/Accepted after revision: 5 August 1996     

RAPD markers suggest genotypic effects on forager specialization in a eusocial wasp

Genetic variability within insect societies may provide a mechanism for increasing behavioral diversity among workers, thereby augmenting colony efficiency or flexibility. In order to assess the possibility that division of labor has a genetic component in the eusocial wasp Polybia aequatorialis, I asked whether the genotypes of workers within colonies correlated with behavioral specialization. Workers specialized by foraging for one of the four materials (wood pulp, insect prey, nectar, or water) gathered by their colonies. I collected foragers on 2 days from each of three colonies and identified the material the foragers were carrying when collected. I produced random amplified polymorphic DNA (RAPD) markers from the genomic DNA of these foragers and estimated genotypic similarity of foragers based on sharing of variable RAPD marker bands. Contingency tests on 20 variable loci per colony showed statistically significant (P <0.05) biases in RAPD marker frequencies among forager types in the three colonies. Patterns of association of RAPD marker bands with specializations were constant in two colonies, but changed between collection days in one colony. RAPD marker biases suggest that division of labor among workers includes a genetic component in P. aequatorialis. Colony-level selection on variation in division of labor is a possible factor favoring the evolutionary maintenance of high genotypic variability (low relatedness) in epiponine wasp colonies and in other eusocial insects. Received: 18 July 1995/Accepted after revision: 1 October 1995     

Drink driving deterrents and self-reported offending behaviours among a sample of Queensland motorists

Problem

The efficacy of drink driving (driving under the influence, DUI) countermeasures to deter motorists from driving over the legal limit is extremely important when considering the personal and economic impact the offending behavior has on the community. This paper reports on an examination of 780 Queensland motorists' perceptions of legal and non-legal sanctions and their deterrent impact on self-reported offending behavior. Method: The data were collected via a telephone survey of motorists recruited from a random sample of all listed telephone numbers in the state, adjusted according to district population figures. Results: The results indicated that there were a range of legal and non-legal factors that were significantly associated with self-reported DUI including: the perceived risk of apprehension and license loss (legal factors); and concerns relating to the possibility of being involved in a crash and hurting another person (non-legal factors). However, additional multivariate analyses indicated that while both legal and non-legal factors significantly predicted self-reported DUI, higher alcohol consumption levels and more favorable attitudes to the behavior also appear to increase the likelihood of DUI. Discussion: The paper will outline the direct implications of the research project such as the development and promotion of countermeasures that both effectively deter motorists and address pro-offending attitudes. Impact on industry: The findings also highlight that DUI remains a relatively common behaviour among some motoring groups and that there is a need to extend current levels of enforcement as well as adopt innovative strategies to enhance the impact of these operations on the offending behaviour.     

Zooplankton encounters in patchy particle distributions

Zooplankton encounter rates are dependent not only on both sensory and swimming performances of the organisms, but also on the distribution pattern of food particles. Increasing evidences indicate that, in natural conditions, phytoplankton is often aggregated in thin layers. In the present contribution we investigate the concomitant effects of motion complexity and habitat fragmentation on the number of encounters realised by virtual continuously moving copepods adopting different motion strategies. Our simulated organisms move in an environment characterised by the presence of thin patches of phytoplankton, and their swim follows five motion rules (pure random walk, correlated random walk with three different time scales, self-avoiding random walk), each characterised by a typical value of the three-dimensional fractal dimension D_3D. Compared to a uniform distribution, for a given motion rule the clustering of prey particles increases the variance of encounters, while no remarkable effect is reported in the average number of particles intercepted. These results broaden our understanding of the behavioural efficiency in freely swimming zooplankters and improve our knowledge of the functioning of aquatic systems.     

Incorporating behaviour into simple models of dispersal using the biological control agent Dicyphus hesperus

We explored the utility of incorporating easily measured, biologically realistic movement rules into simple models of dispersal. We depart from traditional random walk models by designing an individual-based simulation model where we decompose animal movement into three separate processes: emigration, between-patch movement, and immigration behaviour. These processes were quantified using experiments on the omnivorous insect Dicyphus hesperus moving through a tomato greenhouse. We compare the predictions of the individual-based model, along with a series of biased random walk models, against an independent experimental release of D. hesperus. We find that in this system, the short-term dispersal of these insects is described well by our individual-based model, but can also be described by a 2D grid-based biased random walk model when mortality is accounted for.