A computer method to calculate the convolution of statistical distributions

It is becoming increasingly important to determine probability distributions of combinations of random variables. Convolution is a technique by which the distribution of a sum of random variables can be determined. This paper presents some simplifications in order to reduce the numerical integrations and computer time. In addition, the method may be used with empirical nonanalytic distributions. While Monte Carlo methods are also appropriate for calculating the distribution, convolution can give at least as much accuracy as Monte Carlo methods with a reduction in computation. Two applications are presented: one approximates the distribution of percent sand in an area, and the other indicates a method of determining sample size when using the distribution of means to approximate normality.     

Estimation of Scalar Source/Sink Distributions in Plant Canopies Using Lagrangian Dispersion Analysis: Corrections for Atmospheric Stability and Comparison with a Multilayer Canopy Model

Source/sink distributions of heat, water vapour andCO₂ within a rice canopy were inferred using aninverse Lagrangian dispersion analysis and measuredmean profiles of temperature, specific humidity andCO₂ mixing ratio. Monin–Obukhov similarity theorywas used to account for the effects of atmosphericstability on _w(z), the standard deviation ofvertical velocity and _L(z), the Lagrangian timescale of the turbulence. Classical surface layer scaling was applied in the inertial sublayer (z > z_ruf)using the similarity parameter = (z - d)/L, where z is height above ground, d is the zero plane displacementheight for momentum, L is the Obukhov length,and z_ruf 2.3h_c, where h_c iscanopy height. A single length scale h_c, was usedfor the stability parameter 3 = h_c/L in the height range 0.25 < z/h_c < 2.5. This choice is justified by mixing layer theory, which shows that within the roughness sublayer there is one dominant turbulence length scaledetermined by the degree of inflection in the windprofile at the canopy top. In the absence of theoretical or experimental evidence for guidance,standard Monin–Obukhov similarity functions, with = h_c/L, were used to calculate the stabilitydependence of _w(z) and _L(z) in the roughness sublayer. For z/h_c < 0.25 the turbulence length and time scales are influenced by the presence of the lowersurface, and stability effects are minimal. With theseassumptions there was excellent agreement between eddycovariance flux measurements and deductions from theinverse Lagrangian analysis. Stability correctionswere particularly necessary for night time fluxes whenthe atmosphere was stably stratified.The inverse Lagrangian analysis provides a useful toolfor testing and refining multilayer canopy models usedto predict radiation absorption, energy partitioningand CO₂ exchanges within the canopy and at thesoil surface. Comparison of model predictions withsource strengths deduced from the inverse analysisgave good results. Observed discrepancies may be dueto incorrect specification of the turbulent timescales and vertical velocity fluctuations close to theground. Further investigation of turbulencecharacteristics within plant canopies is required toresolve these issues.     

我国北方地区雷电活动的时空特征

利用地闪定位仪（ＤＦ）探测了我国大兴安岭（东北地区）和北京地区的雷电活动。文中详细分析了地闪的空间分布,并讨论了不同地区各雷电参量的时间变化规律。两地区雷电活动的季节变化与日变化均有较大差异。这些差异的产生可能与本地区的天气背景及地理特征有密切的联系。要较深入地了解上述差异产生的可能原因,将雷电活动与天气因子的给合讨论是必需的。     

转移概率预报方法在霜冻预报中的应用

本文在简介转移概率预报方法及最大概率预报原理的基础上,分析了吕梁市1957年～2000年霜冻初、终日期各随机时间序列状态之间的转换规律,结合各序列的初始分布,建立预报模型,并对1995年～2004年的初、终霜日期进行了预报效果检验。结果表明,该方法预报效果好,使用性强,具有良好的应用前景。     

Earthquake catalogues and modelling strategies. A new testing procedure for the comparison between competing models

In this paper we consider the procedures that, on the basis of an earthquake catalogue, yield the magnitude distribution function F _M (F _M generators). In particular, our attention is focused on the F _M generators that are currently used in the frame of the probabilistic seismic hazard analysis at a site. From an engineering point of view, the behaviour of F _M in the range of strong earthquakes is of crucial importance. On the other hand, in general, the statistical validation of F _M in that range is not feasible because of an insufficient number of strong earthquakes in available catalogues.Our investigation is limited to the problem of the comparison between competing F _M generators. We show that a simple empirical F _M generator (called the CFP generator) can be more reliable than generators based on mathematical models. The numerical experiments that support this statement regards two test-sites and two mathematical models of magnitude distribution, namely the truncated-exponential and the characteristic-type models.The main conclusion is the suggestion of a test (called the CFP-test) that can be summarized as follows. Let X be a site for which, with current criteria, a specific F _M ^X generator based on a mathematical model has been selected. At this point, following the comparison approach described in the paper, the F _M ^X generator is compared with the empirical CFP generator. It can happen (actually it happens in many of our numerical experiments) that this test indicates the CFP generator as more reliable than the selected F _M ^X generator.     

Hybridization of the probability perturbation method with gradient information

Geostatistically based history-matching methods make it possible to devise history-matching strategies that will honor geologic knowledge about the reservoir. However, the performance of these methods is known to be impeded by slow convergence rates resulting from the stochastic nature of the algorithm. It is the purpose of this paper to introduce a method that integrates qualitative gradient information into the probability perturbation method to improve convergence. The potential of the proposed method is demonstrated on a synthetic history-matching example. The results indicate that inclusion of qualitative gradient information improves the performance of the probability perturbation method.     

海南省澄迈县雷暴气候特征及其灾害防御

通过统计分析海南省澄迈县1959-2004年逐日雷暴资料,找出雷暴发生时空分布特征,统计1995-2004年10a雷暴发生的影响天气系统,计算出各天气系统影响下雷暴发生的概率。分析结果表明：澄迈县雷暴日46a平均雷暴频次437次,年际变化呈波动减少趋势,5-8月是雷暴发生的高发期,占全年雷暴的53%,16-17时是一天中发生雷暴的最高期,西南方向发生的雷暴略多于其他方向,澄迈县受西南低压槽、南海低压槽和副热带高压等天气系统影响时发生雷暴的概率较大。     

江苏-南黄海地区未来地震形势的统计分析

根据１５００年以来的地震资料,应用数理统计、灰色预测、干支６０周期分析等多种方法,对江苏—南黄海地区未来３～５年地震形势进行分析和预测,结果表明：该区目前处于本世纪第２活跃幕中后期,未来几年内依然存在５～６级地震的危险性,至２０００年７月,累积发震概率将达到０．７～０．８,１９９９—２０００年、２００２—２００３年均有可能发生５级以上中强地震。     

Fractals and the accuracy of geographical measures

The problems of estimating line length, area, and point characteristics in the earth sciences have generated substantial but independent literatures. All three problems are of increasing concern given the current interest in digital capture, processing, and the storage of geographically referenced data. In the case of qualitative maps, all three are shown to be related to Mandelbrot's fractional dimension D (Mandelbrot, 1977) which allows the dependence of each on sampling density to be predicted. The general results are confirmed by simulation on surfaces of constant D. They also imply that certain improvements can be made in a number of previously proposed methods.