遥感数据的遗传-超平面分割算法

在传统的采用超平面作为判别边界的分割方法中，使用统计学的方法确定类判别界面，当区域间的统计值大于区域内的统计值时，就会出现类判决界面难于确定的问题。遗传算法的特点是进行全局优化搜索，利用这种算法优势可以解决传统算法难以解决的超平面空间定位问题，从而解决类判决界面的确定问题。通过EOS／MODIS图像数据的分割对该方法作了详细的介绍，其分类精度明显高于统计学分类方法。     

基于RBF神经网络的遥感影像分类算法研究

提出了运用径向基函数神经网络和K均值法进行遥感影像分类的算法，以实际的遥感影像分类为例，通过与传统的最小距离法进行比较，对RBF神经网络分类器的优点进行了归纳，并就算法实施中的一些问题进行了探讨。分析结果表明，RBF神经网络是一种有效的图像分类器。     

2-D deformation analysis of a half-space due to a long dip-slip fault at finite depth

Closed form analytical expressions of stresses and displacements at any field point due to a very long dip-slip fault of finite width buried in a homogeneous, isotropic elastic half-space, are presented. Airy stress function is used to derive the expressions of stresses and displacements which depend on the dip angle and depth of the upper edge of the fault. The effect of dip angle and depth of the upper edge of the fault on stresses and displacements is studied numerically and the results obtained are presented graphically. Contour maps for stresses and displacements are also presented. The results of Rani and Singh (1992b) and Freund and Barnett (1976) have been reproduced.     

岩体工程质量分级应用研究

岩体质量评价采用的测试技术包括地质勘察、弹性波检测和室内岩石力学试验。结合水布垭水利枢纽大坝建基岩体工程质量评价实例,对岩体工程质量分级的应用进行了研究。     

Theoretical Study on Stochastic Modeling of Combined Gravity-Magnetic-Electric-Seismic Inversion and Its Application

As gravity field,magnetic field,electric field and seismic wave field are all physical fields,their object function,reverse function and compound function are certainly infinite contiuously differentiable func-tions which can be expanded into Taylor (Fourier) series within domain of definition and be further reduced in-to solving stochastic distribution function of series and statistic inference of optimal approximation,This is the basis of combined gravity-magnetic-electric-seismic inversion built on the basis of separation of field and source gravity-magnetic difference-value(D-value)trend surface,taking distribution-independent fault sys-tem as its unit,depths of seismic and electric interfaces of interests as its corresponding bivariate compound re-verse function of gravity-magnetic anomalies and using high order polynomial(high order trigonometric func-tion)approximating to its series distribution,The difference from current dominant inversion techniques is that,first,it does not respectively create gravity-seismic,magnetic-seismic deterministic inversion model from theoretical model,but combines gravity-magnetic-electric-seismic stochastic inversion model from stochastic model;second,after the concept of equivalent geological body being introduced,using feature of independent variable of gravity-magnetic field functions,taking density and susceptibility related to gravity-magnetic func-tion as default parameters of model,the deterministic model is established owing to better solution to the con-tradictioc of difficulty in identifying strata and less test analytical data for density and susceptibility in newly explored area;third,under assumption of independent parent distribution,a real modeling by strata,the prob-lem of difficult plane closure arising in profile modeling is avoided,This technology has richer and more detailed fault and strata information than sparse pattern seismic data in newly explored area,successfully inverses and plots structural map of Indosinian discontinuty in Hefei basin with combined gravity-magnetic-electric-seismic inversion,With development of high precision gravity-magnetic and overall geophysical technology,it is certain for introducing new methods of stochastic modeling and computational intelligence and promoting the develop-ment of combined gravity-magnetic-electric-seismic inversion to open a new substantial and promoting the develop-ment of combined gravity-magnetic-electric-seismic inversion to open a new substantial path.     

黄河流域夏季分区面雨量预报研究

介绍黄河流域分区夏季面雨量预报的研究成果，精心挑选51个具有较好代表性的测站对黄河流域夏季降水的时空演变特征进行分析，使用K均值动态聚类对黄河流域的夏季降水进行了客观分区，并计算出各流域夏季面雨量。通过对黄河流域夏季雨量与500hPa环流，海温、OLR、中纬阻高，高原积雪，欧亚积雪等重要影响因子的关系分析，结合黄河流域夏季面雨量年降和年代际演变特征的分析，研究出黄河流域分区夏季面雨量预测的基本方法和模型，并通过客观化的数学方法建立黄河流域夏季面雨量预测系统，预测系统十年回报的结果显示出具有较的预测技巧。     

用非线性多因子动态组合方法作降水概率预报

将逐时段定点降水预报这个以往一般作为非连续量处理的预报问题转化为对一个连续的降水可能函数的预报问题。降水可能函数为一由实况值确定的函数。因为现在模式变化较快，且许多因子和降水可能函数为非线性关系，文章采用一种新的非线性多因子动态组合方法，通过对降水可能函数的预报，作客观分站降水概率预报。     

日照市短时冰雹定时-定点-定量预报

利用1960～2000年青岛、射阳、徐州3站08时探空资料及日照站08时地面气象资料，分析了日照地区出现的16次降雹天气的时空分布特征，在对降雹天气分型的基础上利用积云数值模式计算降雹因子，用降雹因子和单站要素因子建立历史降雹因子个例库，用历史实况资料建立实况个例库，用距离相似法实现日照地区短时冰雹定时、定点、定量的客观预报。