IMPROVED MAN-COMPUTER INTERACTIVE CLASSIFICATION OF CLOUDS BASED ON BISPECTRAL SATELLITE IMAGERY

In this paper,improvement on man-computer interactive classification of clouds based onhispeetral satellite imagery has been synthesized by using the maximum likelihood automaticclustering(MLAC)and the unit feature space classification(UFSC)approaches.The improvedclassification not only shortens the time of sample-training in UFSC method,but also eliminatesthe inevitable shortcomings of the MLAC method.(e.g.,1.sample selecting and training isconfined only to one cloud image:2.the result of clustering is pretty sensitive to the selection ofinitial cluster center:3.the actual classification basically can not satisfy the supposition of normaldistribution required by MLAC method;4.errors in classification are difficult to be modified.)Moreover,it makes full use of the professionals'accumulated knowledge and experience of visualcloud classifications and the cloud report of ground observation,having ensured both the higheraccuracy of classification and its wide application as well.     

水环境质量评价的灰色聚类法

运用灰色聚类法对甘肃金昌市地下水质进行了评价,并与模糊数学方法作了比较。结果表明灰色聚类法不仅具有模糊数学的优点,又弥补了其不完善之处。因此,用灰色聚类法评价水质,方法简便,结论更符合客观实际。     

Clustering stochastic point process model for flood risk analysis

Since the introduction into flood risk analysis, the partial duration series method has gained increasing acceptance as an appealing alternative to the annual maximum series method. However, when the base flow is low, there is clustering in the flood peak or flow volume point process. In this case, the general stochastic point process model is not suitable to risk analysis. Therefore, two types of models for flood risk analysis are derived on the basis of clustering stochastic point process theory in this paper. The most remarkable characteristic of these models is that the flood risk is considered directly within the time domain. The acceptability of different models are also discussed with the combination of the flood peak counted process in twenty years at Yichang station on the Yangtze river. The result shows that the two kinds of models are suitable ones for flood risk analysis, which are more flexible compared with the traditional flood risk models derived on the basis of annual maximum series method or the general stochastic point process theory. Received: September 29, 1997     

Space-time seismicity and earthquake swarms: Certain observations along the slow-spreading mid-Indian Ocean ridges

The slow spreading mid-Indian Ocean ridge system containing the Carlsberg, Central and Southwest Indian ridges is seismically very active. In the present study, a detailed analysis has been carried out of the data of earthquake sources along different ridge segments in order to investigate the spatial and temporal clustering patterns and to evaluate crustal processes related to the swarm occurrences along these ridges. The spatial and temporal clustering pattern of the recent earthquakes (1980–1990) pertaining to nine major spreading segments and eight fracture zones suggests that the events cluster in greater proportion along the spreading segments than along the fracture zones. We performed a systematic search of earthquake catalogue during the period 1964–1990 by examining the spatio-temporal hypocentral clusters in order to identify the swarm occurrences along these ridges. The search included eighteen prominent sequences, of which, thirteen were earthquake swarms. Except two, all other swarms were found to be occurring mainly on the spreading segments. The maximum magnitude observed in these swarms is m_b = 5.4 and have many events predominantly showing normal faulting mechanisms. The spatial disposition and temporal activity of the events in swarms is much similar to the foreshock-mainshock-aftershock sequences observed along the spreading rift valley zones. These characteristics help us to support that swarms along the slow spreading mid-Indian Ocean ridges are the result of extensional tectonic activity, leading to the development of the median valley topography, a mechanism similar to that proposed by Bergman and Solomon (1990) for the Mid-Atlantic Ridge.     

Evaluation of precursory seismic quiescence in sixteen subduction zones using single-link cluster analysis

This paper describes a new method, single-link cluster analysis (SLC), to evaluate percursory quiescence for shallow earthquakes in sixteen subduction zones, using data from the ISC catalog. To define quiescent regions, we divided the catalog into time intervals with a durationT, overlapping byT/2. We considered all earthquakes having magnitudes larger than some magnitudeM _min, lying within a specified distance of a great circle which is approximately coincident with the trench near a subduction zone. Within each time interval we connected or linked all earthquakes lying within some cutoff distanced of one another. We then projected all these links onto the great circle, and defined a region to be quiescent if it was not covered by the projection of any links. For this study,T was two years,M _min wasm _b =4.9, and we variedd from 100 to 400 km. We defined an earthquake as following quiescence if it occurred within two years following, and within 75 km of a quiescent zone as defined above. The primary conclusion of this study was that earthquakes with surface wave magnitudes 7.2 and greater were about 5–15% more likely to follow quiescence than were the smaller background earthquakes withm _b >-4.9. A chi-squared analysis shows that this result is significant at the 99% level. In contrast, earthquakes with surface wave magnitude of 6.7 to 7.1 were no more likely to follow quiescence than were background earthquakes. Of sixteen individual regions, Central America, Japan, and Peru-Chile were the only regions where large earthquakes were more likely to occur following quiescence than were background earthquakes. For a cutoff link length of 300 km, only in Central America was the difference between large earthquakes and background earthquakes significant at the 95% level of significance. For a cutoff link length of 250 km, the significance level exceeded 95% only in Japan. The SLC method is an objective, quantitative method for evaluating large data catalogs, or for monitoring quiescence in regions where quiescence is conjectured to precede large earthquakes.     

Kohonen网络在烟叶动态分类中的应用

针对输入为高维化学指标数据的烟叶分类问题 ,提出 1种改进的 Kohonen自组织特征映射神经网络的聚类方法。在数据预处理时 ,加入了领域专家经验 ,对输入特征向量中的各个分量分配不同的分类参与度 ;用 Gauss邻域函数替代了标准 Kohonen网络的方形邻域 ;在 2个学习阶段学习率和邻域宽度采用了不同的递减函数。通过应用证明了改进后的 Kohonen网络的收敛效果和聚类精度比 K- means聚类方法和标准的 Kohonen网络都有较大的提高。     

中国不同气候区的高温危险性分析

高温危险性评价是高温灾害风险评价的重要组成部分和基础工作。基于过去60年(1961—2020年)的日最高温度数据,选择高温日数、最高温度、高温强度三个指标,逐站点使用核密度函数估算三个高温指标的概率密度函数,计算了四个重现期(即五年一遇、十年一遇、二十年一遇和五十年一遇)下的三个指标的取值,进而通过K-means非监督聚类得到各指标下的高温危险性等级分布图,然后将各指标叠加,对我国不同气候区的综合高温危险性进行了评估。研究表明：(1)从高温日数来看,发生高温可能性最大的地区位于我国西北干旱(半干旱)区和华中、华南地区的交界部分;(2)从最高温度来看,发生高温可能性最大的区域是西北干旱(半干旱)区和华北地区,华中地区的最高温度在不同重现期下的增强最显著;(3)从高温强度来看,发生高温可能性最大的区域是西北干旱(半干旱)区,西南地区和青藏高原地区内不同区域的高温强度差异较大,需局部防范高温灾害;(4)随着重现期年限的增长,不同高温指标值均明显增长,高值区域面积逐渐增加。全国高温综合高危险性等级地区主要位于西北干旱(半干旱)地区、华北地区南部、华中地区、华南地区和西南地区北部,其余气候区少量...     

复杂介质的三维块状模型快速射线追踪

三维复杂介质的射线追踪,是目前在生产实际中应用最广泛的正演模拟技术。这里采用了块状结构的建模方法构造复杂三维地质体,以试射法作为射线追踪的基本算法。在试射过程中,引入了聚类的基本思想进行聚类加密,提高了射线追踪的效率。为了得到光滑连续的地层曲面,采用基于方位角的平滑处理对曲面进行平滑,可以进一步提高三维射线追踪的计算速度。同时,通过对迭代算法的改进,能够保证复杂模型的快速射线追踪。实际地质模型的射线追踪结果证明,改进后的射线追踪方法能够有效地提高复杂介质三维模型的射线追踪速度。     

中国省域科技创新模式及其时空演变

中国的经济增长模式正在由要素驱动型向创新驱动型转变,适宜的科技创新模式是促进中国经济社会协调可持续发展的关键。基于熵值法构建综合指标评价体系,运用Theil指数对中国科技创新投入与产出水平的时空演变进行分析,并结合K-means聚类分析法将其分别归为相同的5个等级。根据不同的投入—产出组合,得出投入产出协调型、投入领先型与产出领先型3个科技创新类型及其子类。最后,利用多阶段DEA模型,从科技创新投入的角度,将科技创新模式划分为混合驱动型、创新平台驱动型、人力与资本驱动型和人力驱动型四类;从科技创新产出的角度,划分为经济创新导向型、知识与经济创新导向型和知识创新导向型三类模式。对比发现,中国科技创新模式与科技创新类型的时空演变呈现高度的相关性。在此基础上,为中国未来区域科技创新的发展提出若干建议。     

犯罪共生空间的类型识别及其特征分析

以ZG市公共空间盗窃、入室盗窃、寻衅滋事、接触诈骗、抢劫抢夺犯罪为研究对象,采用K均值聚类法识别不同类型的犯罪共生空间;并通过决策树模型分析了不同犯罪共生空间的特征。结果表明,ZG市犯罪共生空间可划分为4种类型：① 无犯罪类型共生区;② 公共空间盗窃和接触诈骗犯罪共生区;③ 所有类型犯罪共生区;④ 入室盗窃、寻衅滋事和抢劫抢夺犯罪共生区。城市中各异的社会环境和建成环境产生了不同类型的犯罪机会,而且各类社会环境和建成环境之间存在着条件交互性作用。研究结果为制定犯罪的联合防控策略和实现有限警力的合理布控并且提高执法效率提供了理论基础。