【摘要】 介绍了分形理论和动力系统重构等现代数学方法和神经网络、模糊聚类、遗传算法等信息处理技术在地球科学稀疏、散乱数据中的优化插值与拟合逼近,以及资料中缺损信息恢复重构等方面的学科动态和研究进展;分析评述了多种数据处理方法的优缺点及其在不同类型、不同特性资料处理中的适用性和局限性;阐述了基于学科交叉融合、优势互补研究思想开展地学资料中非常规复杂数据的信息挖掘、特征提取和优化拓展的技术途径和应用实践;讨论了地球科学中两类重要的数据分析研究思想(基于动力模型约束的变分方法和基于自然几何相似的同化方法),指出它们的交叉融合和优势互补是未来地球科学数据研究中重要的发展趋势和方向。更多还原

【Abstract】 The trends and progress of geoscience research about optimized interpolation and(fitting) method on sparse,scattered data and information recovery from imperfect data by using the modern mathematic method(such as fractal theory and reconstructing dynamic model) and information technology(such as artificial neuron-network,fuzzy clustering and genetic algorithm) are introduced and described.The different advantages and shortcomings of the current data treatment method and their applicability of the localization on different types,and different speciality data are analysed and reviewed.The technology route and application practice on abnormal and complicated geoscience data mining,characteristic extraction and information optimized extension based on the study idea of subject-intercross and mutual advantage-fusion are expounded and discussed. Finally,two kinds of important data analysing ideas of geoscience research(the variational method based on dynamic model restriction and the assimilation method based on natural geometry analogical similitude) are presented and discussed,and their cross-fusion and advantages interchanged are acknowledged as the important development trend and direction of geoscience research in the future.更多还原