复杂区域多目标总体估计空间抽样优化研究

项目名称： 复杂区域多目标总体估计空间抽样优化研究

项目编号： No.41301425

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 天文学、地球科学

项目作者： 胡茂桂

作者单位： 中国科学院地理科学与资源研究所

项目金额： 22万元

中文摘要： 空间抽样是认识和掌握地物对象的一种重要方法, 合理的样本布设方法既能大大节省成本，又可以显著提高抽样调查的效率。在复杂区域调查中，利用一套样本点同时调查环境、生态、气象等多个目标是实际应用中经常面对的重要问题。经典抽样方法未充分考虑到样本空间布局对估计精度的影响，且对空间多指标、多约束调查难以给出合理的解决方案。本研究致力于复杂区域的多指标、多约束空间抽样优化方案研究。首先，针对复杂区域抽样总体估计，提出一种融合空间趋势和空间异质的总体估计模型，建立样本布局与总体估计精度之间的理论关系。在此理论上，结合Pareto多目标优化模型，发展一种新的多指标、多约束空间抽样调查样本设计优化方法，分别得到在指定样本量、指定部分指标估计精度和自适应样本选择的三级空间抽样方案。最后，结合GIS的可视化分析，为决策者提供一个从优化样本方案结合中选择最终样本方案的有效方法。

中文关键词： 空间抽样；空间自相关；异质性；空间趋势；多目标

英文摘要： Spatial sampling is one of the most important methods to know the geographic phenomena and objects. A good sampling scheme can not only save costs but also increase the sampling accuracy and efficiency. It is a common and important problem in real complicated region sampling that multiple objects are expected to be investigated with just one sampling scheme. For example, the objects usually involve environment, ecology, meteorology, and so on. Classical sampling methods have been applied in a great number of domains. However, they do not take spatial autocorrelation of the sampling objects into account for spatial investigation. Furthermore, it is hard for classical sampling methods to investigate multiple objects under multiple constraints simultaneously with just one scheme. In this research, a new spatial sampling optimization method specified for multi-objects and multi-constraints in complicated regions will be proposed. First of all, for a complicated spatial region, the relationship between spatial distributions scheme of samples and the standard error of population estimation will be established based on a sampling model which will fuse both spatial autocorrelation and spatial heterogeneity of the region. Then a multi-objects spatial sampling optimization method will be developed with the Pareto optimiza

英文关键词： spatial sampling；spatial autocorrelation；heterogeneity；spatial trend；multi-objective