【作者】 刘巍；

【导师】 金炜东；

【作者基本信息】 西南交通大学 ， 信号与信息处理， 2012， 博士

【摘要】 ARGO资料作为目前唯一实时/延时可用的全球尺度海洋三维观测资料,为分析揭示海洋环境温盐分布和流场结构、理解和深化海洋环境特性和气候变化规律提供了关键的观测事实基础,其蕴涵的丰富信息资源具有重要的科学意义和应用价值。论文针对ARGO资料稀疏、零散、缺损等空间不规则和时间不连续固有弱点和应用盲区,开展了ARGO稀疏资料插值、缺损数据拟合和三维温盐场重构等客观再分析研究；开展了基于卫星遥感资料和ARGO资料的海表流场估算以及三维流场的反演研究。为发掘和优化利用ARGO信息资源及ARGO资料再分析应用提供理论依据和技术手段；为深化海洋环境演变与气候变化研究,改进提高大气、海洋环境要素的数值预报能力提供数据信息支持。论文主要成果和创新点包括：(1)针对ARGO等稀疏资料的小尺度结构和细节特征刻画等实际问题,改进和发展了分形插值方法模型,通过引入遗传算法对分形插值参数(压缩因子)优化搜索,实现了插值参数的客观优选,改进了分形插值方法对海洋稀疏要素场插值的合理性和有效性。(2)提出了一种小样本插值新方法-信息扩散插值。该方法基于模糊映射思想,通过对稀少数据点的信息进行模糊扩散和插值映射,实现有限数据点信息向其邻近区域点的概率插值；针对正态扩散模型在描述非对称结构数据的局限性,发展了非对称信息扩散插值算法模型-椭圆模型和概率模型。插值试验对比分析验证了该插值方法的合理性和有效性。(3)三维海洋要素场重构。针对常规时空插值方法在分析处理ARGO资料和插值重构ARGO标准化网格产品中存在的问题和不足,提出了一种改进的时空插值算法,构建了三维温盐场反演的算法模型；基于改进的时空插值算法重构的三维要素场误差得到了有效的抑制。(4)针对实际海洋要素场资料中普遍存在的数据缺测问题和常规奇异谱方法中迭代参数K、M选取的主观性和计算效率差等问题,提出和发展了模型参数分段寻优的新方法,能够避免模型参数陷入局部误差收敛进而合理搜索到全局最优解,且迭代计算效率和插补结果准确率得到有效改进和提高。(5)利用卫星高度计资料和卫星遥感海面风场QuikSCAT资料,通过对不同纬度流场计算模型的分段处理,改进了地转流和Ekman流在赤道海区和赤道外海区的不连续问题,反演得到的海表地转流和Ekman流能够合理、准确地刻画全球海表流场的基本结构特征。(6)基于ARGO浮标三维温盐场观测信息,通过将P矢量方法引入等密度面温盐场动力约束,反演得到三维海洋流场。该方法弥补了实际海流观测困难、海流资料难以准确获取的现实问题和不足,反演推算的三维流场可客观表现不同深度的海洋流场的基本结构特征。更多还原

【Abstract】 Being the only real-time/delayed-time available three-dimensional global-scale ocean observation data, Argo data provides critical observation factual basis which contains a wealth of information resources for revealing the marine environment, temperature and salinity distribution and current structure, understanding and deepening of the marine environment features and climate change, and has important scientific significance. Aiming at the sparseness, fragments and discontinuousness of Argo data, the reanalysis research on fitting sparse data and reconstructing the three-dimensional temperature and salinity field is conducted. The surface current field is estimated and the three-dimensional current field is inversed based on the blending of Argo data and satellite remote sensing data. This study can provide a theoretical basis and technical means for exploring and optimizing the use of the Argo information resources and Argo data, and provide data and information support for deepening the research on the evolution of the marine environment and climate change and improving the numerical prediction capability of meteorological and oceanic numerical models.The main achievements and creative results are as follows:(1) Aiming at the small-scale detailed structures and characteristics of the sparse data like Argo data, the fractal interpolation method model is improved. By introducing genetic algorithm to the searching of fractal interpolation parameters (compression factor) to optimize the search, the objective optimization is achieved, and the rationality and effectiveness of the fractal interpolation method on the sparse data is improved.(2) A small sample interpolation method-Information diffusion interpolation is put forward. Based on the idea of fuzzy mapping, by fuzzy diffusing and mapping the sparse data points, this technique can achieve the probability interpolation from the limited data points to its neighboring regions points. In order to the limitations of the normal diffusion model on describing the asymmetric structure of data, the nonnormal diffusion model-the elliptical model and probability model is developed. This model is verified by comparing the results of interpolation experiments.(3) The reconstruction of three-dimensional ocean data field. Aiming at the deficiencies that exist in the Argo data and interpolation reconstruction Argo standardized grid products against the conventional space-time interpolation method, an improved space-time interpolation algorithm is developed, and the three-dimensional temperature and salinity model is conducted. This model can efficiently suppress the errors that exist in the conventional means.(4) Aiming at the common missing data in the actual marine field data and the conventional singular spectrum of the iterative parameter K, M selected subjectivity and poor computational efficiency, a new model parameter segmenting method is put forward. This method can avoid the model parameters into the local error convergence and thus a reasonable search for the global optimal solution can be got. The iterative computational efficiency and accuracy of the interpolation results can be improved and enhanced.(5) Based on the satellite altimeter data and satellite remote sensing of sea surface wind field QuikSCAT data, the inversed surface geostrophic flow and Ekman flow can be got by dealing with the flow model in different latitude segments using different ways to overcome the discontinuity near the equator.(6) Based on the Argo observational three-dimensional temperature salinity field, the three-dimensional ocean flow field is inversed using the P-vector method based on the isopycnal temperature and salinity constraints. This method makes up the actual ocean current observation difficulty and practical problems and deficiencies in available current data. The inversed three-dimensional flow field can objectively reproduce the performance of currents on the different depths.更多还原