海洋卫星资料在气候模式中EAKF同化实验结果分析

发布时间：2018-04-04 01:36

本文选题：气候模式　切入点：集合调整Kalman滤波　出处：《国家海洋局第一海洋研究所》2014年硕士论文

【摘要】：气候变化一直备受国际关注。尽管大部分气候系统模式已经具备全球变化模拟能力，但仍然存在气候漂移、双赤道辐合带、系统误差等问题。充分利用各种观测数据，建立气候模式同化系统也是改善和提高气候模式模拟和预测的有效途径之一。针对气候模式开展同化研究具有较为实际的尝试价值，对提高气候模式的预报水平具有重要意义。本文基于国家海洋局第一海洋研究所发展的气候系统模式(FIO-ESM)，利用集合调整Kalman滤波方法，采用卫星高度计观测海面高度异常(SLA)和卫星观测海面温度(SST)两类数据，开展了海洋卫星资料同化实验，并分析了实验结果。本研究共设计了5组数值实验，包括1组不含数据同化的控制实验和4组同化实验。控制实验由运行了一组模式，分别从不同的初始场开始积分1年；同化实验与控制实验相似，但在积分过程中加入海洋卫星观测数据不断调整。前2组同化实验分别对SLA和SST数据进行同化，后2组实验则同时同化SLA和SST两种数据，但两种数据同化顺序不同。本文利用历史客观分析数据EN3和两套降水数据与数值实验结果进行对比分析，，分析结果均表明：数据同化能够显著改善FIO-ESM中海洋分量的模拟结果。4组同化实验的同化效果在时空上分布特征不同。单独同化SLA或SST数据的同化实验结果表明两种数据对同化结果的影响深度不同，SST改善效果在表层附近较大，而SLA则在次表层最大。联合SLA和SST同化的两组实验结果均比单独同化一类数据的效果更佳；数据同化的先后顺序对同化结果具有一定的影响，但并不显著，先同化SLA的结果略优。SLA和SST数据的观测信息反映的物理过程分别是整层水柱的变化和近表层海洋热量变化，数据物理属性的不同是造成数据同化效果差异的主要原因。同化实验对年平均降水分布有较显著的改善；在热带区域，海洋卫星资料同化有效抑制了气候模式中常见的双赤道辐合带问题，降低了赤道以南的降水量峰值，赤道北侧的降水峰值高于南侧，与观测结果一致；同化实验对低纬度降水强度改善最大，中纬度次之，高纬度改善相对较小。
[Abstract]:Climate change has been the focus of international attention.Although most climate system models already have the ability to simulate global change, there are still some problems such as climate drift, double equatorial convergence zone, system error and so on.It is also one of the effective ways to improve and improve the simulation and prediction of climate model by making full use of all kinds of observational data and establishing a climate model assimilation system.The study of assimilation based on climate model has practical value and is of great significance to improve the prediction level of climate model.Based on the climate system model FIO-ESM developed by the first Oceanographic Research Institute of the State Oceanic Administration, two types of data, SLAs and SSTs, are used to observe sea surface height anomalies by using a satellite altimeter and a satellite observation sea surface temperature (SST), using a set adjusted Kalman filtering method.The experiment of ocean satellite data assimilation is carried out and the experimental results are analyzed.In this study, five groups of numerical experiments were designed, including one control experiment without data assimilation and four assimilation experiments.The control experiment is composed of a set of models which are integrated from different initial fields for one year respectively. The assimilation experiment is similar to the control experiment but the ocean satellite observation data are continuously adjusted in the integration process.The first two groups of assimilation experiments assimilate SLA and SST data respectively, while the latter two groups assimilate both SLA and SST data simultaneously, but the order of assimilation between the two groups is different.In this paper, the historical objective analysis data EN3 and two sets of precipitation data are compared with the results of numerical experiments.The results show that data assimilation can significantly improve the simulation of ocean component in FIO-ESM. 4 groups of assimilation experiments have different spatial and temporal distribution characteristics.The results of the assimilation experiments of SLA or SST data alone showed that the effect of the two data on the assimilation results was different. The improvement effect of the two data on the assimilation results was greater near the surface layer, while SLA was the largest in the subsurface layer.The results of two groups of data assimilation combined with SLA and SST were better than that of single assimilation, the order of data assimilation had a certain influence on the assimilation results, but it was not significant.The results of assimilation of SLA are superior. SLA and SST data reflect the physical process of whole water column and ocean heat change respectively. The difference of physical attributes of data is the main reason for the difference of data assimilation effect.Assimilation experiments have a significant improvement on the annual average precipitation distribution, and in the tropical region, the assimilation of ocean satellite data effectively restrains the problems of the double equatorial convergence zone common in the climate model, and reduces the precipitation peak south of the equator.The peak value of precipitation in the northern side of the equator is higher than that in the south, which is consistent with the observed results, and the assimilation experiment shows that the intensity of precipitation in the low latitude is the greatest, followed by the middle latitude, and the high latitude is relatively small.
【学位授予单位】：国家海洋局第一海洋研究所
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P732.4

【参考文献】