多源卫星臭氧观测的高效全球四维变分同化技术研究

In recent years,the number of ozone instruments onboard satellite increases continuously. They can not only provide abundant information in many areas，such as the stratosphere and ocean， where the conventional observation is usually scarce, but also have huge potential for improving the accuracy of numerical weather predtiction (NWP). Although many operational NWP centers have been successful in satellite ozone data assimilation and achieved some positive results，but it is still in face of a great deal of scientific challenges in the ozone data assimilation community. This study will focus on the key tecnologies for global four-dimensional variational data assimilation (4DVAR) of multi-source satellite ozone observations, mainly including the innovative method of atmospheric constituent data assimilation, exact description of background error characteristic for different scale, the new observation quality control method based on L1/L2 mixed norms, self-adaptive variational bias correction method, chemistry transport model for ozone,the tangent and adjoint programs of stratospheric physical and chemistry parameterization schemes, the new radiative transfer models for ozone instruments, and so on. The target is to improve the analysis quality and medium-range forecasting ability of high-resolution global NWP system by optimal combination and high-efficient assimilation of multi-source satellite ozone data (includes two FY-3 ozone instruments), and especially to discover the potential to improve the accuracy of typhoon forecasts and wind predictions in the middle and upper atmosphere. Therefore, this study is of importance for further developments of satellite data assmilation method, and has great practical values for both the completeness of data assmilation theory and construction of operational 4DVAR systems.

近年来在轨运行的卫星臭氧遥感仪不断增加，丰富的臭氧观测不但能有效弥补平流层等区域观测信息的不足，而且对提高数值天气预报质量具有巨大潜力。虽然许多业务预报中心已成功实现臭氧资料同化，并取得一定正效果；但是仍存在一系列急需解决的科学难题。本项目将开展多源卫星臭氧资料全球四维变分同化的关键技术研究，主要包括：大气成分资料同化新方法、臭氧不同尺度背景误差特征刻画、基于混合范数的观测质量控制、自适应偏差订正、臭氧化学输送模式、平流层物理和化学参数化方案以及卫星观测算子的切线性/伴随模式等。目标是实现不同类型卫星(包括风云三号)臭氧观测的最优融合和高效同化，提高高分辨率全球数值天气预报系统的分析场质量和中期天气预报能力，重点挖掘臭氧同化在改进中高层风场和对流层台风等高影响天气预报上的潜力。本项目的研究是对卫星资料同化方法的进一步探索，对资料同化理论完善和业务同化系统研制都有重要实用价值。

臭氧是影响全球数值天气预报(NWP)效果的重要因素之一，在模式顶扩展到平流层及其以上区域后，如何引入和刻画臭氧成为一个重要科学问题。随着在轨运行的卫星臭氧遥感仪器数量不断增加，丰富的臭氧观测对提高NWP质量具有巨大潜力。本课题主要研究了多源卫星臭氧资料全球四维变分同化的关键技术。首先，完成了大气臭氧资料同化新方法、适合痕量气体的控制变量转换算子和新目标泛函最优化求解的预条件、刻画臭氧不同尺度的小波背景误差协方差模型、自适应偏差订正、基于混合范数的观测质量控制、臭氧化学输送模式、平流层物理和化学参数化方案以及卫星臭氧观测算子的切线性/伴随模式等研究内容。其次，设计和实现了臭氧观测资料同化软件模块，集成到已有的全球大气四维变分资料同化系统后最终形成了全球臭氧观测资料同化系统。最后开展了卫星臭氧资料同化实验，通过对比分析验证了高分辨率臭氧观测资料同化系统的有效性。实验结果表明：高分辨率全球臭氧资料同化系统具有卫星臭氧资料同化能力，能提供三维臭氧浓度分析产品，产品水平分辨率为水平网格距25千米、垂直分层91层；由分析场转换得到的初始场输入全球大气数值预报模式后能生成臭氧、温度、压强、风场等大气变量的10天预报场；在同化卫星臭氧资料后总气柱臭氧含量的预报场分布具有明显变化：臭氧预报场的最大值、最小值有较大变化，总气柱臭氧预报场的高值区和低值区形状明显不同。统计检验结果表明：在同化卫星臭氧观测资料后南北半球500hPa位势高度10天预报的距平相关数值都有明显变化，北半球第一至第十天预报技巧都有一致改进，而且随着预报时间延长提高越大，南半球第一至第八天预报都产生了一定的正效应。总之，卫星臭氧观测的引入增加了新的观测信息、填补了观测信息“盲区”，从而提高了分析场和背景场质量，增加了与观测实况的一致性，提高了其它类型观测的同化利用率。本课题的研究对卫星资料同化方法进一步探索和业务同化系统研制都具有重要科学意义和实用价值。