基于LightGBM的全球海洋次表层温盐遥感预测

随着卫星遥感技术的发展，越来越多的卫星观测数据被应用于预测海洋内部温盐结构信息，而如何有效提高海洋内部温盐信息的预测精度仍是一个挑战。本文应用LightGBM算法结合随机森林算法构建全球海洋次表层（0—1000 m）温度异常（STA）与盐度异常（SSA）的预测模型，模型使用海表卫星观测数据（海表高度异常（SSHA）、海表温度异常（SSTA）、海表盐度异常（SSSA）和海表风场异常水平和垂直分量（USSWA、VSSWA），结合经纬度信息（LON、LAT）作为预测变量，使用Argo次表层温盐数据作为模型训练与测试标记。本文使用五参数模型（SSTA、SSHA、SSSA、USSWA、VSSWA）、带纬度六参数模型（LAT、SSTA、SSHA、SSSA、USSWA、VSSWA）、带经度六参数模型（LON、SSTA、SSHA、SSSA、USSWA、VSSWA）与带经纬度七参数模型（LON、LAT、SSTA、SSHA、SSSA、USSWA、VSSWA）来着重分析LON与LAT在STA、SSA遥感预测中发挥的作用。结果表明LON与LAT在STA、SSA各自预测中发挥不同的作用。在单时相和时序预测STA中LON与LAT对模型的贡献随着深度的增加逐渐增大，而在单时相和时序预测SSA中LON与LAT对不同深度预测模型始终保持较大的贡献。在单时相预测STA与SSA中LON较LAT对模型贡献更大，而在时序预测STA与SSA中LAT较LON对模型贡献更大。经纬度信息是全球海洋次表层温盐机器学习预测的重要参数，可以提高模型的预测精度。同时，LightGBM较随机森林在预测海洋次表层温盐异常时精度更高鲁棒性更强。     

Argo时代东南印度洋上层盐度变化对海平面年代际演变的影响

In the past nearly two decades, the Argo Program has created an unprecedented global observing array with continuous in situ salinity observations, providing opportunities to extend our knowledge on the variability and effects of ocean salinity. In this study, we utilize the Argo data during 2004–2017, together with the satellite observations and a newly released version of ECCO ocean reanalysis, to explore the decadal salinity variability in the Southeast Indian Ocean(SEIO) and its impacts on the regional sea level changes. Both the observations and ECCO reanalysis show that during the Argo era, sea level in the SEIO and the tropical western Pacific experienced a rapid rise in 2005–2013 and a subsequent decline in 2013–2017. Such a decadal phase reversal in sea level could be explained, to a large extent, by the steric sea level variability in the upper 300 m. Argo data further show that, in the SEIO, both the temperature and salinity changes have significant positive contributions to the decadal sea level variations. This is different from much of the Indo-Pacific region, where the halosteric component often has minor or negative contributions to the regional sea level pattern on decadal timescale. The salinity budget analyses based on the ECCO reanalysis indicate that the decadal salinity change in the upper 300 m of SEIO is mainly caused by the horizontal ocean advection. More detailed decomposition reveals that in the SEIO, there exists a strong meridional salinity front between the tropical low-salinity and subtropical high salinity waters. The meridional component of decadal circulation changes will induce strong cross-front salinity exchange and thus the significant regional salinity variations.     

全球海洋次表层温度异常遥感反演的季节时空变化特征

卫星遥感反演海洋内部多时相、大尺度热力结构信息对于了解海洋内部复杂、多维的动力过程有重要意义。本文采用随机森林回归模型，利用卫星遥感观测的海表参量（海表高度异常（SSHA）、海表温度异常（SSTA）、海表盐度异常（SSSA）和海表风场异常（SSWA）），反演不同季节、不同深度层位（1000 m深度以上）的海洋次表层温度异常（STA），并用Argo实测数据作精度验证，采用均方根误差（RMSE）、归一化均方根误差（NRMSE）以及决定系数（R²）评价模型在全球及洋盆尺度上的反演精度。结果显示，全球海洋16个深度层位的平均R²在春、夏、秋、冬四季分别为0.53、0.60、0.54、0.66，NRMSE分别为0.051、0.031、0.043、0.044。随着季节的变化，模型反演性能有所波动。模型在印度洋的反演效果最佳，不同季节、不同深度层位上的平均R²和RMSE分别为0.71和0.18 ℃，而大西洋的反演精度最低，平均R²和RMSE分别为0.46和0.25 ℃。研究表明随机森林模型适用于全球不同季节的STA遥感反演，且在不同洋盆上均有较好的反演效果；不同季节上，上层STA有明显变化信号，空间异质性显著，但300 m以深，STA信号较弱且基本不随季节变化。本研究可为长时序、大尺度海洋内部参量信息遥感反演与重建提供依据，有助于进一步发展深海遥感方法。     

基于大涡模拟和局部滤波同化方法的海洋环流模式

结合最小二乘法极值原理,提出了一种基于局部谱展开的滤波同化方法,把测量数据和数值计算过程中出现的高频短波滤掉,并将高度计数据同化到了求解过程中.结果既增加了数值稳定性,又提高了数值模拟的准确性.针对在海洋环流问题中水平的流动性质和垂直的不同的特点,我们还将大涡模拟的思想和直接涡黏的思想分别应用于水平方向和垂直方向,给出的方法是一种适用于海洋环流和浅水环流问题的大涡模拟湍流模式.对热带和北太平洋一年四季非定常季风作用下环流的数值模拟表明,提出的方法非常有效,数值结果与实际相当吻合.     

多源卫星观测的全球海洋次表层温度异常信息提取

基于表层卫星遥感观测的中深层海洋遥感对于了解海洋内部异常及其动力过程有重要意义。如何从现有的海洋表层遥感观测资料提取海洋内部关键动力环境信息场是具有挑战性的海洋遥感技术前沿。本文采用支持向量回归(SVR)方法,通过卫星遥感观测获取的多源海表参量(海表高度异常(SSHA)、海表温度异常(SSTA)、海表盐度异常(SSSA)和海表风场异常(SSWA)),选择最优参量输入组合,感知海洋次表层温度异常(STA),并用实测Argo数据作精度验证。结果表明SVR模型可准确估算全球尺度的STA(1000 m深度以浅);当SVR输入变量为2个(SSHA、SSTA)、3个(SSHA、SSTA、SSSA)、4个(SSHA、SSTA、SSSA、SSWA)时对应的平均均方差(MSE)分别为0.0090、0.0086、0.0087,平均决定系数(R2)分别为0.443、0.457、0.485。因此,除了SSHA和SSTA外,SSSA与SSWA的输入对SVR模型的估算有积极影响,有助于提高STA的估算精度。在全球增暖与减缓背景下,该研究可为从表层卫星遥感观测提取海洋内部热力异常信息研究提供重要技术支持,有利于拓展卫星对海观测范围。     

LABORATORY MEASUREMENTS OF THE EFFECTS OF VISCOSITY ON SHORT WATER WAVE SPECTRA AND IMPLICATION FOR RADAR REMOTE SENSING OF THE OCEAN SURFACE

Using an X-band radar and a mechanical short wave generator, we measured the dependenceof radar return power on the viscosity of water which was adjusted by changing the watertemperature. From the measurements we drew two inferences: (1) the spectral density of short waterwaves, F, depends strongly on the viscosity of waterF(v)=F_0~mwhere is normalized viscosity, and m=-1.72; and (2) the normalized radar cross section, σ_0(dB), dependsstrongly on the water temperature△σ_0 (dB)=O.217△TAnalyses indicate that these dependencies can be observed in the field only at low wind conditionsat which the Bragg scattering is a dominant mechanism for producing the radar return signals. Theresults of this study can be used to interpret the sharp variability in radar cross section across a watertemperature front and SAR images of oeeanic phenomena.     

A numerical study on the generation of a distinct type of nonlinear internal wave packet in the South China Sea

A distinct type of nonlinear internal-wave packet, with the largest internal solitary wave in the middle of the packet, was regularly observed in the South China Sea during the Asian Seas International Acoustics Experiment in 2001. Data analysis shows that the occurrence of the distinct internal wave packet is closely related with the occurrence of lower-high internal tides; the internal tides are mixed in the experimental area and, thus, there is diurnal inequality between the heights of two neighboring internal tides. Modeling of internal tides and internal solitary waves in a shoaling situation suggests that this type of wave packet can be generated in the South China Sea by the large shoaling of internal solitary waves and internal tides. Both the internal solitary waves and the internal tides come from the direction of Luzon Strait. The initial large internal solitary waves contribute to the occurrence of the largest internal solitary wave in the middle of the packet and the waves behind the largest internal solitary wave, while the shoaling internal tides bring about the nonlinear internal waves in front of the largest internal solitary wave via interaction with the local shelf topography.     

A numerical study on the generation of a distinct type of nonlinear internal wave packet in the South China Sea

A distinct type of nonlinear internal-wave packet, with the largest internal solitary wave in the middle of the packet, was regularly observed in the South China Sea during the Asian Seas International Acoustics Experiment in 2001. Data analysis shows that the occurrence of the distinct internal wave packet is closely related with the occurrence of lower-high internal tides; the internal tides are mixed in the experimental area and, thus, there is diurnal inequality between the heights of two neighboring internal tides. Modeling of internal tides and internal solitary waves in a shoaling situation suggests that this type of wave packet can be generated in the South China Sea by the large shoaling of internal solitary waves and internal tides. Both the internal solitary waves and the internal tides come from the direction of Luzon Strait. The initial large internal solitary waves contribute to the occurrence of the largest internal solitary wave in the middle of the packet and the waves behind the largest internal solitary wave, while the shoaling internal tides bring about the nonlinear internal waves in front of the largest internal solitary wave via interaction with the local shelf topography.