Creation of a gridded dataset for the Southern Ocean with a topographic constraint scheme
Creation of a gridded dataset for the Southern Ocean with a topographic constraint scheme
复制标题
使用地形约束方案创建南大洋网格数据集
DOI：
10.1175/jtech-d-16-0075.1
复制
发表时间：
2016
影响因子：
2.2
通讯作者：
K. I. Ohshima
中科院分区：
文献类型：
--
作者：
Shimada;K.;S. Aoki;K. I. Ohshima
这项研究调查了一种方法，用于创建一个气候数据集，提高了重现性和可靠性的南大洋。尽管观测采样稀疏，但南大洋的主要物理特征是在弱分层和强科里奥利效应下形成的强地形约束。为了提高网格化数据的保真度，将地形约束纳入插值方法，其加权函数包括来自底部深度差和水平距离的贡献。还分析了物理特性的空间变异性，以使用水文数据集估计水平距离和底部深度差异的现实去相关尺度。一个新的网格数据集，地形约束纳入（TCI），然后开发温度，盐度和溶解氧，使用新派生的加权函数和去相关尺度。在可用的网格化数据集之间比较插值值与相邻观测值之间的差（RMS差）的均方根（RMS）。TCI的RMS差异比以前的数据集小12%-21%和8%-20%的潜在温度和盐度，分别证明了地形约束和现实的去相关尺度的有效性。此外，去相关尺度的比较和插值误差的分析表明，在以前的网格数据集采用的去相关尺度是2倍或更多的实际尺度，高估会增加插值误差。本研究提出的插值方法也适用于其他层化程度较弱但采样不足的高纬度海洋。
This study investigated a method for creating a climatological dataset with improved reproducibility and reliability for the Southern Ocean. Despite sparse observational sampling, the Southern Ocean has a dominant physical characteristic of a strong topographic constraint formed under weak stratification and strong Coriolis effect. To increase the fidelity of gridded data, the topographic constraint is incorporated into the interpolation method, the weighting function of which includes a contribution from bottom depth differences and horizontal distances. Spatial variability of physical properties was also analyzed to estimate a realistic decorrelation scale for horizontal distance and bottom depth differences using hydrographic datasets. A new gridded dataset, the topographic constraint incorporated (TCI), was then developed for temperature, salinity, and dissolved oxygen, using the newly derived weighting function and decorrelation scales. The root-mean-square (RMS) of the difference between the interpolated values and the neighboring observed values (RMS difference) was compared among available gridded datasets. That the RMS differences are smaller for the TCI than for the previous datasets by 12%–21% and 8%–20% for potential temperature and salinity, respectively, demonstrates the effectiveness of incorporating the topographic constraint and realistic decorrelation scales. Furthermore, a comparison of decorrelation scales and an analysis of interpolation error suggests that the decorrelation scales adopted in previous gridded datasets are 2 times or more larger than realistic scales and that the overestimation would increase the interpolation error. The interpolation method proposed in this study can be applied to other high-latitude oceans, which are weakly stratified but undersampled.