NOAA最优插值全球海温分析资料（1981-2020）

最优插值（Optimum Interpolation，OI，）海表温度（Sea Surface Temperature， SST）分析是逐周的1°×1°格点数据，月平均资料则由逐周资料线性内插到该月的每一日，再经计算得到。该分析资料使用了卫星SST资料和海冰覆盖模拟的SST资料。在计算该分析资料之前，利用Reynolds（1988）和Reynolds and Marsico（1993）的计算方法，根据误差对卫星资料进行了调整。最优分析的详细描述见文献（Reynolds and Smith，1994）。误差校正提高了OI资料的大尺度准确性。
1981年以后的OI资料于2001年11月被重新计算（现在的OI.V2版本）。新版资料最明显的变化是：随着英国气象局模拟技术的提高，由海冰模拟的SST得到改进。这种变化降低了高纬SST的误差。另外，COADS资料的更新和延伸也有助于1997年船舶观测覆盖面的拓宽，降低了稀疏区（没有这些观测资料而导致的观测稀疏区）的卫星资料偏差（Reynolds, et al.，2002）。

Optimum Interpolation (OI) Sea Surface Temperature (SST) analysis is provided as weekly 1°×1° gridded data. Monthly mean datasets are calculated by first linearly interpolating the weekly data onto every day of the target month, then deriving the monthly averages. This analysis dataset incorporates satellite SST observations and SST data simulated from sea ice cover. Prior to generating this analysis product, satellite data were adjusted based on their respective errors using the methodologies proposed by Reynolds (1988) and Reynolds and Marsico (1993). A detailed description of the optimum interpolation analysis is available in Reynolds and Smith (1994). Error correction enhances the large-scale accuracy of the OI dataset.

The OI dataset covering post-1981 periods was recalculated in November 2001, leading to the current OI.V2 version. The most prominent improvement in the new dataset is the refined SST data simulated from sea ice, enabled by advancements in simulation technologies from the UK Met Office. This modification reduces SST errors in high-latitude regions. Additionally, updates and extensions to the COADS dataset have facilitated the expansion of ship-based observational coverage since 1997, mitigating biases in satellite data over data-sparse regions—regions characterized by insufficient observational data due to the lack of such measurements (Reynolds et al., 2002).