海洋要素的变化存在明显的区域性和季节性的变化特性，本文选择海洋要素中最为突出的海表面温度（SST）要素作为主要分析参数，设计时空变异参数的计算指标，分析时空变异对验证误差影响的关系，通过研究及试验的数据精度验证，证明了时空变异是造成误差的直接原因之一。强烈的时空属性变异，在验证过程中会引入很大的验证误差，处于不同变异等级区划的数据，其验证结果相对误差可达13.08%，变异越剧烈的区域，精度验证效果越差，验证误差就越大，这些误差并非完全是遥感产品的误差，验证结果不具有代表性，不能真实的反映遥感产品的误差特征。对于SST等海洋遥感产品验证时，需要考虑时空变异对验证误差的影响和贡献，合理选择验证试验区域、代表性的评价数据集和科学的评价方法。

Ocean remote sensing is the only effective method for monitoring the marine ecosystem on a global scale in the long term. This method has played a crucial role in the research of carbon cycle, global changing, and ocean acidification and its biological impacts. The validation of ocean color products is quite important for the improvement of the algorithm and the reliability of further application. Given the complex and highly dynamic optical marine characteristics, especially for case Ⅱ water, the evaluation of accuracy and the expression of remote sensing products have become difficult, and considerable research has focused in this field. This study aims to create a scientific evaluation method of the representativeness of ground stations and the heterogeneity of remote sensing pixels of sea surface temperature (SST) verification.
The temporal and spatial variation parameters, including the statistical standard deviation of time series, the spatial variation index, and the frontal index, are designed. The analytic hierarchy process for determining the weight of different indices and the method for regionalizing the interval of a spatiotemporal variation. Verification datasets are classified using a distribution map of the spatiotemporal variation of space level division.
Result of the experimental data shows that spatiotemporal variation is a direct cause of errors. Considering the strong spatiotemporal variation, numerous verification errors are introduced in the verification process, which uses data from different variation grade divisions, and the relative error of the verification result can reach more than 13.08%. The spatial distribution of spatiotemporal variation is large in the Yellow and Bohai Sea regions and small in the East and South China Seas. The seasonal variation is strong in winter and spring and weak in summer and autumn.
In this study, we propose models and methods to reasonably select representative validation datasets and verify the scientific nature of the method from the experimental data. In the region where the spatiotemporal variability is intense, the verification errors are large. These errors are not completely remote sensing product errors. The results of verification are not representative and cannot really reflect the error characteristics of remote sensing products. For SST ocean remote sensing product verification, spatiotemporal variability and its contribution to the validation error must be considered. Moreover, a reasonable selection of evaluation data set, scientific method, and representative test area must be verified.