Department of Earth and Planetary Sciences, Harvard University, Cambridge, Massachusetts 02138; email:
Ann Rev Mar Sci. 2016;8:1-33. doi: 10.1146/annurev-marine-122414-034040. Epub 2015 Aug 31.
Understanding the ocean requires determining and explaining global integrals and equivalent average values of temperature (heat), salinity (freshwater and salt content), sea level, energy, and other properties. Attempts to determine means, integrals, and climatologies have been hindered by thinly and poorly distributed historical observations in a system in which both signals and background noise are spatially very inhomogeneous, leading to potentially large temporal bias errors that must be corrected at the 1% level or better. With the exception of the upper ocean in the current altimetric-Argo era, no clear documentation exists on the best methods for estimating means and their changes for quantities such as heat and freshwater at the levels required for anthropogenic signals. Underestimates of trends are as likely as overestimates; for example, recent inferences that multidecadal oceanic heat uptake has been greatly underestimated are plausible. For new or augmented observing systems, calculating the accuracies and precisions of global, multidecadal sampling densities for the full water column is necessary to avoid the irrecoverable loss of scientifically essential information.
理解海洋需要确定和解释全球积分以及温度（热量）、盐度（淡水和盐含量）、海平面、能量和其他性质的等效平均值。在一个信号和背景噪声在空间上非常不均匀的系统中，历史观测数据稀少且分布不佳，这阻碍了对平均值、积分和气候学的确定，导致潜在的大时间偏差误差，必须在 1%或更好的水平进行校正。除了当前测高-Argo 时代的上层海洋外，对于热量和淡水等数量的平均值及其变化的最佳估计方法，以及对于人为信号所需的水平，没有明确的记录。趋势的低估和高估一样可能；例如，最近的推断认为，多十年的海洋热量吸收被大大低估了，这是合理的。对于新的或增强的观测系统，计算全球、多十年的整个水柱采样密度的准确度和精密度是必要的，以避免不可挽回地失去科学上必不可少的信息。
