全球平均海平面上升的瞬时速率

在确定海平面上升速率时，传统方法是利用最小二乘拟合获取特定时间段内的平均速率.事实上，由于海平面是一种非稳态变化，其速率随着时间变化.本文使用集成经验模态分解获取海平面变化在2002-2014年间的非线性趋势，然后通过三次样条函数平滑拟合非线性趋势得到连续的一阶导数，即为海平面变化的瞬时速率.结果表明，全球平均海平面的瞬时速率先降后升：从2002的2.7 mm·a^-1缓慢下降至2010年的2.5 mm·a^-1，然后上升至2014年的3.8 mm·a^-1.通过分析海平面上升各个贡献成分的瞬时速率，发现该上升主要由海水质量增加引起.在2002-2014年间，格陵兰岛冰川消融对海平面上升瞬时速率的贡献从0.51 mm·a^-1上升至0.85mm·a^-1，南极冰川消融的贡献则从0.12 mm·a^-1上升至0.34 mm·a^-1.陆地水储量对海平面上升起抑制作用，但该抑制作用呈下降趋势，其瞬时速率从-0.24 mm·a^-1增加到0.03 mm·a^-1.比容海平面的瞬时速率表现为下降趋势，从1.6 mm·a^-1减小至1.0 mm·a^-1.这表明在全球尺度上，海水质量对海平面上升的贡献正在增加，截止到2014年，海水质量的贡献已经接近70%.

The instantaneous rate of global mean sea level rise

Sea level changes play an important role in current climate change and cause serious economic consequences. It is crucial to determine the rate at which the sea level rises. Traditionally, least squares method is used to calculate the linear trend of sea level changes for a common period. However, a recent study has shown that least squares method is sensitive to the interannual variability. For example, the global mean sea level (GMSL) rate for 2003-2011 increases from 2.4 mm·a^-1 to 3.3 mm·a^-1 after the interannual variability is corrected. It is well known that sea level changes are non-stationary, which means that the rate varies with time. To exclude the effect of interannual variability and reveal the temporal evolution of the rate of GMSL, we propose a method for computing the instantaneous rate of sea level rise. Firstly, we use Ensemble Empirical Mode Decomposition (EEMD) to extract the non-linear trend from time series of sea level. The combination of the last two intrinsic functions is treated as the non-linear trend. Then we use cubic Spline smoothing to fit the non-linear trend. The first-order derivative derived by cubic Spline smoothing is the instantaneous rate.#br#This method is applied to the GMSL time series (2002-2014) from an inversion result which can close the sea level budget on global scale. This inversion result divided the GMSL into six components including steric sea level, ocean mass contributions from Greenland, Antarctica, glaciers and land hydrology, and a non-stochastic residual signal. Our results show that the instantaneous rate of GMSL slowly decreases from 2.7 to 2.5 mm·a^-1 during 2002-2010. After 2010, the rate of GMSL rapidly increases to 3.8 mm·a^-1 in November 2014. The overall linear trend of steric GMSL is 1.38 mm·a^-1 which is greater than previous studies, however, its instantaneous rate suggests a descending trend. The steric rate decreases from 1.6 mm·a^-1 at 2002 to 1.0 mm·a^-1 at 2014. The rate of Greenland's contribution to GMSL increases from 0.51 to 0.87 mm·a^-1 during 2002-2012, and becomes relative stable during 2012-2014. Antarctica's contribution to GMSL experiences an overall increase in the instantaneous rate which increases from 0.13 to 0.34 mm·a^-1 for the period 2002-2014. The contribution from glacier indicates a slightly decreased trend for 2002-2012, declining from 0.43 to 0.35 mm·a^-1. In contrast to the other mass contributions, hydrology has slowed down the GMSL rise. We note, however, the mitigation is becoming smaller and smaller during 2002-2012. The instantaneous rate increases from -0.24 to -0.02 mm·a^-1. Hydrology starts to make positive contributions to sea level rise after 2013. To close the sea level budget, a non-stochastic residual signal is necessarily derived. This signal only shows an overall trend of 0.22 mm·a^-1 during 2002-2014. However, we find that the instantaneous rate increases rapidly from 0.16 to 0.82 mm·a^-1 for the period 2010-2014. If we attribute this residual signal to the mass-induced sea level rise, then the mass contributions to GMSL is about 70% in 2014.