Coastal sea level trends from TOPEX-Poseidon satellite altimetry and tide gauge data in the Mediterranean Sea during the 1990s
Coastal sea level trends from TOPEX-Poseidon satellite altimetry and tide gauge data in the Mediterranean Sea during the 1990s
复制标题
DOI：
10.1111/j.1365-246x.2007.03424.x
复制
发表时间：
2007-07
影响因子：
2.8
通讯作者：
S. Mangiarotti
中科院分区：
文献类型：
--
作者：
S. Mangiarotti
TOPEX-海神卫星测高和验潮数据的联合分析被用来估计海平面的趋势沿着地中海沿岸从1993年1月至2002年2月。广义最小二乘（GLS）方法允许我们考虑到时空分布的数据和它们各自的误差预算。在不同的分析尺度下，使用试错法估计GLS方法中所涉及的协方差函数的接近最优的参数。然后在沿沿着的153个地点重建组合数据时间序列。沿海地区的全流域平均线性趋势显示海平面上升4.54 ± 0.30 mm yr −1，而仅在开阔洋为4.28 ± 0.30 mm yr −1，即0.26 ± 0.16 mm yr −1的差异（误差较小，因为系统误差不参与误差平衡），这是相当显著的。在局部尺度上，沿海线性趋势范围为-8.5至+13.5毫米/年，比其他公海研究发现的范围（[-24.0至-10.0; +16.0至+29.0]毫米/年）窄得多。然而，一个详细的误差预算表明，最大的估计误差来自空间外推的海平面信号时所需的验潮仪和高度计数据合并。第二个误差源是仪器噪声和由高度计测量的沿岸的污染和由污染验潮仪信号的地壳运动产生的随机噪声。在局部尺度上，估计85%的线性趋势误差不超过2 mm yr-1。
A joint analysis of TOPEX-Poseidon satellite altimetry and tide gauge data is used to estimate sea level trends along the Mediterranean Sea coast from 1993 January to 2002 February. A generalized least squares (GLS) method allows us to take into account the space-time distribution of the data and their respective error budgets. Nearly optimal parameters of the covariance functions involved in the GLS method are estimated at different scales of analysis, using a trial and error method. Combined data time series are then reconstructed at 153 locations along the coast. The basin-wide averaged linear trend for the coastal area shows a sea level rise of 4.54 ± 0.30 mm yr −1 compared to 4.28 ± 0.30 mm yr −1 for the open ocean only, i.e. a 0.26 ± 0.16 mm yr −1 difference (the error is smaller because the systematic error does not take part in the error budget) which is quite significant. At local scales, the coastal linear trends ranges from −8.5 to +13.5 mm yr −1 which is much narrower than ranges found by other open sea studies ([−24.0 to −10.0; +16.0 to +29.0] mm yr −1 ). However, a detailed error budget shows that the largest estimation errors come from the spatial extrapolation of the sea level signal required when tide gauge and altimeter data are merged. The second error source is the instrumental noise and the random noise generated by the littoral contamination on altimeter measurements and from the crustal motions that are contaminating the tide gauge signal. At local scale, it is estimated that 85 percent of the error on linear trends does not exceed 2 mm yr −1 .