基于EEMD及BP神经网络的区域海平面多尺度预测研究

基于时间序列的统计预测模型是现阶段海平面高度预测的主要手段之一,然而海平面变化机理复杂,传统方法对于非平稳非线性的时间序列预测存在较大局限性,预测精度有待进一步提高.本文基于闸坡站长时间(1959～2011年)月均验潮序列,结合集合经验模态分析(Ensemble Empirical Mode Decomposition,简称EEMD)与BP(Back Propagation)神经网络方法,提出一种改进的区域海平面变化趋势预测方法——EEMD-BP建模.本研究首先利用EEMD方法对原始序列进行分解,根据验潮序列中隐含的各个信号的不同频谱特征生成多个本征模函数(Intrinsic Mode Function,简称IMF),达到将时间序列平稳化,提高信噪比的效果.然后由各IMF作为BP神经网络的输入因子,分别预测各IMF的未来变化趋势,最后将输出结果重建得到原始序列的预测值.结果显示,EEMD能有效提取序列中隐含的多时间尺度信号,神经网络能较好地预测海平面未来变化趋势,相对于直接使用BP神经网络进行海平面变化时间序列预测(R=0.76,RMSE=36.74mm,ME=-3.46),EEMD-BP建模预测精度有显著提高(R=0.89,RMSE=28.16mm,ME=2.31).说明EEMD-BP建模首先对非平稳非线性时间序列进行平稳化、降噪等处理,再分别对分解后序列进行预测,有利于提高预测精度.该方法为相关区域海平面变化趋势预测研究提供现实参考意义.     

华北晚古生代陆表海盆地东南缘高频海侵事件对聚煤作用的影响

晚古生代华北陆表海盆地的海平面变化具有复合性海平面变化的特点 ,长周期的海平面变化周期中叠加了中、短期的海平面波动变化 ;长周期海退过程中 ,具多次短周期海侵事件发生 ,长周期海侵过程中也有短周期的海退发生。而短周期的海平面波动变化属于高频海平面变化 ,深刻地影响着陆表海盆地的沉积充填和聚煤作用。总起来看 ,鲁西及邻区的晚古生代陆表海盆地充填沉积是由一种较薄的反映水体向上变浅的复合性旋回序列组成 ,其分界面是由地史上瞬时的相对海平面上升形成的间断事件所致 ,或相对海平面下降期间的暴露事件形成的间断面所产生的。陆…     

中国全新世海平面变化周期与世界未来海平面变化规律

从发现海滩岩断代序列模式后,通过多学科综合研究又发现气候变化周期、闽粤海岸升降周期和海平面变化周期等皆为500a 左右。在13000aB.P.前中国气候变暖,海平面为以上升为主的升降期;近6000多年海平面以周期性上下波动为特征;而近3100年寒冷气候以每1000年7个纬度的速度南移,海平面向上波动幅度变小。按气候变化周期,现阶段为变寒期,将抵消一部分“温室效应”,世界未来50—100年的海面可能与今持平或有所下降。按照冰期和间冰期的时间规律,未来6000—7000年的海平面将下降几十米。     

华南陆缘晚三叠-早、中侏罗世海平面相对升降与古气候演化的地球化学记录

晚三叠-中侏罗世华南陆缘发生多期海侵-海退旋回.元素地球化学特征指示的古盐度高低和古水深深浅记录了海平面的相对升降.华南陆缘存在3个级别的海平面变化.其中, Ⅲ级海平面波动频繁, 由晚三叠世3次波动、早侏罗世4次波动和早、中侏罗世2次波动构成了此间大规模海侵-海退旋回.华南陆缘Ⅰ级海平面相对升降曲线与全球海平面长周期曲线, 在诺利克阶早期-辛涅缪尔阶晚期变化同步, 在辛涅缪尔阶晚期-阿连阶变化相悖.Ⅲ级海平面相对升降曲线与全球海平面短周期曲线, 在瑞替阶-辛涅缪尔阶晚期同发生4次波动, 时间上近乎同步.Sr/Cu值指示的元素迁移与富集反映了气候的温湿与干热.粤中-粤东地区晚三叠-早、中侏罗世经历了温湿-干热-温湿的气候变化.      

海平面上升可能使珠江三角洲灾害增加

由中国科学院、广东省科学院等 13个单位、10 0多位科技人员历时 8a合作完成的一项研究表明 ,珠江三角洲地区沿海的海平面到 2 0 30年可能会上升 30cm。研究内容包括广东海平面变化的历史背景、近期趋势和预测以及海平面上涨后的影响和对策。该项研究的数据来源于广东沿海 11个验潮站的验潮记录。根据数据 ,专家计算出相对海平面上升速率为平均每年 2至 2 5mm。据推断 ,海平面上升30cm可能会加重珠江三角洲的洪水灾害 ,使得原来按 10 0a一遇洪水设计的堤围 ,连 2 0a一遇的洪水都抵御不了。同时 ,地势低平的珠江三角洲届时将有 4 2 %的面积…     

中国东海海平面变化多尺度周期分析与预测

海平面变化规律尤其是对海平面变化周期和上升趋势的研究,已成为国内外科学界研究的热点问题.使用1992-2009年海平面卫星测高仪数据资料,运用小波变换方法对中国东海海平面变化的周平均数据信号进行多尺度周期分析,并通过Winters指数平滑法对未来海平面变化进行预测,结果显示:①1992-2009年东海海平面呈现波动上升...     

海潮及其影响下海岸带地下水位时间序列的周期性和滞后性

运用谱分析原理，对广西北海市海潮及受海潮影响的滨海含水层地下水位时间序列进行分析，求出海潮和观测孔水位变化的周期及地下水位滞后于海潮的时间，分析水位滞后时间与离海岸的垂直距离之间的关系。结果表明，北海市海潮和海岸带地下水位有约344．82h（14．37d）的长周期变化和约24．7h、12．5h的短周期波动，距海岸2175、2350、2375m的3个观测孔地下水位对海潮的滞后时间分别为5．5、6．25和7h，滞后时间随离海岸的距离增大大体上呈线性增加。     

中国沿岸海平面上升与海岸灾害

政府间气候变化专门委员会1995年气候变化评价报告指出,全球海平面在过去的100年里上升了18cm,预测全球海平面将加速上升,在2050年时上升20cm,2100年时上升49cm.根据近40多年的验潮资料分析,中国沿岸海平面上升速率为1.4～2.0mm/a,与全球上升速率一致。由于沿海许多地区的严重地面沉降,地壳垂直升降的不同,以及其它因素的影响,我国沿岸海平面的相对上升各不相同,严重地区的上升速率要大得多。海平面的加速上升必将使我国沿海大部地区的风暴潮等自然灾害更加严重。     

基于小波方法的西安市年降水量多时间尺度特征分析

小波分析具有时频局部化特征,可清晰地给出各种时间尺度周期的强弱分布以及降水量偏多、偏少的变化趋势。应用小波分析的方法,对西安市1951～2006年共56年的年降水量序列进行多时间尺度特征分析。结果表明,西安市年降水量序列在整个时间域内主要受2～6年、6～12年、16～34年时间尺度波动变化影响,不同时间尺度下,降水量偏多、偏少交替变化不相同。年降水量序列主要存在4年、8年、26年左右的主周期,其中,26年为第一主周期。另外,西安市多年降水量呈弱下降趋势,2006年以后的降水应呈正位相,降水量偏多,大约到2018年左右,年降水量总体上都应当偏多。     

海平面上升—地质灾害研究不可忽视的领域

一、关于海平面上升趋势 据国际有关组织对全球数百个验潮站资料的分析计算,近百年来海平面平均以每年1～2mm的速度上升(我国约为1.5mm左右)。据专家估算今后50年内(至2050年),珠江三角洲地区海平面上升幅度50～60cm,长江三角洲地区海平面上升幅度为60～80cm,天津地区海平面上升幅度为70～90cm。从全国范围来看,海平面上升幅度大致在0.5～1m之间。     

近2 Ma BP以来地球轨道参数周期上全球海平面变化机制

海平面的变化往往对区域生态环境、社会经济造成严重影响.通过对全球相对海平面变化(relative sea level,简称RSL)记录的再分析结果,合成了近2 Ma BP以来的全球RSL变化记录,分析了合成RSL记录对原始RSL记录数理特征的继承性,并基于频谱、滤波等数理分析验证了合成RSL记录的合理性.在此基础上,讨论了合成RSL与大气CO₂浓度、中高纬度海域表层海水温度(sea surface temperature,简称SST)、全球大洋底栖氧同位素(δ18O_B)等参数指标间的相关性,结果显示:(1) 合成RSL不仅与原始RSL记录的变化趋势基本一致,继承了原始RSL记录对全球气候变化的响应特征,而且显示出合成RSL记录对地球轨道参数周期变化响应明显;(2) 近2 Ma BP以来,在冰期-间冰期旋回中,合成RSL与δ18O_B变化呈良好的负相关,相关系数r平均值可以达到约0.81,高于合成RSL与大气CO₂浓度及中高纬度海域SST变化的相关系数;(3) 在地球轨道参数周期上,合成RSL与极地冰盖体积(δ18O_B)的变化几乎同时,在偏心率周期上,合成RSL落后于SST和大气CO₂浓度变化;在斜率周期上,合成RSL落后于SST变化而领先于大气CO₂浓度变化.推测这些变化的诱导因素可能是在太阳辐射量改变的前提下,大气CO₂浓度及大洋SST变化对极地冰盖体积产生了差异影响,进而引起海平面发生变化.      

Late Holocene relative sea level rise and the Neoglacial history of the Greenland ice sheet

In West Greenland, early and mid Holocene relative sea level (RSL) fall was replaced by late Holocene RSL rise during the Neoglacial, after 4–3 cal. ka BP (thousand calibrated years before present). Here we present the results of an isolation basin RSL study completed near to the coastal town of Sisimiut, in central West Greenland. RSL fell from 14 m above sea level at 5.7 cal. ka BP to reach a lowstand of ?4.0 m at 2.3–1.2 cal. ka BP, before rising by an equivalent amount to present. Differences in the timing and magnitude of the RSL lowstand between this and other sites in West and South Greenland record the varied interplay of local and non‐Greenland RSL processes, notably the reloading of the Earth's crust caused by a Neoglacial expansion of the Greenland Ice Sheet (GIS) and the subsidence associated with the collapse of the Laurentide Ice Sheet forebulge. This means that the timing of the sea level lowstand cannot be used to infer directly when the GIS advanced during the Neoglacial. The rise in Late Holocene RSL is contrary to recently reported bedrock uplift in the Sisimiut area, based on repeat GPS surveys. This indicates that a belt of peripheral subsidence around the current ice sheet margin was more extensive in the late Holocene, and that there has been a switch from subsidence to uplift at some point in the last thousand years or so. Copyright © 2008 John Wiley & Sons, Ltd.     

Late Holocene Sea Level Reconstructions Based on Observations of Roman Fish Tanks,Tyrrhenian Coast of Italy

We present estimates for late Holocene relative sea level change along the Tyrrhenian coast of Italy based on morphological characteristics of eight submerged Roman fish tanks (piscinae) constructed between the 1st century B.C. and the 2nd century A.D. Underwater geomorphological features and archaeological remains related to past sea level have been measured and corrected using recorded tidal values. We conclude that local sea level during the Roman period did not exceed 58 ± 5 cm below the present sea level. These results broadly agree with previous observations in the region but contrast with recent analysis that suggests a significantly larger sea level rise during the last 2000 years. Using a glacial isostatic adjustment model, we explain how regional sea level change departs from the eustatic component. Our calculation of relative sea level during the Roman period provides a reference for isolating the long‐wavelength contribution to sea level change from secular sea level rise. Precise determination of sea level rise in the study area improves our understanding of secular, instrumentally observed, variations across the Mediterranean. © 2012 Wiley Periodicals, Inc.     

North Indian Ocean warming and sea level rise in an OGCM

The variability in the long-term temperature and sea level over the north Indian Ocean during the period 1958–2000 has been investigated using an Ocean General Circulation Model, Modular Ocean Model version 4. The model simulated fields are compared with the sea level observations from tide-gauges, Topex/Poseidon (T/P) satellite, in situ temperature profile observations from WHOI moored buoy and sea surface temperature (SST) observations from DS1, DS3 and DS4 moored buoys. It is seen that the long (6–8 years) warming episodes in the SST over the north Indian Ocean are followed by short episodes (2–3 years) of cooling. The model temperature and sea level anomaly over the north Indian Ocean show an increasing trend in the study period. The model thermocline heat content per unit area shows a linear increasing trend (from 1958–2000) at the rate of 0.0018 × 10¹¹ J/m² per year for north Indian Ocean. North Indian Ocean sea level anomaly (thermosteric component) also shows a linear increasing trend of 0.31 mm/year during 1958–2000.     

Environmental change in the Ría de Vigo,NW Iberia,since the mid‐Holocene: new palaeoecological and seismic evidence

Pollen analyses (pollen, dinocysts and others) were combined with high‐resolution seismic‐stratigraphy sequences to reconstruct environmental dynamics and regional sea‐level (RSL) changes in a ria in NW Iberia. The chronological framework was established using radiocarbon dating and pollen markers that are related to a number of historical events. Major intensifications of the regional upwelling regime occurred during predominant NAO positive stages, dated to c. 4600–4300, 3800–3600, 3200–2700 and 2600–2400 cal. a BP. The regional Early Highstand System Tract spans from before 4500 to c. 3200 cal. a BP. During this period RSL was still rising, but several short episodes of higher terrestrial influence were detected between c. 4300–3800 and 3600–3200 cal. a BP. A readjustment occurred between 3200 and 2300 cal. a BP, including the first stage of relative sea‐level drop (2.8 ka event) dated to 3200–2800 cal. a BP, with the RSL recovering between 2800 and 2300 cal. a BP. The subsequent Later Highstand System Tract, after 2300 cal. a BP, corresponds to a final period of rising sea level that caused the final inundation of San Simón Bay. Since then, only minor changes in relative sea level can be postulated (e.g. towards the end of the Roman Period, in the Middle Ages and during the Little Ice Age). This new evidence is consistent with most of the available palaeoecological and historical information, but it provides a more detailed, near‐complete succession of simultaneous changes occurring in both the terrestrial and the marine ecosystems.     

格陵兰岛冰盖消融时空特征(2003~2015年)及其对海平面上升的贡献

研究格陵兰冰盖（GrIS）质量变化异常速率可以帮助了解异常气候事件驱动海平面变化的机制.聚焦于2010~2012年GrIS质量变化的异常速率,及其对海平面指纹（SLF）和相对海平面（RSL）变化的贡献.通过联合2003~2015年GRACE月重力场数据和表面质量平衡（SMB）数据,采用mascon拟合法及网格尺度因子恢复泄漏,获得了6个流域的质量变化时空分布.基于海平面变化方程（SLE）并考虑负荷自吸引效应估算了SLF的空间分布.结果表明,2003~2015年间GrIS总质量变化速率分别为-288±7 Gt/a及-275±1 Gt/a;而在2010~2012年间速率相应地增加至-456±30 Gt/a及-464±38 Gt/a,该时期格陵兰西北海岸及东南沿海地区呈现出大量冰盖融化,其对海平面的贡献变化呈现倒“V”型（即先升后降）,而全球平均海平面变化呈现出明显的正“V”型（即先降后升）.另外,GrIS融化对海平面的贡献约为31%,造成全球平均RSL增加了0.07 cm/a,而对斯堪的纳维亚及北欧地区的RSL贡献为-0.6 cm/a,GrIS融化造成的远海地区RSL上升速率比全球平均RSL速率高近30%.      

Late Quaternary sea‐level change and evolution of Belfast Lough,Northern Ireland: new offshore evidence and implications for sea‐level reconstruction

The interplay of eustatic and isostatic factors causes complex relative sea‐level (RSL) histories, particularly in paraglacial settings. In this context the past record of RSL is important in understanding ice‐sheet history, earth rheology and resulting glacio‐isostatic adjustment. Field data to develop sea‐level reconstructions are often limited to shallow depths and uncertainty exists as to the veracity of modelled sea‐level curves. We use seismic stratigraphy, 39 vibrocores and 26 radiocarbon dates to investigate the deglacial history of Belfast Lough, Northern Ireland, and reconstruct past RSL. A typical sequence of till, glacimarine and Holocene sediments is preserved. Two sea‐level lowstands (both max. ?40 m) are recorded at c. 13.5 and 11.5k cal a bp . Each is followed by a rapid transgression and subsequent periods of RSL stability. The first transgression coincides temporally with a late stage of Meltwater Pulse 1a and the RSL stability occurred between c. 13.0 and c. 12.2k cal a bp (Younger Dryas). The second still/slowstand occurred between c. 10.3 and c. 11.5k cal a bp . Our data provide constraints on the direction and timing of RSL change during deglaciation. Application of the Depth of Closure concept adds an error term to sea‐level reconstructions based on seismic stratigraphic reconstructions.     

Holocene and Lateglacial relative sea‐level change in north‐west England: implications for glacial isostatic adjustment models

Relative sea‐level (RSL) change is reconstructed for central Cumbria, UK, based on litho‐ and biostratigraphical analysis from the Lateglacial to the late Holocene. The RSL curve is constrained using ten new radiocarbon‐dated sea‐level index points in addition to published data. The sea‐level curve identifies a clear Lateglacial sea‐level highstand approximately 2.3 m OD at c. 15–17 k cal a BP followed by rapid RSL fall to below ?5 m OD. RSL then rose rapidly during the early Holocene culminating in a mid‐Holocene highstand of approximately 1 m OD at c. 6 k cal a BP followed by gradual fall to the present level. These new data provide an important test for the RSL predictions from glacial isostatic adjustment models, particularly for the Lateglacial where there are very little data from the UK. The new RSL curve shows similar broad‐scale trends in RSL movement predicted by the models. However, the more recent models fail to predict the Lateglacial sea level highstand above present reconstructed by the new data presented here. Future updates to the models are needed to reduce this mismatch. This study highlights the importance for further RSL data to constrain Lateglacial sea level from sites in northern Britain. Copyright © 2012 John Wiley & Sons, Ltd.     

Mid‐ to late Holocene relative sea‐level change in Poole Harbour,southern England

A foraminiferal transfer function for mean tide level (MTL) is used in combination with AMS radiocarbon dated material to construct a record of relative sea‐level (RSL) change from Poole Harbour, southern Britain. These new data, based on multiple cores from duplicate sites, indicate four phases of change during the last 5000 cal. (calendar) yr: (i) rising RSL between ca. 4700 cal. yr BP and ca. 2400 cal. yr BP; (ii) stable to falling RSL from ca. 2400 cal. yr BP until ca. 1200 cal. yr BP; (iii) a brief rise in RSL from ca. 1200 cal. yr BP to ca. 900 cal. yr BP, followed by a period of stability; (iv) a recent increase in the rate of RSL rise from ca. 400–200 cal. yr BP until the present day. In addition, they suggest that the region has experienced long‐term crustal subsidence at a rate of 0.5 mm C¹⁴ yr^?1. Although this can account for the overall rise in MTL observed during the past 2500 yr, it fails to explain the changes in the rate of rise during this period. This implies that the phases of RSL change recorded in the marshes of Poole Harbour reflect tidal range variations or ‘eustatic’ fluctuations in sea‐level. Copyright © 2001 John Wiley & Sons, Ltd.     

Late Weichselian shore displacement in Halland, southwestern Sweden: relative sea-level changes and their glacio-isostatic implications

An analysis of Late Weichselian sediments from basins in southern and central Halland, southwestern Sweden was carried out. Together with evaluation of previously reported data from the area (isolated basins, highest shore levels, glaciofluvial delta levels, periglacial features, and early Holocene shore levels), this has provided the basis for constructing shore displacement curves for the area, covering the approximate time interval 14 000–9000 BP. There is a general fall in relative sea level (RSL) during this period. The average regression rate until the late Younger Dryas ( c . 10 300 BP) is 15–20 mm/yr. During the Younger Dryas the regression is possibly interrupted by minor transgressive phases. The RSL fall is rapid between c . 10 300 and 9500 BP. A halted glacio-isostatic rebound during the Younger Dryas, as the result of glacier growth, is suggested as a cause of the observed pattern of RSL change. Previously reported transgressive events during the early late-glacial phase are discussed with regard to global eustatic rise, glacio-isostatic loading and glaciolacustrine damming events.