• Aucun résultat trouvé

Le Pacifique Tropical

Dans le document La Variabilité Régionale du Niveau de la Mer (Page 175-192)

Les ˆıles du Pacifique sont pArcticuli`erement vuln´erables `a la mont´ee du niveau de la mer car elles pr´esentent souvent une densit´e de population forte et une tr`es faible ´el´evation au dessus du niveau de la mer. Par exemple Funafuti, un atoll des Tuvalus a une densit´e

de population de 1870 hab.km-2

et s’´el`eve seulement entre 0 et 5m au dessus du niveau de la mer. De plus ces ˆıles sont souvent sujettes a des mouvements de subsidence de la croˆute dˆus `a l’activit´e tectonique locale (isostasie thermique de la croˆute oc´eanique par exemple,Pirazzoli [1995]) ou `a l’activit´e anthropique (pompage de l’eau douce souterraine,

urbanisation, etc, Webb and Kench [2010]). Ces facteurs combin´es entre eux font de cette

r´egion l’une des plus expos´ees aux risques associ´es `a la mont´ee du niveau de la mer (Nicholls and Cazenave [2010]). A propos des ˆıles Tuvalu, le 4`eme

rapport du GIEC pr´ecise qu’elles sont menac´ees de submersions significatives voire de disparition totale au cours du XXI`eme si`ecle (Mimura et al.[2007]).

Sur la p´eriode altim´etrique on observe une forte augmentation du niveau de la mer dans l’Ouest du Pacifique (voir Fig. 3.1). En tendance, sur la p´eriode 1993-2010, le niveau de la

mer y a augment´e `a un rythme de 9 `a 14 mm.a-1

, i.e. 3 `a 4 fois plus vite que la moyenne globale sur la mˆeme p´eriode (voir Fig.1.12a). C’est l’augmentation la plus forte observ´ee dans le monde par l’altim´etrie. Elle a ´et´e attribu´ee `a une augmentation de l’intensit´e des vents d’Est sur la r´egion au cours des ann´ees 1990 (Timmermann et al. [2010]; Merrifield and Maltrud [2011]).

Cependant, avec seulement 18 ans d’altim´etrie, il est difficile de savoir si cette aug-mentation est repr´esentative des variations du niveau de la mer au cours des 60 derni`eres ann´ees car dans l’oc´ean Pacifique tropical ces variations sont loin d’ˆetre lin´eaires. Elles pr´esentent une tr`es forte variabilit´e inter-annuelle (environ ±20 cm dans l’Ouest du Paci-fique) avec une p´eriodicit´e de 2 `a 6 ans li´ee `a l’oscillation ENSO et aussi une forte variabilit´e multi-d´ecennale entre -10˚N et +10˚N comme le montre les mar´egraphes, l’altim´etrie et les

3.1 Le Pacifique Tropical

Figure 3.1 – Tendances du niveau de la mer mesur´ees par altim´etrie dans le Pacifique Ouest entre 1993 et 2009. Les r´egions hachur´ees indiquent les zones pour lesquelle la courbe du niveau de la mer local diff`ere significativement d’une droite (pvalue>0.1). Figure adapt´ee deBecker et al.[2011].

mod`eles (Church et al. [2006]; Merrifield [2011]). Nous avons donc propos´e d’analyser les variations du niveau de la mer dans cette r´egion avec les reconstructions pr´esent´ees plus haut dans l’Arcticle intitul´e :”Sea level variations at tropical Pacifique islands since 1950”

R´esum´e de l’Arcticle : ”Sea level variations at tropical Pacifique islands

since 1950” (l’Arcticle original est ins´er´e `a la fin de cette section 3.1).

Dans cette Arcticle, nous analysons la variabilit´e r´egionale du niveau de la mer d’origine climatique dans l’Ouest du Pacifique ainsi que le mouvement de la croˆute terrestre sur 12 ˆıles ´equip´ees de GPS pour estimer localement la mont´ee du niveau de la mer relative sur la p´eriode 1950-2009. La variabilit´e r´egionale du niveau de la mer est estim´ee `a partir de la

re-construction bas´ee sur le mod`ele DRAKKAR/NEMO d´evelopp´ee dans l’´etudeMeyssignac

et al.[2012]. Le mouvement vertical de la croˆute est estim´ee `a partir des solutions GPS de

Santamaria-Gomez et al.[2011],deFadil et al.[2011] et du service GPS international (IGS). Dans un premier temps nous v´erifions que la reconstruction repr´esente bien les variations spatio-temporelles du niveau de la mer dans la r´egion, en comparant avec l’altim´etrie sur la p´eriode 1993-2009 et avec 20 enregistrements mar´egraphiques non-utilis´es dans la recon-struction, sur la p´eriode 1950-2009. La reconstruction montre que le niveau de la mer en moyenne sur l’Ouest du Pacifique (20˚S-15˚N par 120˚E-135˚W) a augment´e `a la vitesse de 1.8 ± 0.5 mm.a-1

depuis 1950 en accord avec la moyenne globale. La variabilit´e r´egionale est forte dans la r´egion, en pArcticulier autour des ˆıles Tuvalus, avec des tendances qui at-teignent 4 mm.a-1sur la p´eriode 1950-2009. Les variations spatio-temporelles du niveau de la mer s’expliquent par les variations spatio-temporelles de l’expansion thermique et sont fortement li´ees `a l’oscillation ENSO. A l’´echelle inter-annuelle, la signature des ´ev`enements El Ni˜no/La Ni˜na est tr`es forte de l’ordre de 20 `a 30cm. On observe aussi une forte

vari-abilit´e multi-d´ecennale avec moins d’amplitude dans le signal r´egional du niveau de la mer avant 1970 et apr`es 2002 et plus d’amplitude entre ces 2 dates. Ceci confirme les r´esultats de Church et al. [2006] pour la p´eriode avant 1970. Pour la p´eriode apr`es 2002, la baisse d’amplitude observ´ee est li´ee `a l’occurence de El Ni˜no/La Ni˜na modoki (Ashok et al.[2007]) plutˆot que de El Ni˜no/La Ni˜na classiques (Behera and Yamagata [2010]). Ceci sugg`ere que la variabilit´e multi-d´ecennale observ´ee dans le niveau de la mer `a l’Ouest du Pacifique s’expliquerait par des variations dans la nature des ´ev`enements El Ni˜no/La Ni˜na plutˆot que dans leur amplitude. L’analyse des s´eries GPS nous a permis d’estimer de mani`ere fiable le mouvement vertical local de la croˆute pour 9 ˆıles de la r´egion : Guam, Pohnpei, Roratonga, Nauru, Funafuti, Pago Pago, Papeete, Noumea et Tarawa. Pour chacune de ces ˆıles, le mouvement vertical n’est pas n´egligeable en comparaison avec la mont´ee locale du niveau de la mer. Ceci montre que lorsqu’on estime le niveau de la mer relatif, il est essentiel de prendre en compte `a la fois les facteurs climatiques qui influencent le niveau de la mer et les autres facteurs qui influencent les mouvements de la croˆute. Pour Guam, Pohnpei et Roratonga, nous trouvons une mont´ee du niveau de la mer relatif en accord

(avec les barres d’erreur) avec la mont´ee du niveau de la mer global (1.8 mm.a-1

). En revanche, pour Nauru, Funafuti, Pago Pago, Papeete, Noumea et Tarawa elle est significa-tivement sup´erieure. Pour Funafuti elle est pArcticuli`erement forte et s’´el`eve `a 5 mm.a-1 sur la p´eriode 1950-2009 confirmant les inqui´etudes de la population locale. Il est `a noter que plus de 15% de cette ´el´evation est dˆue `a la subsidence locale de l’ˆıle dont l’origine est en partie l’activit´e anthropique (voir Webb and Kench [2010]).

Sea level variations at tropical Pacific islands since 1950

M. Beckera,, B. Meyssignaca, C. Letetrelb, W. Llovelc, A. Cazenavea, T. Delcroixa

aLEGOS, UMR5566/CNES/CNRS/UPS/IRD, Toulouse, France

bLIENSs, UMR6250/CNRS/University of La Rochelle, France

cJet Propulsion Laboratory, Caltech, Pasadena, USA

a b s t r a c t a r t i c l e i n f o

Article history:

Received 1 April 2011 Accepted 9 September 2011 Available online 18 September 2011

Keywords:

sea level rise sea level variability tide gauge satellite altimetry ENSO Pacific Islands

The western tropical Pacific is usually considered as one of the most vulnerable regions of the world under present-day and future global warming. It is often reported that some islands of the region already suffer sig-nificant sea level rise. To clarify the latter concern, in the present study we estimate sea level rise and vari-ability since 1950 in the western tropical Pacific region (20°S–15°N; 120°E–135°W). We estimate the total rate of sea level change at selected individual islands, as a result of climate variability and change, plus ver-tical ground motion where available. For that purpose, we reconstruct a global sea levelfield from 1950 to 2009, combining long (over 1950–2009) good quality tide gauge records with 50-year-long (1958–2007) gridded sea surface heights from the Ocean General Circulation Model DRAKKAR. The results confirm that El Niño-Southern Oscillation (ENSO) events have a strong modulating effect on the interannual sea level vari-ability of the western tropical Pacific, with lower/higher-than-average sea level during El Niño/La Niña events, of the order of ±20–30 cm. Besides this sub-decadal ENSO signature, sea level of the studied region also shows low-frequency (multi decadal) variability which superimposes to, thus in some areas amplifies current global mean sea level rise due to ocean warming and land ice loss. We use GPS precise positioning records whenever possible to estimate the vertical ground motion component that is locally superimposed to the climate-related sea level components. Superposition of global mean sea level rise, low-frequency re-gional variability and vertical ground motion shows that some islands of the region suffered significant

‘total’sea level rise (i.e., that felt by the population) during the past 60 years. This is especially the case for the Funafuti Island (Tuvalu) where the“total”rate of rise is found to be about 3 times larger than the global mean sea level rise over 1950–2009.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Sea level rise is a major consequence of global warming, which threatens many low-lying, highly populated coastal regions of the world. The western tropical Pacific Island Region (hereafter called PIR), defined here as the area located between 20°S and 15°N latitude, and 120°E and 135°W longitude, is usually considered as one of the most vulnerable world regions under future sea level rise (Nicholls and Cazenave, 2010). This region is indeed characterized by volcanic archipelagos composed of low-lying islands and atolls where climate-related sea level rise may amplify other stresses caused by natural phe-nomena (e.g., vertical ground motions due to tectonics and volcanism, as well as occurrence of extreme events like storm surges) or human ac-tivities (e.g., ground subsidence due to ground water and/or oil extrac-tion, urbanizaextrac-tion, etc.) (Nicholls et al., 2007). In many cases (e.g., the Tuvalu and Kiribati island chains), much of the land altitude above pre-sent mean sea level rarely exceeds 5 m. The combination of low

elevation, small island size, sensitivity to change in boundary conditions (coastal sea level, waves and currents) and in some cases, high popula-tion density, is a matter of concern, as there is little doubt that sea level will continue to rise in the future (IPCC, 2007). Regularly, the media highlight the case of tropical island inhabitants, referring them as the

first climatic refugees of current global warming. For example, in the Tuvalu islands, it is common to hear that people already see the impacts of sea level rise (Nicholls et al., 2007) and that the situation will worsen in the future.

Tide gauge observations indicate that global mean sea level rose at an average rate of ~1.7 mm/yr since 1950 (Church and White, 2006; Jevrejeva et al., 2006; Holgate, 2007). Satellite altimetry reports faster global mean sea level rise since 1993, of 3.3 +/−0.4 mm/yr (Leuliette et al., 2004; Nerem et al., 2006, 2010; Ablain et al., 2009). However, sea level rise is far from being spatially uniform. Global coverage of satellite altimetry data shows that in the western tropical Pacific re-gion, sea level rose at a rate up to 3–4 times larger than the global mean between 1993 and 2010 (Cazenave and Llovel, 2010; Nerem et al., 2010). Unfortunately, the satellite altimetry record is still too short to conclude that the PIR also displays on the long-term (e.g., since 1950), rates of sea level rise several times higher than the global Global and Planetary Change 80-81 (2012) 85–98

⁎Corresponding author at: 18 Av. E. Belin, 31400 Toulouse, France. Tel.: +33 5 61 33 30 03.

E-mail address:melanie.becker@legos.obs-mip.fr(M. Becker). 0921-8181/$–see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.gloplacha.2011.09.004

Contents lists available atSciVerse ScienceDirect

Global and Planetary Change

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / g l o p l a c h a

Mitchum and Lukas, 1990), satellite altimetry (Merrifield et al., 1999), and modeling results (Busalacchi and Cane, 1985) have shown that the El Niño-Southern Oscillation (ENSO) phenomenon has strong impact on the interannual variability of sea level in that region. In the PIR, El Niño/La Nina events correspond to sea level lows/highs of ~20–

30 cm compared to normal conditions.

In the present study, we estimate sea level change and variability in the PIR since 1950. We then assess the total rate of change at se-lected PIR islands, as a result of the climatic signal (uniform-global mean-sea level rise plus regional variability) and vertical ground mo-tion. The objective of this work is indeed to determine the amount of

“total”sea level change effectively felt by the populations over the last ~ 60 years. For that purpose, we developed a past sea level recon-struction from 1950 to 2009 (this is an update of an earlier work by

Llovel et al., 2009, with a few methodological improvements—see

Section 2.3). The reconstruction method combines nearly one hun-dred, long (1950–2009) good quality tide gauge records with 50-year-long (1958–2007) gridded sea surface heightfields from an Ocean General Circulation Model (OGCM), the DRAKKAR model with-out data assimilation (Barnier et al., 2006; Madec, 2008; Dussin et al., 2009). Sea level is reconstructed globally over a 60-year long time span (1950–2009) (Meyssignac et al., 2011), but here we focus on the PIR and analyze the reconstructed regional variability superim-posed to the uniform global mean sea level rise. The reconstructed sea level is compared to 27 tide gauge records available at PIR island sites equipped with tide gauges (in general, these tide gauge records do not cover the whole 60 years, and just a few of them, 7 out of 27, were used in the global sea level reconstruction). When possible, we also estimate vertical crustal motions deduced from GPS (Global Positioning System) solutions. This allow us to estimate the“total”

sea level rise felt locally at a few PIR islands, i.e., the climatic compo-nents (global mean rise plus regional variability) and ground motion. We focus on Funafuti (Tuvalu Islands) because it is there that the largest total sea level rise is observed over the 1950–2009 time span. The paper is organized as follows. InSection 2we describe the data sets used in this study and present the past sea level reconstruc-tion. InSection 3, we analyze sea level spatial patterns in the PIR over 1950–2009 (the 60-year long reconstruction). InSection 4, we ana-lyze three distinct sub-regions under ENSO influence and compare tide gauge-based, reconstruction-based and satellite altimetry-based sea level variations. We present the main results and address the question“how past and recent sea level changes have locally affected the PIR?”. InSection 5, we determine the“total”sea level trend due to the climatic component (uniform global mean rise plus regional var-iability), and vertical ground motion using GPS solutions.Section 6

summarizes our main results and conclusions.

2. Data

2.1. Satellite altimetry

For the altimetry data, we use the DT-MSLA“Ref”series provided by Collecte Localisation Satellite (CLS; http://www.aviso.oceanobs. com/en/data/products/sea-surface-height-products/global/msla/index. html). This data set is used over the time span from January 1993 to December 2009. It is available as 1/4°× 1/4° Mercator projection grids at weekly interval. The DT-MSLA“Ref”series are based on the combi-nation of several altimetry missions, namely: Topex/Poseidon (T/P), Jason-1 and 2, Envisat and ERS 1 and 2. It is a global homogenous inter-calibrated data set based on a global crossover adjustment using T/P and then Jason-1 as reference missions. Moreover, the use of recent orbit solutions for Jason-1 and T/P (GSFC—Goddard Space Flight Cen-ter-orbit computed with the ITRF2005 terrestrial reference frame; Alta-mimi et al., 2007) allows to remove previous heterogeneity between global hemispheric mean sea level trends (Ablain et al., 2009). ERS

Traon et al. (1998) and Ducet et al. (2000). Usual geophysical correc-tions are applied: solid Earth, ocean and pole tides, wet and dry tropo-sphere, ionosphere (see Ablain et al., 2009 for more details) and inverted barometer (Carrere and Lyard, 2003; Volkov et al., 2007).

2.2. Tide gauges

We use monthly mean sea level data from 27 tide gauge stations (listed inTable 1) included in the Revised Local Reference data set of the Permanent Service for Mean Sea Level (PSMSL;Woodworth and Player, 2003, http://www.pol.ac.uk/psmsl). We consider only tide gauges that have at least 30 years of data between 1950 and 2009. Note that only tide gauge records covering the whole studied period, 1950–2009, were considered to perform the global sea level reconstruction (seeSection 2.3). Where necessary, the PSMSL records were extended through 2009 using fast delivery data from the University of Hawaii Sea Level Center (UHSLC; http://ilikai.soest. hawaii.edu/uhslc). At 12 sites (Saipan, Yap, Pohnpei, Majuro, Kanton, Tarawa, Rabaul, Funafuti, Kapingamarangi, Christmas, Rikitea and Rarotonga), recent data are lacking in both PSMSL and UHSLC. Thus we completed the tide gauge time series using altimetry data beyond 1993. For that purpose, we interpolated the gridded altimetry data in a radius of 1° around the tide gauge position. Then the missing tide gauge data were replaced by the interpolated altimetry data. This was done only at sites with at least 5-year overlap and a correlation ≥0.9 between the tide gauge and altimetry time series. This 5-year overlap is deduced from the Bonett's formula (Bonett and Wright, 2000) and is based on a desired Pearson's correlation coefficient of 0.9 ± 0.1, with 95% level of confidence (for more details on the meth-od, seeBonett and Wright, 2000).

Tide gauge location (latitude, longitude) and data length are listed inTable 1. We corrected the tide gauge time series for the inverted barometer response of sea level to atmospheric loading using surface pressurefields from the National Centers for Environmental Predic-tion (NCEP) (Kalnay et al., 1996) (data available at http://www. ncep.noaa.gov/). Tide gauge data were also corrected for the Glacial Isostatic Adjustment— GIA— effect using the ICE5G-VM4 model fromPeltier (2004). The GIA correction is small in the PIR and ranges between−0.1 mm/yr and−0.3 mm/yr. Thus using this particular GIA model versus another has little impact of the sea level results.

For a number of reasons (changes in instrumentation, earthquakes or other natural of anthropogenic factors; seeBecker et al., 2009), gaps and discontinuities may affect the tide-gauge time series. When small gaps (≤4 consecutive years) are observed in the tide-gauge record, we reintroduce missing data by linearly interpolating the time series. Outliers were detected using the Rosner's test (Rosner, 1975) and removed. Annual and semi-annual cycles were re-moved through a least-squaresfit of 12-month and 6-month period sinusoids. In a last step, to be consistent with the time resolution of the sea level reconstruction (seeSection 2.3), we averaged monthly tide gauge time series to obtain annual averages.

2.3. Past sea level reconstruction

Several previous studies have attempted to reconstruct past de-cade sea level in two dimensions (2-D), combining sparse but long tide gauge records with global gridded (i.e., 2-D) sea level (or sea level proxies) time series of limited temporal coverage (Smith, 2000; Chambers et al., 2002; Church et al., 2004; Berge-Nguyen et al., 2008; Llovel et al., 2009; Ray and Douglas, 2010; Church and White, 2011). Satellite altimetry available since 1993 has shown that sea level is not rising uniformly. But altimetry-based spatial trend patterns essentially reflect decadal variability rather than low-frequency trends because the altimetry record is still short. Thus on longer time spans, regional sea level trends are expected to be

Table 1

Locations, time spans and sea-level trends for the PIR tide gauges when the tide gauges are extended by altimetry. The correlation coefficient and the root mean square with the annual reconstructed sea level (RESL) are given for the tide gauge time span. The RESL trends and errors from 1950 to 2009 are also given in the two last columns. The symbol * corresponds to significant trends (p-valueb0.1). Stars★correspond to tide gauges used in the reconstruction.

Station group

Station Tide-gauge RESL (tide gauge time

span)

RESL 19502009 Lon Lat Start End (extended by

Dans le document La Variabilité Régionale du Niveau de la Mer (Page 175-192)