• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Long-term Regional Dynamic Sea Level Changes from CMIP6 Projections

    2021-02-26 08:22:06BrunoFERREROMarcosTONELLIFernandaMARCELLOandIlanaWAINER
    Advances in Atmospheric Sciences 2021年2期

    Bruno FERRERO, Marcos TONELLI, Fernanda MARCELLO, and Ilana WAINER

    University of S?o Paulo, S?o Paulo 05508120, Brazil

    ABSTRACT

    Key words:dynamic sea level,CMIP6,sea level rise,signal-to-noise,time of emergence

    1.Introduction

    Even though human-induced sea level rise (SLR) will continue over the next century at least (Church et al., 2013),the expected regional expression of its global average is rather variable across the ocean basins (Slangen et al., 2014;Bordbar et al., 2015; Meyssignac et al., 2017), influenced by a range of processes (Clark et al., 2015; Slangen et al.,2017). This regional distribution is influenced both by anthropogenic and natural external forcings and by intrinsic climate variability (Marcos et al., 2017). Detection consists of determining if a given signal actually corresponds to an externally forced change or simply falls within possible fluctuations from natural internal variability of the coupled climate system (Stott et al., 2010).

    Large ensembles (LEs) of climate simulations offer a powerful approach to assess climate change detection.While single-model LEs target the uncertainty arising from internal climate variability (i.e., from natural interactions in the coupled ocean–atmosphere–land–biosphere–cryosphere system), multi-model ensembles also include structural uncertainty (arising from differences in model formulation), but offer, in turn, unique insights on the forced climate response since they provide information from the perspective of diversified efforts to simulate the climate system. Despite the exclusive value of a single-model LE in addressing more specifically the intrinsic climate variability, there is no evidence that any particular model is more realistic at climate projections than others of its class (Deser et al., 2020).

    In LE analysis, the ensemble mean, which is referred to as the signal (S), represents the forced response (i.e., the anthropogenic climate change); while the ensemble spread,or noise (N), represents the uncertainty arising from two sources: internal variability and model structural differences. Therefore, ensemble spread in single-model LE arises only from internal variability, while in multi-model LEs the spread is due to both the model configuration and internal variability. A third source of projection uncertainty is associated with the possible radiative forcing scenarios.

    Little et al. (2015) assessed the sources of uncertainty for modeled sea level change in CMIP5 projections and suggested that the combined effect of temperature biases,upper-ocean stratification, and vertical mixing impact the thermosteric sea change across models. Moreover, they discuss that differences in atmospheric models lead to discrepancies in surface fluxes and feedback, which increase uncertainties in multi-model LEs.

    The Intergovernmental Panel on Climate Change(IPCC) assessments rely on multi-model climate projections based on new alternative development scenarios of future emissions and land-use changes [the Shared Socioeconomic Pathways, or SSPs (O’Neill et al., 2016) and the forcing levels of CMIP5’s Representative Concentration Pathways, or RCPs (van Vuuren et al., 2011)], produced with integrated assessment models participating in phase 6 of the Coupled Model Intercomparison Project (CMIP6; Eyring et al., 2016).

    The SSPs’ narrativeillustrate possible anthropogenic drivers of climate change over the 21st century (departing from the historical runs) ranging from sustainable to fossilfueled development (Riahi et al., 2017):

    ● SSP1 —Sustainability —Taking the Green Road:Low challenges to mitigation and adaptation;

    ● SSP2 —Middle of the Road: Medium challenges to mitigation and adaptation;

    ● SSP3 —Regional Rivalry —A Rocky Road: High challenges to mitigation and adaptation;

    ● SSP4 —Inequality —A Road Divided: Low challenges to mitigation, high challenges to adaptation;

    ● SSP5 —Fossil-fueled Development —Taking the Highway: High challenges to mitigation, low challenges to adaptation.

    Beyond improving the understanding of the climate system and characterizing societal risks and response options,numerical climate change projections provide relevant information regarding the emergence of anthropogenically forced trends above internal variability (Carson et al., 2019). This particular time, when S exceeds N above a particular threshold with no turning back below it, is referred to as the time of emergence (ToE).

    The dynamic sea level (DSL) is the spatiotemporally dependent sea surface topography referenced to the Earth’s geoid and it is influenced by ocean currents, local mass balance and density changes in the water column (Cazenave and Remy, 2011; Griffies and Greatbatch, 2012; Gregory et al., 2013; Richter et al., 2013). DSL does not count for global mean SLR and does not contain any other sea level signal, such as zero global mean thermosteric sea level, land ice melt, land motion, or inverse barometer effects; it is defined to have a global mean of zero (Gregory et al.,2019). According to several past studies, regional sea level changes are usually detected by examining DSL changes(e.g., Slangen et al., 2014; Bordbar et al., 2015; Hu and Bates, 2018).

    Here, we investigate the most up-to-date projected DSL for the 21st century as simulated by state-of-the-art global climate models and Earth System Models (ESMs) under the auspices of the CMIP6 project (Table 1). First, the results from the 50-member CanESM5 ensemble (Swart et al., 2019) are assessed for two historical CMIP6 experiments: historical(1850 to 2014, full-forcing) and historical-natural (1850 to 2020; natural-only with no anthropogenic forcing). The trends in sterodynamic sea-level [DSL plus global mean thermosteric sea-level, as defined in Gregory et al. (2019)] from these experiments are compared for consistency against satellite altimetry data from the Archiving, Validation and Interpretation of Satellite Oceanographic data (AVISO, https://www.aviso.altimetry.fr).

    Projected DSL responses to three distinct SSPs are then presented for the 50-member CanESM5 ensemble mean and a CMIP6-27-model ensemble mean. The three CMIP6 scenarios used are: the SSP1-2.6 (sustainability, low emissions,mitigation – year 2100 forcing of 2.6 W m), SSP3-7.0 (business as usual, medium to high emissions — year 2100 forcing of 7.0 W m), and SSP 5-8.5 (fossil-fueled development, high emissions — year 2100 forcing of 8.5 W m).Our goal is to provide a global picture of projected regional DSL based on the agreement between both the CanESM5 and the CMIP6-27-model ensemble sets. Furthermore, we discuss the projected DSL outcomes from distinct SSP scenarios and how they differ between the single- and multimodel ensemble approaches.

    2.Materials and methods

    Two sets of experiments are used in this study: one from a single-model LE, in which ensemble members differ only by small round-off level temperature variations in their initial condition fields, and one corresponding to a multi-model LE, composed of 27 different models from the CMIP6 archive. Our focus is to analyze the future projection experiments (SSPs) for both these sets. These are the three “Tier 1 ScenarioMIP projections” that were available at the time of writing: SSP1-2.6, SSP3-7.0, and SSP5-8.5 –covering the period from 2015 to 2100. Besides the SSP experiments, a set of two historical experiments are also analyzed for the single-model LE, as detailed below.

    2.1.Initial condition CanESM5 ensemble

    For the single-model analysis we employ the 50-member ensemble output from the Canadian Earth System Model version 5 (CanESM5, Swart et al., 2019) retrieved from the CMIP6 archive (hereafter “CanESM5 ensemble”).To compare with satellite-derived observational data, we use the historical scenario with full forcing (i.e., anthropogenic plus natural external forcings and natural internal variability, from 1850 to 2014, Fig.1b) and the historical scenario with natural forcings only (10 members with all-natural external forcings, e.g., volcanoes, solar, but no anthropogenic emissions, from 1850 to 2020, Fig.1c). Each historical realization starts at a different year (with 50-year intervals) from the piControl experiment, yielding differences in multidecadal ocean variability among members (Swart et al., 2019).

    Fig.1. Annual mean linear trend (mm yr?1) for the years 1993–2018 of (a) total sea level from satellite observations(AVISO+), (b) sterodynamic sea-level (DSL plus global mean thermosteric sea level) from CanESM5 historical + SSP585(50 members average) and (c) from CanESM5 historicalnatural (10 members average). The altimeter-derived sea levels refer to the ocean topography with respect to the geoid. Figure A1 in the appendix provides information about the CanESM5 intra-ensemble spread in the observed period.

    There are two subset variants within this ensemble: 25 members use a conservative wind-stress field interpolation passed from the atmospheric model to the ocean model and the other 25-member subset uses a bilinear remapping scheme. These variants do not produce distinguishable responses on transient climate or global scale dynamics(Swart et al., 2019); hence, for our purposes, we can assume this to be a 50-member ensemble with the spread generated only by initial condition characteristics. The analyses were conducted using yearly means of the CanESM5 ensemble DSL results.

    2.2.CMIP6 multi-model ensemble

    The outputs from the 27 Earth System Models (ESMs)used in this study (Table 1; hereafter “CMIP6 ensemble”)are from the CMIP6 archive. The models selection was based on the availability of the DSL variable (or “zos” in CMIP convention — with zero global-area mean and not including inverse barometer depressions from sea ice) for the three “Tier 1 ScenarioMIP projections”. The model output submissions for the CMIP6 archive do not have the same number of ensemble members, so to compute our multi-model ensemble mean of the projected sea level change for the 21st century we are using a single ensemble member from each model (usually the ‘r1i1p1f1’), and every model is given the same weight for the ensemble statistics. To provide an estimate of the internal variability we are also using the last 200 years from the control run (under constant pre-industrial forcing: piControl).

    2.3.Methods

    Prior to performing the CMIP6 ensemble analysis, we interpolated the DSL data from 27 different models onto a common 1° × 1° grid using bilinear interpolation with the same land–ocean mask, excluding the marginal seas and interior lakes like the Mediterranean Sea, Red Sea, Arabian Gulf, Black Sea, Caspian Sea, Baltic Sea, and Hudson Bay.

    Four key metrics in LE analysis were then obtained for the regional DSL data for both the CanESM5 and the CMIP6 ensemble results over the 21st century for each of the three SSPs:

    (i) S — the forced response — which is the ensemble mean of the absolute linear trend in annual mean DSL spanning the period from 2015 to 2100. Trends are used to estimate sea level changes as well as other climate signals in order to highlight nonstationary behavior in the time series.This low-frequency signal can be assessed in terms of prediction, therefore providing a clearer indication of the future long-term movements in the series (Visser et al., 2015) and playing a role in climate change detection (Church et al.,2013).

    (ii) N — the ensemble spread representing uncertainty –which is the standard deviation from the linear trend field among the ensemble members. Using model spread to define N should include additional model errors, whereas in the real-world N should be natural forcing and internal variability only.

    (iii) S/N, as the ratio between the ensemble mean andthe ensemble spread, i.e., between forced response and uncertainty.

    Table 1. Earth System Models used.

    (iv) ToE — the forced response detection time —defined as the decade over which a trend (S) will be statistically above the unforced sea level internal variability [i.e., N(Carson et al., 2019)]. Here, ToE is computed as the year when the DSL time series (with no thermosteric component) at each grid point exceeds two standard deviations of the monthly mean DSL from the piControl experiment,using the same approach as in Bordbar et al. (2015) and Lyu et al. (2014). The computation was performed for each ensemble member separately, and the resulting ToEs were averaged to obtain an ensemble mean ToE.

    The CanESM5 and CMIP6 ensembles were both de-drifted using their respective piControl runs to remove potential spurious trends caused by model equilibrium adjustment rather than by external forcing. Results do not change significantly, consistent with Gupta et al. (2013), who assessed surface properties other than DSL to report that“when considering multimodel means of surface properties,drift is negligible”. The authors also investigated the steric sea level, which integrates the water column, to conclude that the deep ocean can be dominated by model drift.

    3.Results

    The regional patterns of sea level change obtained from satellite-derived observations (Fig.1a) arise from a variety of combined factors within the climate system, which, in turn, emerge from internal climate processes as well as from external forcings. Numerical experiments run with transient forcings enable us to untangle the sea level response to external forcings emerged from natural and anthropogenic drivers, as in Figs. 1b and c, respectively. As argued by Swart et al. (2019), the trends identified in CanESM5 historical-natural simulations (Fig.1c) “are far smaller than would be expected from anthropogenically forced trends, confirming that the model is suitably stable to evaluate centennial-scale climate change.”

    The sterodynamic sea level trends (with global mean thermosteric sea level included) from 1993 to 2018 indicate good agreement between the satellite altimetry product(Fig.1a) and the CanESM5 ensemble (Fig.1b), both depicting major global-scale features such as the midlatitude band of positive trends. When comparing the results from both the altimetry data (Fig.1a) and the CanESM5 full-forcing(Fig.1b) against the CanESM5 historical-natural (Fig.1c;note different color scale) — the latter accounting only for natural external forcings and internal variability — it can be noted that the most evident trends in Figs. 1a and b generally depart from the trends of the spatial pattern with no anthropogenic forcings (Fig.1c). This suggests that a scenario in which only natural forcings existed would still bear the regional distribution of the full-forcing SLR scenario,albeit with much smaller magnitudes, as expected. The differences seen between the altimetry data (Fig.1a)/CanESM5 full-forcing results (Fig.1b) and the CanESM5 historical-natural are therefore attributable to anthropogenic external forcing.

    Although sea level changes have been best explained by models driven by both natural and anthropogenic forcings, the anthropogenic forcing still seems to be the leading factor in explaining the magnitude of observed changes,while most of the variability in models seems to be caused by natural forcing, as shown by Slangen et al. (2014), when comparing full-depth observations of thermosteric sea level change to a range of single-forcing experiments done with 28 CMIP5 climate models. This study, as well as that of Marcos and Amores (2014), shows that the majority of observed global mean thermosteric sea level distribution in the second half of the 20th century is of anthropogenic origin.

    Results for the future projections show common regional sea level changes relative to the global mean in both the CanESM5 and the CMIP6 ensembles (Figs. 2 and 3). The tendency to greater SLR (depicted by S) appears strongly in the Arctic (where N is also higher), the northwestern Pacific, the northern extension of the Gulf Stream, the North Indian Ocean, as well as in the northern limits of the Antarctic Circumpolar Current (ACC). In the Southern Ocean, there is a common negative change closer to the Antarctic Continent and in the southeastern Pacific.

    These trends, either positive or negative, are more intense in the SSP5-8.5 scenario and display a regional distribution consistent with previous studies. Based on an ensemble of 21 CMIP5 model projections, the RCP4.5 relative regional sea level anomaly shown by Slangen et al.(2014) [e.g., Fig.7 in Slangen et al. (2017)] is similar to the spatial patterns displayed in Figs. 2 and 3. The authors also found the highest regional sea-level changes along the North Atlantic coastal regions, and along the ACC. Features such as DSL rise in high latitudes and polar regions, as well as the Southern Ocean belt-like structure (“meridional dipole”), are also robust in CMIP3 and CMIP5 model ensembles (Yin et al., 2010; Pardaens et al., 2011; Yin,2012; Bilbao et al., 2015). Likewise, Carson et al. (2015)found large negative sea surface height changes in the Southern Ocean, relative to the global signal.

    By analyzing DSL projections from both CMIP5 and CMIP6 models, Lyu et al. (2020) concluded that they exhibit very similar features and intermodel uncertainties. Relevant improvements from CMIP5 to CMIP6 include a better representation of the location of Southern Hemisphere westerly wind stress, which in turn could be partially responsible for improving the Southern Ocean simulated mean sea level. Larger DSL changes in the North Atlantic and Arctic are also projected by CMIP6 models, which are associated with a larger weakening of the Atlantic Meridional Overturning Circulation. It is suggested that the inclusion of models with larger climate sensitivity in CMIP6, in comparison with that of the CMIP5 ensemble, might have contributed to this increase in projected DSL.

    Fig.2. Top: CanESM5 ensemble-mean of the linear DSL trends (mm yr?1). Middle: Spread of the ensemble trends (DSL trends’standard deviation, mm yr?1). Bottom: The signal-to-noise ratio. The DSL trends were computed for all 50 members between 2015 and 2100 for the three SSP projections (left: SSP1-2.6; center: SSP3-7.0; right: SSP5-8.5). Figures A2 and A3 show the time series of CanESM5 global mean thermosteric sea level and local sterodynamic sea level (1850–2100) for the historical simulation and the assessed SSP scenarios.

    Fig.3. CMIP6 multi-model mean (top) linear DSL trends (mm yr?1), spread of multi-model trends (DSL trends’ standard deviation,mm yr?1; middle), and the signal-to-noise ratio (bottom). The DSL trends were computed for all 27 models between 2015 and 2100 for the three SSP projections (left: SSP1-2.6; center: SSP3-7.0; right: SSP5-8.5).

    For the CanESM5 ensemble, greatest N, i.e., largest spread (Fig.2, middle) occurs mainly at high latitudes, with values greater than 1 in the Arctic and the Southern Ocean.High values of projected DSL spread indicate an uncertainty that arises from internal climate variability. N does not vary significantly among the higher emissions projections, but is particularly larger in the Arctic for the SSP1-2.6. This suggests that the uncertainty associated with internal variability does not accompany the change in external forcing magnitude in our results, direct or indirectly. Additionally, for the CanESM5 ensemble, N is larger in regions where S is higher, suggesting that there is more uncertainty in the DSL projection over regions expected to see the biggest changes, in agreement with the results from Hu and Bates (2018) and Slangen et al. (2015).

    The great advantage of using the S/N is provided by the ability to assess the significance of the DSL trends. For instance, Hu and Deser (2013) discuss that the magnitudes of the sterodynamic trends in the Southern Ocean and Arctic (2000–60) are smaller than the 95% uncertainty, and therefore statistically non-significant. Here, the magnitude (S)and spread (N) of the DSL trends are somewhat similar in the Arctic for both CanESM5 (Fig.2) and CMIP6 (Fig.3) results. Nonetheless, the North Atlantic dipole pattern along the Gulf Stream (North Atlantic Current) is rather clear and displays high S/N values in both ensembles for the scenarios SSP3-7.0 and SSP5-8.5 (Figs. 2 and 3), which seems to be linked to the weakening of the Meridional Overturning Circulation (Yin et al., 2009; Pardaens et al., 2011;Bouttes and Gregory, 2014; Hu and Bates, 2018). Carson et al. (2015) also reported high S/N values along the northeastern coast of the United States for results from the Max Planck Institute Earth System Model with low resolution,while the combined analysis (including CMIP5 outputs) of Carson et al. (2016) suggested that, among densely populated regions, the New York City and the northeast of North America are projected to undergo the largest changes in relative sea level during the 20th and 21st centuries.

    For both the CanESM5 and CMIP6 ensembles, the regional changes projected by the SSP3-7.0 and SSP5-8.5 scenarios are fairly similar, while standing out in comparison with the changes projected by the lower-emission SSP1-2.6 (Figs. 2 and 3). This is consistent with the small differences found between the middle-RCP4.5 and the stronger-RCP8.5 forcing scenarios, by Carson et al. (2015),using 21 models from the CMIP5 archive.

    In the Southern Ocean, noticeable S/N values point to high trends — more so for the SSP5-8.5 than for the SSP3-7.0 scenario, with S reaching five times the magnitude of N in the CMIP6 ensemble and up to 20 times in the CanESM5 ensemble. Forget and Ponte (2015) suggested that on long time scales wind stress variability will impact regional sea level variability across the global oceans. The stronger DSL response in the Southern Ocean has been associated with the strengthening of the westerlies (Landerer et al., 2014),which yields enhanced Ekman transport northward and intensifies the meridional sea surface height gradient. This in turn is intensified by an increased meridional temperature gradient.

    Fig.4. CanESM5 ToE: Decade when the DSL time series exceeds the range of natural variability (defined by two standard deviations of the monthly mean DSL simulated in the control experiment). The illustrated ToE corresponds to the averaged ToE from all 50 members.

    Fig.5. CMIP6 ToE: Decade when the DSL time series exceeds the range of natural variability (defined by two standard deviations of the monthly mean DSL simulated in the control experiment) in each model. The illustrated ToE corresponds to the averaged ToE from all 27 models.

    Furthermore, the magnitude of this strong DSL response in the Southern Ocean could be related to surface fluxes (Clark et al., 2015). Warm sea surface temperature(SST) biases in the Southern Ocean have been reported for CMIP5 simulations (Sallée et al., 2013; Meijers, 2014) to be linked to deficiencies in atmospheric processes (Hyder et al., 2018). The same atmospheric surface flux anomalies responsible for causing the SST bias (Hyder et al., 2018) are also the source of variations in heat and freshwater fluxes as well as a poleward shift of the westerly winds (Slangen et al., 2015). Ultimately, ACC changes due to wind stress variations are possibly the main drivers of the local DSL change.

    Figures 4 and 5 display the ToE for the CanESM5 and CMIP6 ensembles, respectively. The regional distribution of S/N (bottom of Figs. 2 and 3) is consistent with the spatial distribution of the decade of ToE from Figs. 4 and 5.Higher S/N can be associated with smaller uncertainties.Although the trend detection falls in similar regions for the three projections, SSP1-2.6 yields much later ToE covering a smaller area relative to the other SSPs. The CanESM5 DSL S emerges roughly two decades prior to the one of the CMIP6 ensemble, which reflects the much smaller natural forcing generated by the CanESM5 simulations, where N is a function only of initial conditions. For the CMIP6 simulations, N comes from both differences in initial conditions and also in model structure.

    Although the RCP8.5-CMIP5 results from Lyu et al.(2014) show that, as a result of large internal variability, the ToE of sea-level change for DSL only (their Fig.2a) occurs before the year of 2080 over only a small fraction of the ocean area, our CMIP6 (and also CanESM5) DSL results show ToEs occurring before 2080 over a majority of the global ocean, more in agreement with those results of Lyu et al. (2014) which consider the DSL plus the global mean thermosteric sea level (their Fig.2b).

    4.Conclusions

    We provide an assessment of 21st century projected changes in regional DSL as simulated within the auspices of the CMIP6 project in three “Tier 1 ScenarioMIP” scenarios:SSP1-2.6 (mitigation), SSP3-7.0 (business as usual), and SSP5-8.5 (high forcing). Detection analyses are performed on DSL outputs from a 50-member single-model ensemble mean (CanESM5) as well as from a multi-model ensemble mean of 27 distinct ESMs participating in CMIP6 (Table 1).The single- and multi-model analyses show a consistent forced response (signal) in regional sea level projections, in which regions displaying higher S/N (low uncertainty)match regions expecting earlier ToEs, overall. The DSL signal is projected to emerge following an upward (i.e., rising)trend in the Arctic and northwestern Atlantic, and following a downward trend (i.e., decline) in the Southern Ocean and Southeast Pacific—with respect to the global-area mean.

    Although internal variability can change in response to forcing (Lehner et al., 2020), here we show that as the signal progressively increases, the DSL noise does not vary substantially across different future scenarios. It means that variations in the external forcing do not seem to significantly impact the ensembles’ spread associated with the uncertainty arising from internal climate variability (and from model differences for the CMIP6 ensemble) with respect to the DSL results analyzed here. N is reduced in the single- relative to the multi-model LE, since the latter accounts for model differences other than just internal variability. In the CanESM5 ensemble, N is larger in regions where S is higher, suggesting that there is more uncertainty in the DSL projection over regions expected to see the largest changes[in agreement with Hu and Bates (2018) and Slangen et al.(2015)].

    The CanESM5 ensemble historical experiment driven by natural forcing alone cannot fully reproduce the current pattern of SLR, although its resulting regional sea level distribution resembles that of the full-forcing historical experiment, with relatively much reduced magnitudes, as expected.

    The SSP3-7.0 and SSP5-8.5 scenarios project similar and equivalent DSL changes — as an adjustment to the external forcing — whereas the changes over the 21st century in the sustainable scenario (SSP1-2.6) are shown to be mostly dominated by internal variability, i.e., DSL signals remain almost entirely within the envelope of internal climate variability. This is consistent with Lyu et al. (2014),who showed that the ToE for total sea level exhibits little dependence on the emissions scenario, and occurs considerably earlier than that for the surface warming. It should be noted, however, that the total sea level variable considered by the authors also accounts for the global thermosteric mean sea level, which certainly has an emissions scenario dependence, differently from DSL.

    Regional projections of sea level changes are strongly linked to global-scale processes that could exceed the range of local natural variability even in a more sustainable scenario. Nevertheless, by “Taking the Green Road” we would be approaching a reality where most of the DSL changes throughout the 21st century would be dominated by internal processes of the climate system rather than by external radiative forcing imbalance.

    Acknowledgements.

    We acknowledge the World Climate Research Programme, which, through its Working Group on Coupled Modelling, coordinated and promoted CMIP6. We thank the climate modeling groups for producing and making available their model output, the Earth System Grid Federation (ESGF) for archiving the data and providing access, and the multiple funding agencies that support CMIP6 and ESGF. Grants CNPq-MCTINCT-594 CRIOSFERA 573720/2008-8 and Coordenacao de Aperfeicoamento de Pessoal de Nivel Superior-Brasil (CAPES)-Finance Code 001; FAPESP 2015/506861; 2017/16511-5; 2018/14789-9.

    APPENDIX

    Fig.A1. CanESM5 sterodynamic sea level spread (standard deviation,mm yr?1) in the observed period (1993–2018).

    Fig.A2. Time series of global mean thermosteric sea level (cm) from the CanESM5 ensemble for each SSP departing from the historical simulation. Shaded area indicates the spread for each scenario.

    Fig.A3. Time series of local (45°W, 38°S) mean sterodynamic sea level (cm) from the CanESM5 ensemble for each SSP scenario departing from the historical simulation. Thin lines indicate individual ensemble members and solid lines indicate ensemble means for each scenario.

    午夜久久久久精精品| 国产在线男女| 国产伦精品一区二区三区四那| 欧美高清成人免费视频www| 久久久久免费精品人妻一区二区| 十八禁网站网址无遮挡 | 亚洲三级黄色毛片| 丝袜喷水一区| 亚洲欧美日韩东京热| 亚洲第一区二区三区不卡| 免费看a级黄色片| 久久99蜜桃精品久久| av播播在线观看一区| 国产三级在线视频| 久久久久精品久久久久真实原创| 日韩欧美三级三区| 日本爱情动作片www.在线观看| 美女xxoo啪啪120秒动态图| 寂寞人妻少妇视频99o| 精品人妻偷拍中文字幕| 亚洲怡红院男人天堂| 亚洲av福利一区| 嫩草影院精品99| 一区二区三区四区激情视频| 91精品伊人久久大香线蕉| 2021少妇久久久久久久久久久| 精品一区二区三区视频在线| 亚洲精品456在线播放app| 高清在线视频一区二区三区| 97热精品久久久久久| 国国产精品蜜臀av免费| 国产视频首页在线观看| 一二三四中文在线观看免费高清| 久久久久九九精品影院| 成人亚洲精品av一区二区| 国产大屁股一区二区在线视频| 99久久人妻综合| 亚洲人成网站在线观看播放| 色网站视频免费| 日本与韩国留学比较| 国产欧美另类精品又又久久亚洲欧美| 欧美性猛交╳xxx乱大交人| 日韩亚洲欧美综合| 大片免费播放器 马上看| 亚洲欧美一区二区三区国产| 欧美3d第一页| 久久久a久久爽久久v久久| 少妇熟女aⅴ在线视频| 1000部很黄的大片| 色综合亚洲欧美另类图片| 高清av免费在线| 久久6这里有精品| 久久6这里有精品| 午夜激情福利司机影院| 国产亚洲av片在线观看秒播厂 | 亚洲自拍偷在线| 一级a做视频免费观看| 18禁在线无遮挡免费观看视频| 免费看日本二区| 成年女人在线观看亚洲视频 | 少妇人妻一区二区三区视频| a级毛片免费高清观看在线播放| 亚洲综合精品二区| 中文字幕制服av| 国产日韩欧美在线精品| 欧美不卡视频在线免费观看| 国产乱人视频| 国产伦精品一区二区三区视频9| 激情五月婷婷亚洲| 一级毛片久久久久久久久女| 美女xxoo啪啪120秒动态图| 一级毛片久久久久久久久女| 免费观看性生交大片5| 搡老妇女老女人老熟妇| 黑人高潮一二区| 国产精品国产三级专区第一集| 99久久九九国产精品国产免费| 一级毛片黄色毛片免费观看视频| av专区在线播放| 最近2019中文字幕mv第一页| 一级a做视频免费观看| 日韩精品青青久久久久久| 久久久久久久久久成人| 亚洲av日韩在线播放| 国产成人91sexporn| 一级毛片久久久久久久久女| 麻豆成人午夜福利视频| 亚洲图色成人| 丝瓜视频免费看黄片| av黄色大香蕉| 亚洲三级黄色毛片| 欧美高清成人免费视频www| 人体艺术视频欧美日本| 男女啪啪激烈高潮av片| 最近最新中文字幕免费大全7| 91久久精品国产一区二区三区| 久久韩国三级中文字幕| 久久久久网色| 女的被弄到高潮叫床怎么办| 午夜免费男女啪啪视频观看| 三级男女做爰猛烈吃奶摸视频| 亚洲欧美精品专区久久| 免费av毛片视频| 国产视频首页在线观看| 成年人午夜在线观看视频 | 亚洲天堂国产精品一区在线| 日日摸夜夜添夜夜添av毛片| 国产精品一区二区性色av| 久久精品夜夜夜夜夜久久蜜豆| 日韩三级伦理在线观看| 噜噜噜噜噜久久久久久91| 我的女老师完整版在线观看| 欧美一区二区亚洲| 人妻少妇偷人精品九色| 久久久久久久大尺度免费视频| 免费黄频网站在线观看国产| 看十八女毛片水多多多| 一个人看的www免费观看视频| 国产免费一级a男人的天堂| 美女主播在线视频| 国产精品久久久久久久电影| 在线播放无遮挡| 欧美3d第一页| 欧美zozozo另类| 成人国产麻豆网| 一个人看的www免费观看视频| 伊人久久国产一区二区| 最近手机中文字幕大全| 亚洲精品乱久久久久久| 欧美三级亚洲精品| 国产免费又黄又爽又色| 精品一区二区三区视频在线| 特级一级黄色大片| 国产成人aa在线观看| 国产精品爽爽va在线观看网站| 2022亚洲国产成人精品| 内射极品少妇av片p| 成年av动漫网址| 欧美成人一区二区免费高清观看| 爱豆传媒免费全集在线观看| 麻豆久久精品国产亚洲av| 国产午夜精品一二区理论片| 国产亚洲一区二区精品| 久久97久久精品| www.色视频.com| 91久久精品电影网| 国产伦理片在线播放av一区| 亚洲精品中文字幕在线视频 | 国产黄片美女视频| 真实男女啪啪啪动态图| 老司机影院成人| av一本久久久久| 日本色播在线视频| 亚洲成色77777| 亚洲在线自拍视频| 国产美女午夜福利| 午夜亚洲福利在线播放| 亚洲欧洲国产日韩| 神马国产精品三级电影在线观看| 成人高潮视频无遮挡免费网站| 丝袜喷水一区| 非洲黑人性xxxx精品又粗又长| 日本爱情动作片www.在线观看| 亚洲国产av新网站| 国产免费福利视频在线观看| 91精品一卡2卡3卡4卡| 精品99又大又爽又粗少妇毛片| 成人无遮挡网站| 一边亲一边摸免费视频| 亚洲乱码一区二区免费版| 亚洲色图av天堂| 免费观看精品视频网站| 久久久久性生活片| 少妇猛男粗大的猛烈进出视频 | 国产精品国产三级专区第一集| 成年版毛片免费区| 日韩av在线免费看完整版不卡| av播播在线观看一区| 大话2 男鬼变身卡| 亚洲无线观看免费| 极品少妇高潮喷水抽搐| 亚洲欧美一区二区三区国产| 国精品久久久久久国模美| 一本—道久久a久久精品蜜桃钙片 精品乱码久久久久久99久播 | 亚洲国产最新在线播放| 2018国产大陆天天弄谢| av国产免费在线观看| 高清午夜精品一区二区三区| 国产亚洲精品av在线| 精品人妻一区二区三区麻豆| 亚洲成人久久爱视频| 亚洲欧美日韩卡通动漫| 日韩三级伦理在线观看| 国产午夜精品久久久久久一区二区三区| 波野结衣二区三区在线| 国产欧美日韩精品一区二区| 中文字幕免费在线视频6| 亚洲国产精品成人久久小说| 久久6这里有精品| 国产黄a三级三级三级人| 亚洲精品日韩av片在线观看| 久久精品久久久久久噜噜老黄| 亚洲精品乱久久久久久| 欧美日韩在线观看h| 日本免费在线观看一区| 天天躁夜夜躁狠狠久久av| 国产精品久久久久久精品电影| 日本免费a在线| 如何舔出高潮| 黄色配什么色好看| 3wmmmm亚洲av在线观看| 亚洲精品,欧美精品| 18禁在线播放成人免费| av女优亚洲男人天堂| 真实男女啪啪啪动态图| 国产爱豆传媒在线观看| 激情 狠狠 欧美| 久久精品久久久久久久性| 国产在线男女| 波野结衣二区三区在线| 日韩伦理黄色片| 毛片一级片免费看久久久久| 国产 一区 欧美 日韩| 全区人妻精品视频| 热99在线观看视频| 成人漫画全彩无遮挡| 国产免费视频播放在线视频 | 两个人的视频大全免费| 一边亲一边摸免费视频| 国产永久视频网站| 久久久国产一区二区| 青春草国产在线视频| kizo精华| 婷婷色麻豆天堂久久| 国产综合精华液| 国产在线一区二区三区精| 大陆偷拍与自拍| 亚洲一级一片aⅴ在线观看| 午夜老司机福利剧场| 亚洲精品一区蜜桃| 黄色配什么色好看| 男人舔女人下体高潮全视频| 午夜免费鲁丝| 亚洲伊人色综图| 免费高清在线观看日韩| 高清不卡的av网站| 99国产精品免费福利视频| 国产一区亚洲一区在线观看| av在线app专区| 久久99蜜桃精品久久| 女性被躁到高潮视频| www.av在线官网国产| 夫妻午夜视频| 国产亚洲av片在线观看秒播厂| 亚洲熟女精品中文字幕| 人妻系列 视频| 国产免费福利视频在线观看| 日韩免费高清中文字幕av| 欧美人与善性xxx| xxx大片免费视频| 中文字幕制服av| 丝袜人妻中文字幕| 2021少妇久久久久久久久久久| 精品酒店卫生间| av一本久久久久| 国产精品久久久av美女十八| 99久久中文字幕三级久久日本| 丝袜在线中文字幕| 国产精品无大码| 国产成人aa在线观看| 国产熟女欧美一区二区| 久久午夜福利片| 国产精品二区激情视频| 久久影院123| 欧美在线黄色| 国产精品亚洲av一区麻豆 | 在线观看一区二区三区激情| 美女国产视频在线观看| 天天躁夜夜躁狠狠躁躁| 久久99蜜桃精品久久| 人妻少妇偷人精品九色| 纵有疾风起免费观看全集完整版| 亚洲精品国产av成人精品| 欧美激情极品国产一区二区三区| 我的亚洲天堂| 免费在线观看完整版高清| 亚洲内射少妇av| 亚洲少妇的诱惑av| 如日韩欧美国产精品一区二区三区| 久久久久精品性色| 日本免费在线观看一区| 亚洲精品国产av成人精品| 精品酒店卫生间| 美女xxoo啪啪120秒动态图| 一区福利在线观看| 国产精品免费大片| 日韩精品免费视频一区二区三区| 啦啦啦在线观看免费高清www| 日日啪夜夜爽| 久久精品国产自在天天线| 一级黄片播放器| 欧美精品一区二区大全| 亚洲欧美中文字幕日韩二区| 久久精品国产亚洲av高清一级| av卡一久久| 国产成人午夜福利电影在线观看| 午夜久久久在线观看| 一个人免费看片子| 国产麻豆69| 亚洲国产成人一精品久久久| 波多野结衣一区麻豆| 人妻系列 视频| 伊人亚洲综合成人网| 不卡av一区二区三区| 色网站视频免费| 午夜福利视频精品| 亚洲经典国产精华液单| 卡戴珊不雅视频在线播放| 在线精品无人区一区二区三| 激情五月婷婷亚洲| 80岁老熟妇乱子伦牲交| 夫妻午夜视频| 另类亚洲欧美激情| 日韩av免费高清视频| tube8黄色片| 亚洲伊人久久精品综合| 中文字幕制服av| 男人操女人黄网站| 蜜桃国产av成人99| 99香蕉大伊视频| 久久精品久久久久久久性| 亚洲精品日韩在线中文字幕| 久久精品夜色国产| 国产精品国产三级专区第一集| 亚洲欧美一区二区三区国产| 亚洲一区二区三区欧美精品| 国产乱来视频区| 日韩av不卡免费在线播放| 在线 av 中文字幕| 国产成人精品久久二区二区91 | 在线观看www视频免费| 日韩av不卡免费在线播放| 9热在线视频观看99| 亚洲第一区二区三区不卡| 日韩免费高清中文字幕av| 黑丝袜美女国产一区| 午夜福利视频精品| 夫妻性生交免费视频一级片| 亚洲欧美一区二区三区久久| 中文字幕另类日韩欧美亚洲嫩草| 高清欧美精品videossex| 熟女电影av网| 亚洲四区av| 国产熟女欧美一区二区| 午夜福利影视在线免费观看| 制服丝袜香蕉在线| 亚洲精品日本国产第一区| 制服丝袜香蕉在线| 丝袜脚勾引网站| 边亲边吃奶的免费视频| 久久精品夜色国产| 一区二区三区精品91| 少妇 在线观看| 日韩av在线免费看完整版不卡| 777久久人妻少妇嫩草av网站| 在线看a的网站| 超碰成人久久| av网站免费在线观看视频| 亚洲欧美中文字幕日韩二区| 午夜日本视频在线| 亚洲五月色婷婷综合| 麻豆av在线久日| 亚洲欧美精品综合一区二区三区 | 欧美 亚洲 国产 日韩一| 国产熟女欧美一区二区| 亚洲精品日本国产第一区| 国产成人精品无人区| 在现免费观看毛片| 精品人妻偷拍中文字幕| 麻豆精品久久久久久蜜桃| 国产黄频视频在线观看| 婷婷成人精品国产| 亚洲色图 男人天堂 中文字幕| 婷婷成人精品国产| 亚洲av成人精品一二三区| 晚上一个人看的免费电影| 看非洲黑人一级黄片| 国产爽快片一区二区三区| 精品国产一区二区久久| 免费黄频网站在线观看国产| 极品人妻少妇av视频| 亚洲国产欧美日韩在线播放| 免费在线观看黄色视频的| 欧美成人午夜免费资源| 中国三级夫妇交换| 成年女人在线观看亚洲视频| 国产精品一区二区在线不卡| 在线精品无人区一区二区三| 国产成人a∨麻豆精品| 久久久国产精品麻豆| 99九九在线精品视频| 国产有黄有色有爽视频| 老司机影院成人| 男女下面插进去视频免费观看| 黄色视频在线播放观看不卡| 国产黄频视频在线观看| 国产精品久久久久久久久免| 国产一区二区在线观看av| 久久久国产欧美日韩av| 在线观看三级黄色| 考比视频在线观看| 日韩成人av中文字幕在线观看| 国产欧美亚洲国产| 黄色毛片三级朝国网站| 一级毛片黄色毛片免费观看视频| 高清欧美精品videossex| 亚洲一码二码三码区别大吗| 一级黄片播放器| 制服诱惑二区| av线在线观看网站| 美女国产视频在线观看| 男女无遮挡免费网站观看| 午夜激情久久久久久久| 黄片小视频在线播放| 久久99热这里只频精品6学生| 久久久久久久久久久免费av| 夜夜骑夜夜射夜夜干| 亚洲精品aⅴ在线观看| 少妇人妻精品综合一区二区| 免费观看a级毛片全部| 丝瓜视频免费看黄片| 18禁裸乳无遮挡动漫免费视频| 免费少妇av软件| 多毛熟女@视频| 宅男免费午夜| 久久韩国三级中文字幕| 国产免费福利视频在线观看| 国产日韩欧美视频二区| 又大又黄又爽视频免费| 午夜影院在线不卡| 国产精品一区二区在线不卡| av有码第一页| 欧美国产精品va在线观看不卡| 日韩一本色道免费dvd| 热re99久久精品国产66热6| 五月伊人婷婷丁香| 日本91视频免费播放| 国产精品免费大片| 久久精品国产亚洲av涩爱| 卡戴珊不雅视频在线播放| 你懂的网址亚洲精品在线观看| 丝袜脚勾引网站| 青草久久国产| 日本爱情动作片www.在线观看| www.av在线官网国产| 国语对白做爰xxxⅹ性视频网站| 在线看a的网站| 国产日韩欧美视频二区| 人人妻人人澡人人看| 欧美精品高潮呻吟av久久| 有码 亚洲区| 亚洲成av片中文字幕在线观看 | a级片在线免费高清观看视频| 精品少妇久久久久久888优播| 成人免费观看视频高清| 99久国产av精品国产电影| 国产黄色免费在线视频| 中文字幕亚洲精品专区| 亚洲国产欧美在线一区| 高清视频免费观看一区二区| 丰满少妇做爰视频| 1024视频免费在线观看| 久久久久久久大尺度免费视频| 国产精品亚洲av一区麻豆 | 超色免费av| 18禁观看日本| 啦啦啦在线观看免费高清www| 18禁国产床啪视频网站| 色播在线永久视频| 美女大奶头黄色视频| 国产一级毛片在线| 女人精品久久久久毛片| 热99久久久久精品小说推荐| 午夜福利视频精品| 中文天堂在线官网| 亚洲国产成人一精品久久久| 自拍欧美九色日韩亚洲蝌蚪91| 赤兔流量卡办理| 免费观看无遮挡的男女| 午夜福利网站1000一区二区三区| 美女视频免费永久观看网站| 欧美日韩一级在线毛片| 国产一级毛片在线| 免费女性裸体啪啪无遮挡网站| 欧美日韩一区二区视频在线观看视频在线| 天美传媒精品一区二区| 男女无遮挡免费网站观看| av片东京热男人的天堂| 秋霞在线观看毛片| 美女脱内裤让男人舔精品视频| 男男h啪啪无遮挡| 精品酒店卫生间| 91在线精品国自产拍蜜月| 久久精品aⅴ一区二区三区四区 | 天美传媒精品一区二区| 国产成人精品福利久久| 亚洲国产最新在线播放| 国产免费福利视频在线观看| 亚洲欧美精品综合一区二区三区 | 亚洲色图综合在线观看| 免费在线观看完整版高清| 啦啦啦中文免费视频观看日本| 国产精品不卡视频一区二区| 亚洲情色 制服丝袜| 黄色一级大片看看| 亚洲婷婷狠狠爱综合网| 欧美日韩一区二区视频在线观看视频在线| 国产精品熟女久久久久浪| 免费观看av网站的网址| 人人妻人人添人人爽欧美一区卜| 欧美+日韩+精品| 国产精品 欧美亚洲| 考比视频在线观看| 亚洲欧洲精品一区二区精品久久久 | 日产精品乱码卡一卡2卡三| 亚洲欧美日韩另类电影网站| 大陆偷拍与自拍| 大片免费播放器 马上看| 永久网站在线| 午夜激情av网站| 如日韩欧美国产精品一区二区三区| 九九爱精品视频在线观看| 99久久中文字幕三级久久日本| 97在线人人人人妻| 桃花免费在线播放| 精品国产一区二区三区四区第35| 制服人妻中文乱码| 亚洲人成网站在线观看播放| 黄片播放在线免费| 亚洲中文av在线| av国产久精品久网站免费入址| 亚洲av电影在线进入| 嫩草影院入口| 久久人妻熟女aⅴ| 热99国产精品久久久久久7| 国产成人精品无人区| 欧美日韩精品网址| 男女啪啪激烈高潮av片| 日韩一区二区三区影片| 欧美亚洲 丝袜 人妻 在线| 久久精品久久久久久久性| 日本午夜av视频| 999精品在线视频| 日本午夜av视频| 久久久久久久国产电影| 国产一区二区在线观看av| 丝袜喷水一区| 日韩av免费高清视频| 久久毛片免费看一区二区三区| 青春草亚洲视频在线观看| 久久精品国产亚洲av涩爱| 国产免费视频播放在线视频| 大陆偷拍与自拍| 久久久久网色| 成人国产麻豆网| 热99久久久久精品小说推荐| 美国免费a级毛片| 国产一区二区三区av在线| 精品国产一区二区三区四区第35| 卡戴珊不雅视频在线播放| 亚洲av国产av综合av卡| 涩涩av久久男人的天堂| 我的亚洲天堂| av网站免费在线观看视频| 亚洲av在线观看美女高潮| 久久热在线av| 欧美xxⅹ黑人| 天天躁夜夜躁狠狠躁躁| 中文字幕最新亚洲高清| 天堂中文最新版在线下载| 黄色 视频免费看| 国产亚洲午夜精品一区二区久久| 亚洲经典国产精华液单| 天堂俺去俺来也www色官网| 日韩中文字幕视频在线看片| 精品酒店卫生间| 韩国av在线不卡| 大码成人一级视频| 伊人亚洲综合成人网| 伊人久久国产一区二区| 综合色丁香网| 交换朋友夫妻互换小说| 中文字幕制服av| 久久ye,这里只有精品| 欧美xxⅹ黑人| 欧美成人午夜免费资源| 两个人看的免费小视频| 国产女主播在线喷水免费视频网站| 亚洲成人一二三区av| 午夜久久久在线观看| 国产av精品麻豆| 欧美成人精品欧美一级黄| av一本久久久久| 午夜福利在线观看免费完整高清在| 久久久久久久国产电影| 制服人妻中文乱码| 亚洲一码二码三码区别大吗| 亚洲av在线观看美女高潮| 我的亚洲天堂| 天天躁夜夜躁狠狠久久av| 久久久国产欧美日韩av| 久久 成人 亚洲| 国产精品偷伦视频观看了| 久久婷婷青草| 一本大道久久a久久精品|