Seventeen climate models from CMIP6 were examined to assess the expected behavior of seven atmospheric/ocean variables in the Caribbean Basin and the Seaflower Biosphere Reserve (SBR) during the twenty-first century, under two socioeconomic scenarios (SSP2-4.5 and SSP5-8.5). Additionally, an ensemble is made with the five models with the best oceanic resolution in the Caribbean Sea. Precipitation shows significant negative trends in most of the projected periods, while air and sea surface temperature, surface salinity and mean sterodynamic sea level (SDSL) have significant positive trends. Air temperature in SBR will probably increase by 2 °C compared to the preindustrial period after 2050 (SSP5-8.5) or 2060 (SSP2-4.5). The warming trend in the region could extend the hurricane season and/or increase hurricane frequency, affect ecosystems like coral reefs and mangroves, and intensify ocean stratification. For the same period, SDSL is expected to rise in SBR between ~24.2 and 39.9 cm. If all contributing factors are included, an increase of up to ~95 cm (SSP5-8.5) could be expected by the end of the twenty-first century. This sea level rise would modify the ecological balance and enhance flooding, affecting tourism and risking the disappearance of the low-elevation islands.
1 Introduction
Since the last century, the Earth’s climate has changed due to the increase of greenhouse gases in the atmosphere. Climate change has many negative impacts and is highly dependent on regional dynamics. Threats resulting from climate change are expected to grow, enhancing problems across the planet. Understanding the behavior of the main atmospheric and oceanic variables is of vital importance since climate is determined by different factors including the dynamics and composition of the atmosphere, the ocean, ice and snow cover, and land surface and its features as a coupled system (IPCC 2014). Coastal areas are particularly affected by climate change, since sea level rise will increase flooding, coastal erosion, storm surges, and saltwater intrusion, menacing human welfare and generating large economic losses (Tsyban et al. 1990; Nicholls and Lowe 2004; IPCC 2014). The Caribbean Sea is being affected by coastal erosion (Rangel-Buitrago et al. 2015), which is expected to increase as well as coastal flooding through the twenty-first century, as a consequence of extreme sea level increases driven by sea level rise (Torres and Tsimplis 2014).
To assess future climate behavior, Global General Circulation Models (Atmosphere–Ocean General Circulation Models or AOGCMs) have been created to simulate dynamic physical processes in the ocean, atmosphere, cryosphere, land, and Earth system interactions. These models are considered the most accurate tool to understand the response of the Earth system to different projected greenhouse gas emission rates into the atmosphere (IPCC 2014). Results from AOGCMs are developed by different institutions and research groups around the world and are evaluated in the Coupled Model Intercomparison Project (CMIP). In this chapter, the results of CMIP Phase 6 models (Meehl et al. 2014; Eyring et al. 2015) are reported.
Anzeige
The Intergovernmental Panel On Climate Change (IPCC) was created in 1998 by two United Nations agencies: the World Meteorological Organization (WMO) and the United Nations Environment Programme (UNEP) (Agrawala 1997). The main strategy of the IPCC is to enable political action by providing a scientific definition of the climate state and variation, including consequences and measures for adaptation and mitigation (Berg and Lidskog 2018). This is done through the preparation of Assessment Reports, which include the analysis of the latest CMIP model results. CMIP6 was created with the intention of answering the following questions: (1) How does the Earth system respond to different forcings? (2) What are the origins and consequences of systematic model biases? and (3) How to assess future climate changes given climate variability, predictability, and uncertainties in scenarios? (Meehl et al. 2014). CMIP6 is an improved version of the former CMIP5 because scientific groups were focused on developing intercomparison studies based on their own strategic goals, implying a diversity of Endorsed Model Intercomparison Projects (MIPs) to fill scientific gaps when compared to previous CMIP phases (Eyring et al. 2015).
The Caribbean Sea is in an intertropical region (Fig. 1). It is the largest marginal sea of the Atlantic Ocean with a surface extension of 2.52 × 106 km, almost twice as large as the Gulf of Mexico (Gallegos 1996). The Caribbean is connected on the northwest to the Gulf of Mexico through the Yucatan Channel and on the north and east to the Atlantic Ocean, separated by the Antilles. The Archipelago of San Andrés, Providencia, and Santa Catalina (hereafter, the archipelago) is composed of nine islands located in the Colombia Basin (Table 1). The climate is regulated by the meridional position of the Intertropical Convergence Zone (Andrade 2000) creating a windy-dry season (December–April) and rainy-warm season (August–October) (Etter et al. 1987; Angeles et al. 2010), with different regional ocean responses in the basin (Torres and Tsimplis 2012; Torres et al. 2022 in press). Eastward trade winds dominate the Caribbean, including a strong low-level jet around 15°N reaching speeds of ~12 ms−1 in the windy season (Andrade 2000).
A map highlights a closed region starting from the south of Puerto Rico, Hispaniola, and Cuba, and stretching along the Yucatan Basin, Caiman Basin, Gulf of Mosquitos, Gulf of Darien, Venezuela Basin, Canoe Trench, Granada Basin, Anegada Passage and ending to the south of Puerto Rico.
Fig. 1
Map of the Caribbean Sea. Bathymetric data from (GEBCO 2019), including the location of San Andrés (green star) and other islands (red dots) of the archipelago. Notation of islands is indicated in Table 1. The red polygon indicates Caribbean Sea limits used to obtain spatial averages of variables. The yellow polygon represents the Seaflower Biosphere Reserve (Coralina-Invemar 2012; Borrero-Pérez et al. 2019)
Table 1
Location of the islands in the archipelago (DIMAR 2019)
The Seaflower Biosphere Reserve (hereafter SBR) is located in the Colombia Basin (Fig. 1). It has a total area of 180,000 km2, of which 65,018 km2 is a protected marine area. Precipitation ranges between 0.8 mm day−1 in the dry season to 10.6 mm day−1 in the rainy season. Surface air temperature varies between 26.5 °C in the dry season and 28.1 °C in the rainy season. The trade winds average speed during the year is 4.5 m s−1, reducing in magnitude between September and October, and intensifying at the beginning of the year and in the month of July (Coralina-Invemar 2012). Sea surface temperature has a range between 26 °C in the dry season and 29.5 °C in the rainy season. Sea surface salinity typically presents values below 35.5 PSU, being affected by the freshwater contribution from the Orinoco and Magdalena rivers and rainfall (Coralina-Invemar 2012). Extreme events are dominated by hurricanes from June to November (Ortiz 2012) but are also affected by cold fronts from January to March (Ortiz et al. 2013).
Atmospheric and ocean behavior in the Caribbean Sea has been studied in the last decade, including the assessment of atmospheric temperature, pressure and wind (Montoya-Sánchez et al. 2018; Rodriguez-Vera et al. 2019; Hamed and Yunfang 2020; Bustos and Torres 2022) sea surface temperature and salinity (Ruiz et al. 2012; Beier et al. 2017) and mean sea level rise (Torres and Tsimplis 2013). Projected regional changes of these variables during the twenty-first century, using results from the CMIP5 under different radiative concentration scenarios, were recently assessed by Bustos and Torres (2021).
Anzeige
It is important to study projected regional and local atmospheric and ocean responses to climate change, to evaluate possible threats during the twenty-first century. Furthermore, it is necessary to understand the local risk, and its dependence on different Socioeconomic Scenarios Pathways (SSP), to develop accurate adaptation and mitigation plans to reduce impacts associated with climate change. These kinds of assessments are especially important in the Caribbean Sea and SBR, as small islands and the developing countries in the basin have constraints on adaptive capacity (Nicholls et al. 2007).
Therefore, the main objective of this chapter is to assess the projected behavior of atmospheric pressure, ambient temperature, wind, precipitation, ocean temperature, and salinity at the ocean surface, as well as mean sea level in the Caribbean Sea and SBR to the end of the twenty-first century. We used results from 17 CMIP6 models, under two Socioeconomic Pathways (SSP2-4.5 and SSP5-8.5) scenarios. The chapter is organized as follows. Section 2 describes the datasets and methods used. Sections 3 and 4 include results and discussions of atmospheric and oceanic projections, respectively. Section 5 presents a summary and conclusions.
2 Data and Methods
We used 17 CMIP6 models (hereinafter M17) used by Chen et al. (2020), but excluding UKESM1-0-LL, MCM-UA-1-0, CAMS-CSM1-0, CNRM-CM6-1 and replacing NESM3 and MIROC-ES2L with ACCESS-CM2 and ACCESS-CM1.5. We use the “historical” run before 2014 and projections between 2014 and 2100 to assess atmospheric pressure (SLP), air temperature (Ta), precipitation (Pr), and wind at sea level, as well as sea surface temperature (SST), sea surface salinity (SSS) and mean sea level for different periods between 1850 and 2100. However, reported trends are referenced to 2005 to facilitate comparison with the CMIP5 results reported by Bustos and Torres (2021). For the projections, two scenarios are used for projections. SSP2-4.5, which is the medium-forcing scenario, with a 4.5 W m−2 mean radiation, a peak of emissions in 2040, and then stabilization of the rate for the rest of the century. SSP5-8.5 is the high-forcing scenario, with an 8.5 W m−2 mean radiation, and a peak of emissions in 2080 (O’Neill et al. 2017; Riahi et al. 2017; Gidden et al. 2019).
The SSPs are based on 5 narratives describing different levels of socioeconomic development (Riahi et al. 2017). SSP1-sustainable development, SSP2-middle of the road development, SSP3-regional rivalry, SSP4-inequality, and SSP5-fossil-fuel-driven development. Full details of the SSPs are described by O’Neill et al. (2017). Data and model descriptions were accessed from the Program for Climate Model Diagnosis and Intercomparison (PCMDI) (http://cmip-pcmdi.llnl.gov/mips/cmip6/). Model data was downloaded from the Earth System Grid Federation (https://esgf-node.llnl.gov/projects/cmip6/).
Climate models used for sea level projections do not explicitly resolve mesoscale processes in the ocean (Penduff et al. 2010; Serazin et al. 2015). Only some effects of these processes, which depend on spatial resolution, are included in the models, generating errors in the circulation that affect the regional sea level projections. Therefore, adequate sea level projections in regions like the Caribbean Sea can only be obtained with high-resolution models that are able to capture mesoscale processes (van Westen et al. 2020). Consequently, we show results from two ensembles. One using all 17 CMIP6 models, and the other using only the five models with the best oceanic resolution (0.25°–1.00°) in the Caribbean Sea (ACCESS-CM1.5, ACCESS-CM2, GFDL-ESM4, MPI-ESM1-2-HR, and MRI-ESM2-0), hereafter M5-CAR. In this chapter, we emphasize results from the latter.
The model ensemble was obtained by interpolating and averaging each model’s results to a common grid (OM4 MOM6 for the five-model ensemble and T63 spectral for the 17-model ensemble) with a horizontal resolution of 1.875° × 1.25° (1.00° × 0.25°) for the atmospheric (oceanic) component to avoid errors introduced by extrapolation. Annual time series were obtained by averaging monthly data from the AOGCMs. To evaluate the Caribbean Sea and SBR regional behavior, we calculated the spatial mean and standard deviation from all model nodes in the study area (red and yellow polygons in Fig. 1, respectively). We found time series trends fitting a simple linear regression model, with an intercept using ordinary least squares. A significant error in each estimation was calculated at a 95% confidence level. Serial correlation was not used for the analysis, consequently, the 95% confidence intervals may be slightly narrower. To compute anomalies, the reference period (1976–2005) was subtracted from the end of the century (2071–2100) averaged values, with the aim of reducing the model’s interannual variability that could induce bias in the results.
We compared the five-model ensemble (hereafter M5) spatial results with satellite data linearly interpolated to the model nodes (Bustos and Torres 2021). Average satellite sea surface temperature for 1993–2005 from the COBE mission (Tokyo Climate Center 2020) and 2000–2005 sea surface salinity from the Aquarius mission (Jet Propulsion Laboratory 2020) were compared. Ocean mean circulation patterns in the Caribbean were computed using monthly files from OSTM/Jason-2 absolute dynamic topography anomalies for the 1993–2005 period (NOAA 2020a) and compared to the model’s mean sea level height above the geoid (SSH). Comparison showed that M5-CAR correctly reproduces the most important mesoscale features in the Caribbean (not shown).
Three main factors increase global mean sea level. First, thermal expansion (thermosteric change). Second, changes in the mass of seawater in the ocean (barystatic change). Third, global halosteric effects are far smaller than either thermosteric or barystatic changes (Griffies et al. 2016). Two variables are used in the CMIP6 AOGCMs to assess mean sea level. ZOSTOGA represents the global mean thermosteric sea level (GMTSL) that is affected by thermal expansion, representing roughly one-third to one-half of the observed global mean sea level rise in the 20th and early 21st centuries (Church et al. 2011; Gregory et al. 2013; Hanna et al. 2013). ZOS, defined as dynamic sea level (Griffies and Greatbatch 2012; Griffies et al. 2014), reflects the fluctuations due to ocean dynamics taking into account the redistribution of mass and changes in circulation (Yin 2012; Meyssignac et al. 2017). To assess total changes in local sea level, these two variables were added (Huang and Qiao 2015; Gregory et al. 2019) and reported in this chapter as sterodynamic sea level (SDSL) following van Westen et al. (2020).
AOGCMs models in CMIP6, similarly to CMIP5, do not include land-ice melting and other smaller contributions to sea level (IPCC 2014). Additionally, most CMIP6-based global climate models will have unreliable values for barystatic changes (Nowicki et al. 2016) as their dynamics are difficult to simulate (Dyurgerov and Meier 2004; Kaser et al. 2006; Henderson-Sellers and McGuffie 2012). Therefore, these changes can be estimated through the Ice Sheet Model Intercomparison Project (ISMIP6) (Nowicki et al. 2016). Finally, global halosteric effects in a CMIP simulation are associated with inaccurate estimates of ocean mass changes in these models, and represent a small fraction of the volume change that results from adding freshwater to the ocean (Wunsch et al. 2007).
Climate models often exhibit spurious trends that are unrelated to external forcing and internal climate variability, especially in oceanic variables. We calculated and removed this model drift from oceanic variables in the Caribbean using the model piControl simulation with a 500-yr length for all the models analyzed. We applied the full linear drift method recommended by Gupta et al. (2013).
3 Projections of Atmospheric Variables
3.1 Surface Air Temperature (Ta)
Spatially averaged Ta time series for the Caribbean and SBR (Fig. 2a) show large interannual variability under both SSPs projected scenarios. However, the time series are dominated by a trend after 1970. The uncertainty related to the models’ internal variability increases towards the end of the century and is larger under SSP5-8.5 than SSP2-4.5 (shaded area). The Ta spatial average for the M5-CAR in 1976–2005 for the CAR (SBR) is 26.86 ± 0.19 (27.06 ± 0.19) °C (Fig. 2a). All models indicated positive and statistically significant trends for Ta in all analyzed periods (Table 2). Ta warming trends in the M5-CAR produced by the historical model run are in good agreement with regional trends determined from in-situ data (Peterson et al. 2002; Stephenson et al. 2014; Jones et al. 2016) and CMIP5 models (Bustos 2020; Bustos and Torres 2021) for the different periods.
Two graphs labeled surface air temperature time series anomalies refereced to the 1976 to 2005 multi model mean value, and the precipitation time series anomalies referenced to the 1976 to 2005 multi model mean value. Both graphs plot curves for M 1 7 C A R, M 5 C A R, and M 5 S F B. The curves in the first graph depict almost similar pattern of remaining constant for some time and then increasing.
Fig. 2
Spatially averaged time series. For the Caribbean from 17 model ensemble (M17-CAR in green), 5 model ensemble (M5-CAR in blue) and SBR (M5-SBR in red). a Air Temperature (°C) and b Precipitation (mm day−1) anomalies referenced to the 1976–2005 averaged period. The 1850–2014 results are from historical experiments. The 2014–2100 projection is for SSP2-4.5 (solid line) and SSP5-8.5 (dashed line) scenarios. Trends values are presented in Table 2. The 5–95% range calculated across models is included as a shaded area for M5-CAR. Clear blue for historical and SSP2-4.5, dark blue for SSP5-8.5
×
The strongest Ta trends are for the SBR in the projected periods, with values of 2.88 ± 0.28 °C cy−1 (SSP2-4.5) and 3.52 ± 0.22 °C cy−1 (SSP5-8.5) for 2005–2050. For 2005–2100, the trends are 1.95 ± 0.11 °C cy−1 (SSP2-4.5) and 3.39 ± 0.08 °C cy−1 (SSP5-8.5). Under SSP2-4.5, the weaker trends of 2005–2100 relative to 2005–2050 are attributable to radiative emissions decline after 2040 in that scenario (O’Neill et al. 2017). However, because the trends are always positive, Ta is expected to continue to increase in the CAR and the SBR during the entire twenty-first century.
Table 2
Adjusted trends per century for spatially averaged atmospheric and oceanic variables
Experiment
Historical
SSP2-45
SSP5-85
Variable
Model/Per
1850–2005
1960–2005
2005–2050
2005–2100
2005–2050
2005–2100
Atmospheric variables
Ta
M17-CAR
0.44 ± 0.04*
1.37 ± 0.23*
2.94 ± 0.22*
2.21 ± 0.10*
3.67 ± 0.20*
4.33 ± 0.09*
M5-CAR
0.27 ± 0.05*
1.50 ± 0.27*
2.89 ± 0.26*
1.96 ± 0.12*
3.54 ± 0.20*
3.53 ± 0.08*
M5-SBR
0.25 ± 0.05*
1.44 ± 0.31*
2.88 ± 0.28*
1.95 ± 0.11*
3.52 ± 0.22*
3.39 ± 0.08*
SLP
M17-CAR
0.08 ± 0.03*
0.05 ± 0.21
−0.04 ± 0.22
0.03 ± 0.06
0.07 ± 0.25
0.28 ± 0.07*
M5-CAR
0.09 ± 0.07*
−0.14 ± 0.51
0.03 ± 0.43
−0.02 ± 0.14
−0.14 ± 0.40
0.45 ± 0.15*
M5-SBR
0.07 ± 0.06*
−0.06 ± 0.47
0.01 ± 0.38
−0.02 ± 0.12
−0.10 ± 0.35
0.47 ± 0.13*
Wind
M17-CAR
0.03 ± 0.02*
0.03 ± 0.17
0.11 ± 0.10*
0.06 ± 0.05*
0.10 ± 0.12
0.24 ± 0.04*
M5-CAR
0.01 ± 0.05
−0.14 ± 0.39
0.19 ± 0.28
0.11 ± 0.09*
−0.04 ± 0.30
0.18 ± 0.09*
M5-SBR
0.01 ± 0.09
−0.14 ± 0.66
0.18 ± 0.44
0.10 ± 0.15
−0.17 ± 0.50
0.23 ± 0.16*
Pr
M17-CAR
−0.17 ± 0.03*
−0.25 ± 0.26
−0.09 ± 0.20*
−0.15 ± 0.07*
−0.45 ± 0.28*
−0.92 ± 0.09*
M5-CAR
−0.19 ± 0.18*
−0.49 ± 0.48*
−0.21 ± 0.39
−0.14 ± 0.13*
−0.32 ± 0.21*
−1.16 ± 0.14*
M5-SBR
−0.36 ± 0.13*
−0.46 ± 0.78
−0.24 ± 0.76
−0.14 ± 0.12*
−0.73 ± 0.56*
−1.79 ± 0.23*
Oceanic variables
SST
M17-CAR
0.26 ± 0.04*
0.92 ± 0.25*
2.27 ± 0.21*
1.89 ± 0.07*
2.60 ± 0.28*
3.29 ± 0.27*
M5-CAR
0.08 ± 0.06*
1.47 ± 0.53*
2.81 ± 0.37*
2.36 ± 0.14*
3.35 ± 0.38*
3.43 ± 0.12*
M5-SBR
0.05 ± 0.04*
1.39 ± 0.57*
2.79 ± 0.42*
2.39 ± 0.16*
3.33 ± 0.45*
3.37 ± 0.42*
SSS
M17-CAR
0.03 ± 0.01*
0.13 ± 0.07*
0.48 ± 0.06*
0.54 ± 0.03*
0.69 ± 0.11*
1.00 ± 0.04*
M5-CAR
0.09 ± 0.03*
0.19 ± 0.18*
0.95 ± 0.22*
0.36 ± 0.07*
1.14 ± 0.19*
1.45 ± 0.06*
M5-SBR
0.08 ± 0.02*
0.11 ± 0.09*
0.99 ± 0.13*
0.34 ± 0.04*
1.17 ± 0.12*
1.41 ± 0.07*
SSH
M17-CAR
0.03 ± 0.01*
0.13 ± 0.07*
0.48 ± 0.06*
0.54 ± 0.03*
0.64 ± 0.11*
1.00 ± 0.04*
M5-CAR
−2.41 ± 0.49*
−2.11 ± 2.90
−2.62 ± 2.28*
−2.42 ± 0.79*
0.63 ± 3.13
−2.44 ± 0.98*
M5-SBR
−2.49 ± 0.67*
0.19 ± 0.42
−1.09 ± 3.09
−1.87 ± 1.18*
0.85 ± 2.95
−0.28 ± 1.18
GMTSL
M17-CAR
2.56 ± 0.14*
5.94 ± 0.68*
20.41 ± 0.96*
22.64 ± 0.40*
20.87 ± 2.33*
32.19 ± 1.37*
M5-CAR
3.49 ± 0.12*
6.05 ± 0.76*
25.73 ± 0.54*
30.64 ± 0.42*
28.58 ± 0.96*
42.89 ± 1.49*
SDSL
M17-CAR
1.15 ± 0.19*
4.72 ± 0.13*
24.49 ± 1.30*
24.78 ± 0.53*
22.10 ± 1.35*
22.21 ± 0.48*
M5-CAR
1.08 ± 0.50*
3.93 ± 3.36*
23.11 ± 2.24*
28.23 ± 0.92*
29.21 ± 3.35*
40.46 ± 1.89*
M5-SBR
1.00 ± 0.68*
6.24 ± 4.47*
24.65 ± 2.96*
28.77 ± 1.23*
32.66 ± 4.00*
42.62 ± 1.85*
For the Caribbean Sea using 17 Multi-model Ensemble (M17-CAR); 5 Multi-model Ensemble (M5-CAR); and for the Seaflower Biosphere with the 5 multi-model ensemble (M5-SBR). Two CMIP6 Socioeconomic Pathways (SSP) are assessed. Significant trends (95% confidence) are indicated with (*). Air temperature (Ta), atmospheric pressure (SLP) wind speed, and Precipitation (Pr) trends indicated in °C cy−1, hPa cy−1, ms−1 cy−1, and mm day−1 cy−1, respectively. Sea surface temperature (SST), sea surface salinity (SSS), sea surface above geoid (SSH), global averaged steric sea level (GMTSL), and local mean sterodynamic sea level (SDSL) trends indicated in °C cy−1, PSU cy−1, and cm cy−1, respectively
In addition to the time series, we also assessed Ta spatial behavior for the 2071–2100 averaged period from the M5-CAR. The Ta spatial average for 2071–2100 is estimated to be 28.79 ± 0.21 (29.52 ± 0.51) °C, under SSP2-4.5 (SSP5-8.5) scenarios, respectively. Differences from the reference period 1976–2005 represent a nearly homogeneous increase of 1.93 (2.66) °C under SSP2-4.5 (SSP5-8.5) scenarios (Fig. 3b, c).
Three columns of five images each depict the historical from 1976 to 2005, S S P 2 4.5 anomalies 2071 to 2100 reference 1976 to 2005, and S S P S 8.5, reference 1976 to 2005.
Fig. 3
Air temperature (°C, first row), wind (m s−1, second row), precipitation (mm day−1, third row), sea surface temperature (°C, fourth row) and sea surface salinity (PSU, fifth row) spatial behavior in the Caribbean Sea from the 5-model ensemble (M5-CAR). For the 1976–2005 averaged period (first column) after the “historical” run. Surface air temperature, precipitation, sea surface temperature and sea surface salinity anomalies for 2071–2100 relative to 1976–2005 under SSP2-4.5 are shown in the second column, except wind which shows the 2071–2050 mean behavior. Third column as second column, but under the SSP5-8.5 scenario
×
We also studied Ta and SST seasonal changes, because they are important for sea level sub-regional behavior and extremes in the Caribbean (Torres and Tsimplis 2012, 2013). Regardless of SSP scenarios used, an increase in Ta is expected in all months at the end of the century (2071–2100), maintaining seasonality with maximum values between June–November (rainy season) (Fig. 4). The annual range increases from 2.05 °C in 1976–2005 to 2.09 °C (2.12 °C) in 2071–2100 for SSP2-4.5 (SSP5-8.5) scenarios respectively.
Three multiline graphs labeled seasonal behavior of Ta and S S T seaflower biosphere M M M C M I P 6, seasonal behavior of P r seaflower biosphere M M M C M I P 6, and seasonal behavior of wind seaflower biosphere M M M C M P 6. The second and third graphs depict highly fluctuating trends.
Fig. 4
Seasonal behavior for 1976–2005 and 2071–2100 averaged periods in the SBR from the 5-model ensemble (M5-SBR) under SSP2-4.5 and SSP5-8.5 projected scenarios. a Surface Ambient temperature (Ta) and Sea surface temperature (SST). b Precipitation (Pr), and c wind speed
×
The preindustrial (1860–1900) Ta mean in the CAR (SBR) has a spatial mean of 26.47 ± 0.09 (26.70 ± 0.10) °C for the M5 (Fig. 2a). In line with the Paris Agreement (IPCC 2014; United Nations 2015), a 2 °C increase in Ta relative to the preindustrial period has been defined as the limit at which the planet could experience risks and impacts associated with climate change in the attempt to meet the sustainable development goals (United Nations 2015). Under SSP2-4.5, Ta in the CAR (SBR) would be near the limit established by the Paris Agreement 28.47 (28.70) °C, about 2059 (2060), whereas under SSP5-8.5 this limit would be reached after 2046 (2050) (Fig. 2a).
Taylor et al. (2018) studied air temperature and precipitation from 42 CMIP5 models of the Caribbean for the period 1861–2100 under RCP-4.5. Most models indicated that air temperature would attain an increase of 2 °C between 2033 and 2062 relative to the preindustrial period (1861–1900). Similarly, Bustos and Torres (2021) found 2060 (2040) as the attainment dates for the 2 °C limit under RCP4.5 (RCP8.5) scenarios respectively. This indicates that the Ta increase according to the M5-CAR is close to the upper limit (highest trends) compared to the results from Taylor et al. (2018).
3.2 Sea Level Pressure (SLP)
SLP in the Caribbean Sea has modest trends <0.1 hPa cy−1 that are significant only for the 1850–2005 historical period (Table 2), indicating a steady behavior during the previous century. In addition, significant trends are only observed for CAR (SBR) for the 2005–2100 projected period under the SSP5-8.5 scenario, with values of 0.45 ± 0.15 (0.47 ± 0.13) hPa cy−1 (Table 2). Time series are not shown as they are dominated by climate models’ interannual variability. Besides, spatial differences in time (not shown) are basin-wide coherent.
3.3 Surface Wind
Surface wind time series are dominated by large interannual variability (not shown), as was also found for sea level pressure (SLP). Significant coherent trends in wind speed are found only for 2005–2100 under SSP5-8.5, with a value of 0.23 ± 0.16 ms−1 cy−1 in M5-SBR (Table 2). The regional wind pattern shows an increase in the area of wind speed >8 ms−1, without significant changes in direction for the 2071–2100 period under both SSPs scenarios, relative to 1976–2005 (Fig. 3e, f). In the 1976–2005 period, the average wind speed for the M5-CAR (M5-SBR) is 6.05 ± 0.17 (5.99 ± 0.29) ms−1, with values >8 ms−1, indicating the position of the Caribbean low-level jet (Fig. 3d) and resembling the typical wind behavior in the basin. The largest increase in wind speed is for 2071–2100 referenced to the 1976–2005 period, under the SSP5-8.5 in M5-SBR (spatial mean of 0.32 ms−1). Similar results were found from the CMIP5 models (Bustos and Torres 2021).
We executed a seasonal analysis of long-term variation of the surface wind, since this variable dominates seasonal changes in the regional ocean circulation (Torres and Tsimplis 2012). Thus, we investigated wind time series from the dry-windy season (December–January–February) and rainy-warm season (September–October–November) using both SSPs scenarios (not shown). For the 1976–2005 period, wind speed in M5-SBR was stronger in the dry season (7.06 ± 0.53 ms−1) when compared to the rainy season (4.71 ± 1.12 ms−1). For the same area in the 2071–2100 period under the SSP5-8.5, the mean wind speed in the dry season is 6.56 ± 0.47 ms−1, and 5.43 ± 0.90 ms−1 in the rainy season. Thus, M5-SBR models indicate that wind speed differences between the dry and rainy seasons will decrease without changing their seasonality (Fig. 4c). Besides, note that the maximum wind speed month in M5-SBR is projected to change from December (1975–2005) to July (2071–2100—SSP5-8.5). These results are in agreement with Bustos and Torres (2022), who also show different spatial patterns in wind intensification based on CMIP6 models, especially seen in the Caribbean low-level jet with seasonal different responses.
Costoya et al. (2019) assessed wind energy projections in the Caribbean for the twenty-first century using downscaling techniques in seven AOGCM models from CMIP5. They found that the maximum annual wind speed increase in the CAR would be ~0.4 ms−1 by 2100 under RCP8.5, referenced to the 2005 value. Thus, their increase in regional wind is about twice the result from the M5-CAR in the same period with the SSP5-8.5 scenario (0.18 ms−1 in Table 2).
Evaluating future changes in surface wind is important, because various authors have shown the dominance of wind in driving the CAR ocean circulation (Brenes and Trejos 1994; Torres and Tsimplis 2012; Montoya-Sánchez et al. 2018). We found that significant positive trends in M5-SBR for 2005–2100 (SSP5-8.5) do not result from a homogeneous wind speed increase in all months. Although wind direction and seasonal behavior seem not to have significant changes (Figs. 3d–f and 4c), seasonal variations in wind speed could significantly affect regional circulation. Therefore, future changes in the Caribbean surface wind should include a seasonal assessment.
3.4 Precipitation (Pr)
The M5-CAR and M5-SBR show significant negative trends for most of the periods and projected scenarios analyzed (Table 2). The largest trends are in the 2005–2100 period for SSP5-8.5 (−1.79 ± 0.23 mm day−1 cy−1 in M5-SBR). Regardless of the significant coherent trends, model projections of this variable have large differences (shading area in Fig. 2b); therefore, results must be treated with caution. Pr is expected to decrease in the CAR and the SBR during the twenty-first century, as trends are negative. It appears that the future radiative emission rate significantly affects Pr, as was seen for Ta, due to significantly larger trends in SSP5-8.5 when compared to SSP2-4.5 (Table 2). These results are in good agreement with the reduction of Pr in the Caribbean founded by Almazroui et al. (2021), also using CMIP6 models.
The Pr spatial average for the M5-CAR (M5-SBR) in 1976–2005 is 2.76 ± 0.22 (3.82 ± 0.34) mm day−1 (Fig. 3g). Similar values were found by Lee and Wang (2014) in a shorter period (1980–2005) using CMIP5 models. The Pr M5-SBR spatial average for 2071–2100 is expected to be 3.78 ± 0.34 (2.84 ± 0.51) mm day−1, under SSP2-4.5 (SSP5-8.5) scenarios, respectively. Differences from the reference period 1976–2005, represent a decrease in SBR of −0.04 (−0.98) mm day−1 under SSP2-4.5 (SSP5-8.5) scenarios, respectively (Fig. 3h, i).
We also studied Pr seasonal changes for the SBR. All scenarios show a bimodal distribution with maximum values from June to November (rainy season). Under SSP2-4.5, the Pr annual pattern in 2071–2100 does not show a large difference (Fig. 4) when compared to the 1976–2005 behavior. The largest difference is a reduction in Pr during the rainy season for the 2071–2100 SSP5-8.5 projection. The annual range decreases from 6.64 mm day−1 in 1976–2005 to 6.21 (4.19) mm day−1 in 2071–2100 for SSP2-4.5 (SSP5-8.5) scenarios respectively. These projections under the SSP5-8.5 scenario coincide with the results of Karmalkar et al. (2013) where a reduction of Pr in the Caribbean islands was identified under the SRES A2 scenario using 15 GCMs from CMIP3 and a Regional Climate Model. These are important results, as inhabited islands in the SBR depend on the rainy season to collect fresh water for their yearly consumption. The shown reduction of precipitation in the rainy season (>35% under SSP5-8.5) should be tracked carefully, in order to define convenient mitigation plans if data starts to confirm such projections.
4 Projections of Oceanic Variables
4.1 Sea Surface Temperature (SST)
Significant and positive trends are found for SST in all experiments and projected periods for the CAR and SBR (Table 2). SST behavior was similar to Ta as expected, due to heat fluxes between the atmosphere and ocean. We found the smallest trends for 1850–2005, with a noticeable increase over 1960–2005 with values of 1.47 ± 0.53 (1.39 ± 0.57) °C cy−1 for the M5-CAR (M5-SBR), respectively. However, in the historical run, significant SST trends are within the bounds of the interannual variability (Fig. 5a).
Two graphs labeled sea surface temperature time series anomalies referenced to the 1976 to 2005 multimodel mean value, and the S D S L time series anomalies referenced to the 1976 to 2005 multimodel mean value. Both graphs plot curves for M 1 7 C A R, M 5 C A R, and M 5 S F B. Both curves depict an almost similar pattern of remaining constant for some time and then increasing.
Fig. 5
Spatially averaged time series for the Caribbean Sea from the 17 model ensemble (M17-CAR in green), 5 models ensemble (M5-CAR in blue) and SBR (M5-SBR in red): a Sea surface temperature (°C); b Sterodynamic sea level (cm) anomalies referenced to the 1976–2005 period. The 1850–2014 results are from historical experiments. The 2014–2100 projection is for SSP2-4.5 (solid line) and SSP5-8.5 (dashed line) scenarios. Trend values are presented in Table 2. The 5–95% range calculated across models is included as a shaded area for M5-CAR. Clear blue for historical and SSP2-4.5, dark blue for SSP5-8.5
×
Warming trends throughout the Caribbean for the second half of the twentieth century (Table 2) are consistent with those evidenced by Peterson et al. (2002) and Villegas et al. (2021). Similarly, Antuña-Marrero et al. (2016), using information reconstructed from the International Comprehensive Ocean–Atmosphere Dataset, found a trend for 1972–2005 of 1.41 ± 0.67 °C cy−1. Deser et al. (2010) found regional trends in SST with a range of 0.4–1.6 °C cy−1 for the period 1900–2008, using observed data and a model-based reconstruction. Trends from the historical run are smaller compared to the results of Bustos and Torres (2021) using CMIP5 models.
Significant positive trends dominated SST in all the projected periods analyzed for the Caribbean, regardless of the model or SSP scenario used (Table 2). For 2005–2050, the trend for the SBR is 2.79 ± 0.42 (3.33 ± 0.45) °C cy−1 under SSP2-4.5 (SSP5-8.5) respectively. For 2005−2100, the trends are between 2.39 ± 0.16 (3.37 ± 0.42) °C cy−1 under SSP2-4.5 (SSP5-8.5). For the projected periods, most SST trends are smaller than Ta trends. According to the IPCC (2014), SST is expected to increase globally between 0.70 and 2.40 °C by 2100 under RCP4.5. Therefore, under a similar projection scenario (SSP2-4.5) trends found for the CAR and SBR are at the maximum among these values.
Because of this positive trend, by 2071–2100 SST in the M5-CAR (M5-SBR) is expected to be 29.40 ± 0.19 (29.26 ± 0.18) °C or 30.56 ± 0.39 (30.35 ± 0.38)°C, under SSP2-4.5 or SSP5-8.5, respectively. That represents a temperature increase in the Caribbean of 2.04–3.20 °C and for SBR 2.09–3.18 °C (Fig. 3k, l), over the value for the 1976–2005 averaged period (27.36 °C in M5-CAR and 27.17 °C in M5-SBR) (Fig. 3j), depending on the radiative scenario used.
SST warming trends in the CAR produced by the SSP5-8.5 scenario run are in good agreement with regional trends determined from the 26 CMIP5 models ensemble presented by Alexander et al. (2018) with values between 2.5 and 3.5 °C cy−1 for the Gulf of Mexico and the Caribbean in the 1976–2099 period. Similar results were obtained by Bustos and Torres (2021).
SST is warmest between September and November and coldest in February (Fig. 4a), resembling the seasonal behavior in the Caribbean (Torres and Tsimplis 2012). Regardless of the projected scenario, all months in the 2071–2100 period are expected to be warmer than the warmest month in the 1976–2005 period. Several studies have shown diverse factors that can affect the formation of hurricanes (Goldenberg et al. 2001; Wang and Lee 2007), of which SST increase is probably the most important. Because warmer sea surface temperatures enhance tropical cyclone formation in the Atlantic (Grinsted et al. 2013), the hurricane season in the Caribbean runs from June to November (NOAA 2020b), corresponding to the warmer months. Therefore, M5-SBR projections for the end of the century show that all year round there would be sufficient heat in the ocean to permit hurricane formation, which in the future could extend the hurricane season in the basin, among other impacts.
SST increase could also affect the biosphere in the local upwelling areas. Taylor et al. (2013), using monthly observations from the CARIACO Ocean Time-Series in the southern Caribbean Sea between 1996 and 2010, identified that the SST increase would intensify ocean stratification, reducing the delivery of upwelled nutrients to surface waters, generating an ecological state of change in the planktonic system.
Bustos and Torres (2021) showed that the SST increase in the Caribbean expected from the CMIP5-ACCESS1.0 model could have severe consequences for hurricane frequency, its season length, and in coral bleaching by the end of the twenty-first century. This chapter builds on these results using a 5-model ensemble from the CMIP6. Results are consistent, although the CMIP6 shows that most of the SST trends for the projected periods are larger compared to the CMIP5-ACCESS model results.
4.2 Sea Surface Salinity (SSS)
For all experiments using the M5-CAR and M5-SBR, positive and significant SSS trends are seen in all periods (Table 2). The smallest trends are for 1850–2005 with values of 0.09 ± 0.03 (0.08 ± 0.02) PSU cy−1 for CAR (SBR), respectively. Nonetheless, in the historical run, significant SSS trends do not extend beyond the interannual variability (not shown). The SSS trends for 2005–2100 in the M5-CAR are 0.36 ± 0.07 (1.45 ± 0.06) PSU cy−1 and in SBR 0.34 ± 0.04 (1.41 ± 0.07) PSU cy−1 under SSP2-4.5 (SSP5-8.5) (Table 2). Therefore, SSS trends in the Caribbean Sea are sensitive to the projected radiative scenarios.
The M5-CAR mean salinity field for the 1976–2005 period is 35.51 PSU (Fig. 3m), while the expected SSS mean for the 2071–2100 period is 35.96 ± 0.10 (36.87 ± 0.22) PSU and for M5-SBR 36.19 ± 0.13 (37.07 ± 0.25) PSU under SSP2-4.5 (SSP5-8.5), respectively. The difference of ~0.91 (0.89) PSU for CAR (SBR) between the two SSP scenarios by the end of the twenty-first century indicates the sensitivity of this variable to the radiative forcing scenario used. Due to the positive trends, for 2071–2100 under SSP2-4.5, SSS increases uniformly in the basin between 0.3 and 0.6 PSU (Fig. 3n), while under SSP5-8.5, SSS anomalies are between 1.0 and 1.5 PSU (Fig. 3o), with larger values toward the Cayman Basin and the Yucatan Peninsula.
The SSS increase shown in the present study is similar to the results presented by Bustos and Torres (2021) for the CMIP5-ACCESS1.0 model. In their study, they suggested that positive SSS trends were related to SST increase, due to evaporation enhancement. In this chapter, the 5-model ensemble from CMIP6 results also shows an increase in SST which could lead to an increase in evaporation. Besides, we also include an assessment of precipitation. We show negative trends in precipitation, which would also increase SSS due to rainfall decrease and have an effect reducing local river freshwater fluxes.
4.3 Sea Level
To assess the future sea level behavior in the Caribbean Sea and SBR, we present two analyses, each one including different periods under two SSP scenarios. First, we determined SSH, GMTSL and sterodynamic sea level (SDSL) spatially averaged trends (Table 2, Fig. 5b), as well as SDSL trends spatial behavior (Fig. 6) to study regional patterns. Second, expected sterodynamic sea-level rise is presented for the SBR and the archipelago (Fig. 7, Table 1).
Six color-coded maps a to f labeled historical S D S L trends 1850 to 2005, S S P 2 4.5 b, S D S L trends 2005 to 2050, c, S D S L trends 2005 to 2100, d, S D S L trends 1960 to 2005, e, S D S L trends 2005 to 2050, and f, S D S L trends 2005 to 2100.
Fig. 6
Sterodynamic sea level trends (SDSL) in cm cy−1 for the Caribbean Sea from 5 models ensemble. Periods a 1850–2005 and d 1960–2005 from historical run. Period 2005–2050 using b SSP2-4.5 and e SSP5-8.5. Period 2005–2100 using c SSP2-4.5 and f SSP5-8.5. Isolines every 2 cm cy−1. Note different trend ranges in panels
Six color-coded maps a to f labeled historical S D S L rise 1976 to 2005, b, S D S L rise 2021 to 2050, c, S D S L rise 2071 to 2100, d, S D S L rise 1976 to 2005, e, S D S L rise 2021 to 2050, and f, S D S L rise 2071 to 2100.
Fig. 7
Sterodynamic sea level (SDSL) rise in (cm) for the archipelago from the 5-model ensemble. Periods 1976–2005 referenced to a 1850–1879 and d 1931–1960 from the historical run. Period 2021–2050 using b SSP2-4.5 and e SSP5-8.5 projection scenarios. Period 2071–2100 using c SSP2-4.5 and f SSP5-8.5 projection scenarios. Panels b, c, e, and f referenced to 1976–2005 averaged period. Isolines every 1 cm. Note different height ranges in each panel. Brown Polygons represent the Seaflower Biosphere Reserve (Coralina-Invemar 2012; Borrero-Pérez et al. 2019)
×
×
Trends in SSH from the historical run in M5-CAR and SBR are only significant and negative in the 1850–2005 period (Table 2). On the contrary, GMTSL trends are significant and positive for the Caribbean in all periods of the historical run, making all SDSL trends in this period significant and positive. SDSL trends in 1960–2005 are more than double the 1850–2005 trends.
All significant SSH trends for the CMIP6 projected periods in M5-CAR and SBR are negative (Table 2), contrary to CMIP5-ACCESS results (Bustos and Torres 2021). However, all GMTSL and SDSL trends in CMIP6 projected periods are positive and significant, regardless of their radiative scenario. Therefore, positive GMTSL trends prevail over negative SSH trends projected in the Caribbean. In the 2005–2100 period for the Caribbean (SBR), SDSL trends under SSP5-8.5 are 40.46 ± 1.89 (42.62 ± 1.85) cm cy−1 respectively (Table 2). The largest contribution from GMTSL to SDSL trends in the Caribbean and SBR, compared to SSH contribution, indicates that regional sea level rise will be largely due to global ocean temperature increase rather than regional effects (Figs. 6 and 7).
SDSL trends in the Caribbean dominate the interannual variability with a clear exponential behavior under SSP5-8.5 through the end of the century (Fig. 5b). An acceleration of 1.6 ± 0.1 mm yr−1 cy−1 was reported for Cristobal sea level in the southern Caribbean, for the period 1908–2009 (Torres and Tsimplis 2013). In the same period, the M5-CAR SDSL time series shows an acceleration of 0.56 ± 0.16 mm yr−1 cy−1. Additionally, for the 2005–2100 period an acceleration of 1.62 ± 0.17 (2.64 ± 0.21) mm yr−1 cy−1 is expected under the SSP2-4.5 (SSP5-8.5) scenario respectively. The SDSL acceleration observed in Fig. 5b, is mainly due to GMTSL acceleration with values of 1.27 ± 0.14 (2.21 ± 0.19) mm yr−1 cy−1 for the 2005–2100 period under SSP2-4.5 (SSP5-8.5) scenario, respectively. Note that this acceleration is larger in the 5-model ensemble than in the 17-model ensemble (Fig. 5b). As the main difference between the two model ensembles is the spatial resolution in the intertropical ocean, this indicates that projected GMTSL variations can depend on climate models’ resolution in addition to the climate sensitivity of the models used in an ensemble.
SDSL trends are not spatially homogeneous in the study area. For 1850–2005, M5-CAR has larger trends (>1.3 cm cy−1) toward the north-eastern Caribbean (Fig. 6a). On the contrary, larger trends (>5 cm cy−1) for 1960–2005 appear toward the Yucatan Peninsula and in the SBR (4–5 cm cy−1) (Fig. 6d). For the projected 2005–2050 period, trends range between 15–27 cm cy−1 under SSP2-4.5 (Fig. 6b) and 17.5–29 cm cy−1 under SSP5-8.5 (Fig. 6e). In both cases, trends are larger toward the Cayman Sea and smaller toward the eastern Caribbean. For 2005–2100, trends are 27–30 cm cy−1 under SSP2-4.5 (Fig. 6c). However, the strongest trends are for the SSP5-8.5 scenario, with a range of 41–44 cm cy−1 (Fig. 6f) with larger values in the Venezuela and Granada basins. Modeled SDSL trends in the Caribbean show strong spatial variability and changes with time, which coincides with observed regional sea level behavior assessed from observed data (Torres and Tsimplis 2013).
We also assessed SDSL rise in the SBR, which results from the reported trends using the M5 ensemble (Fig. 7). In a 45-year period of the projection run (2021–2050 referenced to 1976–2005), the SDSL is expected to rise in the islands of the archipelago >10.1 (>11.3) cm under SSP2-4.5 (SSP5-8.5) (Fig. 7b, e). Additionally, in a 95-year period (2071–2100 referenced to 1976–2005), the SDSL is projected to rise under the same scenarios >24.2 (>38.8) cm (Fig. 7c, f). The largest SDSL rise for this period is projected for Bajo Nuevo under SSP245 (24.69 cm) and for Roncador under SSP585 (39.89 cm), among the islands in the SBR (Table 1). However, the sea level is projected to rise in all the islands with differences <2 cm (Fig. 7). According to projections from CMIP6 of Jevrejeva et al. (2020), the GMTSL increase by 2081–2100 is expected to be between 12.8–23.6 cm (18.6–34.6 cm) relative to 1995–2014 under SSP2-4.5 (SSP5-8.5) scenarios, respectively. Reported results of SDSL rise from the SBR with the 5-model ensemble are just above the upper limit of the projections obtained in the mentioned study. Although SDSL rise includes the SSH’s smaller contribution (Table 2) and the periods assessed are slightly different, results are in good agreement.
M5-CAR using CMIP6 models shows an underestimation of SDSL trends when compared to previous studies using CMIP5 results (Palanisamy et al. 2012; Gupta et al. 2013;). In the archipelago, SDSL rise from the CMIP5-ACCESS1.0 model was projected to be >14.5 (>16) cm for 2021–2050 and >32.6 (>45.5) cm for 2071–2100 under RCP4.5 (RCP8.5), respectively, referenced to the 1976–2005 period (Bustos and Torres 2021). Therefore, the 5-model ensemble from CMIP6 projects a smaller SDSL increase in the study area when compared to just one model from CMIP5, but with similar spatial patterns under each scenario. This difference is mainly due to differences in the GMTSL trend, assessed by Bustos and Torres (2021) using the ZOSGA variable, which represents the total change in global mean sea level due to thermosteric changes, water flux input and salinity influences on density. In CMIP6, the ZOSGA variable is not available, therefore in this study, we use ZOSTOGA, which only accounts for the contribution due to thermal structure changes, which partially explains the smaller GMTSL trends reported in Table 2.
Conversely, Jevrejeva et al. (2020) reported an increase and larger variance for the projected GMTSL rise for the twenty-first century, from the comparison of a 15-CMIP6 model ensemble with 20-CMIP5 models (they used ZOSTOGA in their comparison). They discuss probable causes for these differences between the CMIP, mainly associated with a new generation of climate models as well as a new set of scenarios of concentrations, emissions, and land use. Besides, they assess explanations for large uncertainties in the simulation of GMTSL. Other factors associated with the models’ physics such as parametrizations, circulation, and mixing schemes can also influence the SDSL projections (Swart et al. 2019).
5 Summary and Final Remarks
Seventeen CMIP6 models were used to evaluate atmospheric and oceanographic trends over the twenty-first century in the Caribbean Sea (CAR) and the Seaflower Biosphere Reserve (SBR), under two different radiative emission scenarios SSP2-4.5 and SSP5-8.5. We selected five models with the highest oceanic resolution in the Caribbean Sea to obtain the best projections for the study area due to their better ability to resolve mesoscale processes (van Westen et al. 2020).
Surface air temperature (Ta) shows significant positive trends in M17-CAR, M5-CAR, and M5-SBR ensembles (Fig. 2a). A 2 °C increase compared to the preindustrial period, defined in the Paris Agreement as a limit for greatly increased global risks, would be expected between 2046 and 2059 (2050–2060) for the CAR (SBR), depending on the radiative scenario used. On the contrary, precipitation (Pr) shows significant negative trends in most of the periods for M5-CAR and M5-SBR, therefore a reduction of Pr in the study area is expected by the end of the century (Fig. 2b). Besides, considerable Pr reduction was identified under the SSP5-8.5 scenario in the rainy season for 2071–2100 (Fig. 4b), which can reduce freshwater availability in the SBR inhabited islands. Therefore, it is important to continue monitoring these variables, to develop opportune adaptation plans to minimize freshwater supply problems for the population of the archipelago.
Sea level pressure and surface wind show a large interannual variability with the largest trends for the 2005–2100 period under the SSP5-8.5 scenario in the CAR and SBR, but without considerable spatial changes by the end of the century (Fig. 3e, f). Besides, wind speed seasonality in the SBR is not projected to change, however, an intensification from June to October can be expected (Fig. 4c). Besides, projected changes in surface wind speed will probably have different spatial responses in the study area (Bustos and Torres 2022). As wind is a major driver of seasonal sea level changes in the region (Torres and Tsimplis 2012), such wind changes might force ocean dynamics to sub-annual variations.
Sea surface temperature (SST) is expected to increase by the end of the twenty-first century compared to the baseline period 1976–2005, between 2.03–3.20 °C in the Caribbean and 2.09–3.18 °C in the SBR under SSP2-4.5 and SSP5-8.5 scenarios respectively (Fig. 3k, f). SST trends are all positive and significant with higher values for the projected periods (Fig. 5a). Furthermore, SST seasonality will not change in the region. However, all months at the end of the century (2071–2100) will be warmer than the warmest month in the baseline period (1976–2005), regardless of the radiative scenario used (Fig. 4a). This temperature increase in the ocean could intensify coral bleaching events, extend the hurricane season, and/or increase hurricane frequency in the basin by the end of the century (Saunders and Lea 2008; Eakin et al. 2009; Bender et al. 2010; Lough et al. 2018). However, other important factors modulate the genesis and development of tropical cyclones such as wave perturbation in trade winds, upper troposphere divergence, low vertical wind shear, and air temperature profiles in the atmosphere (IPCC 2014). Therefore, only the expected SST trends shown in this study are not enough to conclude about future cyclone-related changes in the Caribbean Sea.
Similarly, sea surface salinity (SSS) trends are positive and significant in all periods, with larger values under the SSP5-8.5 scenario, coinciding with the expected reduction in precipitation and the Ta and SST increase, which will probably enhance evaporation in the region. Besides, SST and SSS positive trends will compensate for surface density changes. Trends indicate that SST warming will overcome SSS increase, therefore, surface water will become lighter, which could affect local upwelling areas due to ocean stratification intensification (Torres et al. 2022 in press), reducing the delivery of upwelled nutrients affecting planktonic systems (Taylor et al. 2013).
In the islands of the SBR, depending on the SSP scenario, SDSL is expected to rise between ~24.2 and 39.9 cm for 2071–2100 compared to the baseline period 1976–2005, due to trends up to 42.62 ± 1.85 cm yr−1 in the SBR (SSP5-8.5). An SDSL acceleration of 0.56 ± 0.16 mm yr−1 cy−1 is observed for the 1908–2009 period in the Caribbean from the 5-model ensemble, which coincides with previous sea level acceleration reported from a long tide-gauge time series in the region (Torres and Tsimplis 2013). Besides, an acceleration of 1.62 ± 0.17 (2.64 ± 0.21) mm yr−1 cy−1 is found under the SSP2-4.5 (SSP5-8.5) scenario for the 2005–2100 period. This acceleration is due to GMTSL behavior, which protrudes from the model ensemble with the best oceanic resolution in the region (Fig. 5b). Therefore, the assessment of GMTSL trends might be sensitive to the model’s ocean resolution in the tropics, as this is an important area for heat storage and redistribution.
We found that GMTSL dominates SDSL trends. SDSL trends from CMIP6 M5-CAR are lower in all cases when compared to the ACCESS-CMIP5 model (Bustos and Torres 2021), however, slightly different variables were used in both studies to assess GMTSL (ZOSTOGA in the former and ZOSGA in the later as discussed in Sect. 4.3). On the contrary, Jevrejeva et al. (2020) reported GMTSL (ZOSTOGA) trends increase in CMIP6 models when compared to CMIP5, and assessed probable causes for these differences. Accuracy in projections of future sea level change depends on the ability of climate models to reproduce the components of sea level rise over the twenty-first century and simulate future changes across a range of emission scenarios. Differences between CMIP5 and CMIP6 thermosteric sea level projections differ mainly due to a new generation of climate models (Eyring et al. 2015) and a new set of scenarios of concentrations, emissions, and land use (O’Neill et al. 2016). However, large uncertainties in the Caribbean and global sea level projections remain due to the climate models’ limited ability to reproduce future thermosteric sea level changes.
Using a 9–model ensemble from CMIP6, Sung et al. (2021) identified for the 2081–2100 period, compared to 1986–2005, a future global sea level mean (range) increase of 0.28 m (0.17–0.38 m) and 0.65 m (0.52–0.78 m) under SSP1-2.6 and SSP5-8.5 scenarios respectively. They included the contribution from changes in ocean density (mainly from thermal expansion), land ice melting from glaciers and ice sheets, groundwater, and Global Isostatic Adjustment (GIA). From these, ocean-related processes and glacier melting are major contributors to SLR. On a global scale, the former contributes 54% (42%) and the latter 32% (52%) for SSP1-2.6 (SSP5-8.5). Therefore, the SDSL rise reported in this chapter, which only accounts for ocean thermosteric changes, will probably contribute nearly half of the total sea level rise expected for the Caribbean Sea by the end of the century. If we use this global % contribution (42%) in a coarse calculation, total sea level rise by the end of the century would be up to ~95 cm for SSP5-8.5 in the islands of the archipelago.
Such mean sea level rise will interact with the tide, sea level seasonal cycle and meteorological extremes (Torres and Tsimplis 2014), enhancing flooding and erosion, risking the complete submersion of the low elevation islands of the archipelago (height <2 m—Table 1). Besides, erosion will reduce beaches’ extension, which can affect tourism, the main income for some Caribbean island countries (IPCC 2014). Sea level rise will also affect the biosphere. For example, the ecological balance around mangroves will change (Bacon 1994), affecting a large number of species around this area of high biological productivity in the SBR (Urrego et al. 2009). These ecosystems provide feeding and breeding grounds for birds, reptiles, fish, and invertebrates, including many endemic vulnerable threatened and endangered species (Prato and Newball 2016). For all these reasons, it is important to continue strengthening ocean and atmosphere monitoring in the archipelago, including projections from higher spatial resolution climate models and the use of downscaling techniques to improve the dynamics of some air-ocean mesoscale processes. In parallel to the former, there is a need to enhance nations’ adaptive capacity, in order to build and improve regional and local mitigation and adaptation plans, and to reduce climate change impacts in the Seaflower Biosphere Reserve by the end of the twenty-first century.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
