Warming Projections of Eastern Mediterranean in CMIP6 Simulations According to SSP2-4.5 and SSP5-8.5 Scenarios ^†

Abstract

This study investigates the future temperature changes in the climate-vulnerable region of the Eastern Mediterranean. The results from seventeen (17) CMIP6 (6th Phase of Coupled Model Intercomparison Project) model simulations are analyzed. The analysis is focused on the SSP2-4.5 and SSP5-8.5 scenarios. The ERA5 reanalysis is used as a reference dataset to investigate the performance of CMIP6 simulations to accurately reproduce the mean temperature in the Eastern Mediterranean region. The results show that CMIP6 model simulations vary regarding their efficiency for capturing the mean temperature. Future projections show that significant warming is shown during the last period of the 21st century. The continental Balkan and Turkish regions are recognized as the most affected areas regarding future warming. The increase in temperature spatially ranges (in local scale) from 1.5 °C to 4.5 °C for the SSP2-4.5 scenario and from 3.0 °C to 8.0 °C for the SSP5-8.5 scenario. Finally, the seasonal analysis indicates that summer (JJA) shows the maximum temperature increase compared with the other seasons.

1. Introduction

The Mediterranean region is located in the mid-latitudes and characterized by temperate climate conditions. Additionally, it is at a climatic crossroad where the humid and mild European climate as well as the hot and arid African climate affect the Mediterranean climate [1]. This sensitive area shows warming 20% faster than the global average [2,3,4]. Furthermore, during the last decades the warming has seemed to accelerate [3,5]. Future global climate model (GCM) projections (CMIP5 and CMIP6 simulations) show that drying and warming rates will be increased over the Mediterranean region in comparison to other areas [1]. The maximum changes are shown during the summer period. Additionally, the projections show that the future mean temperature, summer extremes and the frequency of heatwaves will increase [6]. The warming in combination with the decline of precipitation over the Mediterranean area will be some of the dominant factors that increase the climate and socioeconomic risks for this area [1].

GCMs are state-of-the-art tools for climate science. The sixth phase of the Coupled Model Intercomparison Project (CMIP6) involves the latest generation of GCMs that runs in numerous institutes and research centres globally. The main questions that CMIP6 is expected to give an answer to are (a) the impact of climate forcing on the Earth’s climate, (b) the better understanding of model simulation biases, (c) the alternation of future climate (climate variability, predictability, etc.) [7]. The fifth generation reanalysis dataset of ECMWF (ERA5) combines observations with modelling techniques to provide numerous meteorological and climatological parameters during the period from 1940 to the present. The ERA5 dataset contributes significantly to climate and environmental investigations [8].

This work studies (a) the ability of seventeen (17) CMIP6 model simulations to reproduce the temporal and distributional features of mean temperature averaged over the Eastern Mediterranean using data from ERA5 as the reference, (b) whether the simulations provide warming projections for various model simulations of CMIP6 over one high-interest area regarding its climate vulnerability. In general, the findings from CMIP6 project show stronger warming compared with the previous CMIP phase [1]. This study investigates the future changes in the Eastern Mediterranean (EMed) temperature for two SSPs (SSP2-4.5 and SSP5-8.5; period from 2015 to 2100) with reference to a historical period from 1970 to 2005 (considered as the basis period). Finally, the calculation of the temperature differences between the future and basis period over annual, monthly and seasonal scales indicates the season that shows the maximum temperature changes.

This work is organized as follows: In Section 2 (“Materials and Methods”) the data and methodology that are used in this study are presented. In Section 3 (“Results”) the findings of the analyses and a short discussion are presented. Finally the main results are concluded in Section 4 “Conclusions”.

2. Materials and Methods

2.1. Data and Region of Study

The monthly mean near-surface temperature (tas) of seventeen (17) new-generation model simulations available in the sixth phase of the Coupled Model Intercomparison Project (CMIP6) which underpins the sixth Assessment Report of the IPCC [9] are studied here.

Table 1. List of CMIP6 model simulations that are used in this study.

Model	Institute (Country)	Resolution (lon/lat)	Ensemble
ACCESS-CM2	Australian Community Climate and Earth System Simulator Climate Model Version 2 (Australia)	192 × 144	r1i1p1f1
ACCESS-ESM1-5	Australian Community Climate and Earth System Simulator Earth System Model Version 1.5	192 × 145	r1i1p1f1, r2i1p1f1, r3i1p1f1
AWI-CM-1-1-MR	Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research	384 × 192	r1i1p1f1
CAMS-CSM1-0	Climate Academy of Meteorological Sciences—Climate Simulation Model	320 × 160	r1i1p1f1
CanESM5	Canadian Centre for Climate Modelling and Analysis, Environment and Climate Change Canada	128 × 64	r1i1p1f1
CMCC-CM2-SR5	Fondazione Centro Euro-Mediterraneo sui Cambiamenti Climatici, Italy	288 × 192	r1i1p1f1
CNRM-CM6-1-HR	Centre National de Recherches Meteorologiques, Centre Europeen de Recherche et de Formation Avancee en Calcul Scientifique, France	256 × 128	r1i1p1f2
GFDL-ESM4	National Oceanic and Atmospheric Administration, Geophysical Fluid Dynamics Laboratory, USA	360 × 180	r1i1p1f1
GISS-E2-1-G	Goddard Institute for Space Studies, USA	144 × 90	r1i1p1f2
HadGEM3-GC31-LL	Met Office Hadley Centre, UK	92 × 144	r1i1p1f3
INM-CM5-0	Institute for Numerical Mathematics, Russian Academy of Science, Russia	180 × 120	r1i1p1f1
IPSL-CM6A-LR	Institut Pierre Simon Laplace, France	144 × 143	r2i1p1f1
KACE-1-0-G	National Institute of Meteorological Sciences/Korea Meteorological Administration, Climate Research Division, Republic of Korea	192 × 144	r1i1p1f1
MIROC6	Japan Agency for Marine-Earth Science and Technology, The University of Tokyo, Japan	256 × 128	r1i1p1f1
MIROC-ES2L	Japan Agency for Marine-Earth Science and Technology, The University of Tokyo, Japan	128 × 64	r1i1p1f1
MPI-ESM1-2-LR	Max Planck Institute for Meteorology, Germany	192 × 96	r1i1p1f1
MRI-ESM2-0	Meteorological Research Institute, Japan	128 × 64	r1i1p1f1

provides elements regarding the simulations that are used in this study. In particular, the temperature results from the CMIP6 model simulations, under two Shared Socioeconomic Pathways scenarios (SSPs), are studied during the future period. SSPs are scenarios that consider alternative socioeconomic and demographic development. They are used by the climate change research community to investigate possible future climate change [10]. Generally, SSP2-4.5 is the “middle of the road” (moderate) scenario where the radiative forcing rises to 4.5 W/m² till 2070 and then drops. The SSP5-8.5 scenario is the extreme scenario where the radiative forcing rises to 8.5 W/m² till the end of the 21st century. All the SSPs involve the socioeconomic axes of urbanization, education, population and economic development as drivers for the evolution of future climate [11].

ERA5 monthly mean temperature data over the region of the Eastern Mediterranean, which are freely available in Copernicus Climate Change Service Climate Data Store (https://cds.climate.copernicus.eu/ accessed on 16 September 2024), are retrieved and analyzed. These data are used as the reference dataset. The analysis focuses on a geographical window that includes the region of the southeastern Mediterranean (15° to 40° E, 30° to 45° N, hereafter EMed; Figure 1).

2.2. The Performance of CMIP6 Models to Simulate the Annual Mean Temperature Averaged over EMed

The Kling–Gupta Combined Statistical Index (KGE) is calculated for each one of Global Climate Models (GCMs) using the temperature data from ERA5 as the reference dataset for the historical period from 1970 to 2005 (basis period) [12]. In general, KGE is considered as an appropriate index to investigate the accuracy of GCMs regarding their efficiency to reproduce climatic variables such as temperature [13], reflecting both the temporal and distributional features. For the calculation of KGE the following Equation (1) is used:

$$KGE=1-\sqrt{{\left(r-1\right)}^2+{\left({\displaystyle \raisebox{1ex}{${\sigma }_s$}\!\left/ \!\raisebox{-1ex}{${\sigma }_0$}\right.}-1\right)}^2+{\left({\displaystyle \raisebox{1ex}{${\mu }_s$}\!\left/ \!\raisebox{-1ex}{${\mu }_0$}\right.}-1\right)}^2}$$

(1)

where r is the Pearson’s correlation coefficient between CMIP6 model simulations and ERA5, ${\sigma }_s$ and ${\sigma }_0$ are the standard deviations of CMIP6 model simulations and ERA5 and the ${\mu }_s$ and ${\mu }_0$ are the averages of CMIP6 model simulations and ERA5. The KGE values range from −∞ to 1. The values that are closer to 1 correspond to a better model performance [14].

2.3. Temperature Projections and Trend Analysis over EMed: The Last Period of 21st Century

The time series of annual mean temperature averaged over the EMed regions is calculated in order to investigate the temperature evolution over EMed according to the SSP2-4.5 and SSP5-8.5 scenarios. Furthermore, the differences between the last period of the 21st century and the historical basis period are calculated for each of the model simulations. For the calculation of the statistical significance of the differences, the two-tailed t-test at 99% is used. For the future period from 2070 to 2100, where the maximum temperature changes are identified, the temperature trend for each of the model simulations is calculated. For the statistical significance of the calculated temperature trends (°C/decade), the Mann–Kendall statistical significance test is revealed at the 99% level. In order to illustrate the results of these analyses, related maps are constructed. Finally, a boxplot of seasonal (DJF, MAM, JJA and SON) and annual (Ann.) as well as monthly mean temperature anomalies between the last period of the 21st century (for SSP2-4.5 and SSP5-8.5) and the basis historical period is constructed in order to investigate the temperature changes over the seasons.

3. Results

3.1. Accuracy of CMIP6 Simulations to Capture and Reproduce the Annual Mean EMed Temperature

The calculation of the temperature bias of the CMIP6 annual mean EMed temperature relative to ERA5 data ranges from −3.9 °C to 2.7 °C. Nine out of seventeen (9/17) simulations show a temperature bias lower than $1.5\ast {\sigma }_0$ (Figure 2). Additionally, the calculation of the KGE index shows that the ACCESS-ESM1-5, CMCC-CM2-SR5, INM-CM5-0, IPSL-CM6A-LR and MIROC6 present moderate KGE values (0.4 ≤ KGE < 0.47). These GCMs provide a better performance than the other CMIP6 simulations in capturing the temporal and distributional features of annual mean EMed temperature. In particular, the MIROC-ES2L and GFDL-ESM4 show KGE values between 0.3 and 0.4 (0.3 ≤ KGE < 0.4). The other GCMs present a positive but lower than 0.3 KGE (KGE < 0.3). Note that CanESM5, GISS-E2-1-G and HadGEM3-GC31-LL show negative KGE values (KGE < 0). The analysis by Zareian et al. [13] has shown that for the western Turkey area the KGE (as a measure for CMIP6 spatial temperature performance) ranges from −0.2 to 0.4 and the majority of model simulations have shown that the KGE values fluctuates between 0.2 and 0.4 [13].

3.2. Future Temperature Changes over EMed

The analyses of CMIP6 model simulations show that all GCMs present an increase in the averaged, annual mean EMed temperature until the end of the 21st century. Figure 3 shows the anomalies of mean EMed temperature (with respect to basis period; 1970–2005) for the historical and future periods both for the SSP2-4.5 and SSP5-8.5 scenarios. The maximum changes are shown according to SSP5-8.5 scenario, where the EMed warming ranges from 4 °C to 8 °C at the end of 21st century. For the SSP2-4.5 scenario, EMed warming ranges from 2 °C to 4 °C. The majority of GCMs show that after the middle 21st century the CMIP6 simulations show an increased temperature trend in the SSP8.5 scenario compared with the SSP2-4.5 scenario (Figure 3). Tebaldi et al. [15] have shown that the global mean temperature increases at the end of 21st century by about 4.5 °C (according to the SSP5-8.5 scenario) and about 2 °C (according to the SSP2-4.5 scenario) with respect to the period from 1995 to 2014. The synergy between the decline of aerosols (changing the radiative forcing) and the reduction in near-surface soil moisture possibly explains this warming trend [3].

The CMIP6 simulations show that the maximum temperature changes are shown during the last period of the 21st century. Focusing on the period from 2070 to 2100, the differences in annual mean EMed temperature (with respect to the basis period) show that the SSP5-8.5 scenario presents increased temperatures compared with the SSP2-4.5 scenario, showing similar spatial patterns but a (clear) higher temperature for SSP5-8.5 (Figure 4). In particular, for the SSP2-4.5 scenario, CMIP6 model simulations show an increase in EMed temperature during the future period that ranges locally from 1.5 °C to 4.5 °C (Figure 4A). The analysis shows that the CMIP6 model mean presents an increase for the EMed temperature related to the basis period of about 3.4 °C (95% CI: [2.89, 3.83]). The maximum changes are presented over the continental areas. For the SSP5-8.5 scenario the GCMs show an increased EMed temperature which varies from 3 °C to 8 °C. In particular, the EMed temperature for the CMIP6 model mean shows a warming with respect to the basis period of about 5.1 °C (95% CI: [4.27, 5.69]). The maximum changes (for both SSPs) are presented over the continental area (mainly over Turkey and north Greece) (Figure 4). The findings show that the EMed warming varies among the different GCMs (Figure 4). Previous studies have shown that the southeastern Mediterranean (and north latitudes) shows the maximum warming over Europe of ~3 °C to 4 °C [6,16,17,18]. The differences among the CMIP6 for projecting the long-term warming are related to the climate sensitivity of each model simulation. This feature in combination with different future climate scenarios (SSPs) reduces the uncertainties and improves the ability of GCMs to capture the possible future climate changes [19].

Figure 5 shows the temperature trend (°C/decade) for the last period of the 21st century both for the SSP2-4.5 and SSP5-8.5 scenarios. In 13 out of 17 (13/17) model simulations the maximum warming trend is presented over continental areas of Turkey and the Balkan Peninsula indicating that these areas are more prone to climate change and extreme conditions compared with the other EMed regions. For the SSP2-4.5 scenario the warming ranges from 0.0 to 0.4 °C/decade (Figure 5A). The SSP5-8.5 scenario shows increased trends (compared with the SSP2-4.5 scenario) that range from 0.4 to 1.5 °C/decade (Figure 5B).

In order to study the seasonal temperature differences in the future period 2070 to 2100 with respect to the basis period the monthly mean temperature anomalies averaged over EMed and the boxplot of seasonal temperature anomalies are calculated (Figure 6 and Figure 7). Both future scenarios show that for all seasons the temperature is increased by about ~3.0 and ~5.0 °C for the SSP2-4.5 and SSP5-8.5 scenarios, respectively. The summer season (JJA) shows the maximum temperature increase compared with the other seasons ranging from ~3.0 °C to ~5.0 °C for the SSP2-4.5 scenario and from ~5.0 °C to ~7.0 °C for the SSP5-8.5 scenario. The analysis of CMIP5 model simulations under the RCP8.5 scenario has shown that the temperature in the Mediterranean is projected to increase by about 20% more than the global mean warming, with the maximum warming to be shown during the summer (JJA) season (about 50% more than the global mean). The increased warming over the Mediterranean and the reduced precipitation due to climate change categorize this region as one of the most responsive regions to climate stress and future resilience [20].

4. Conclusions

This work focuses on investigating the future temperature changes over the EMed using CMIP6 model simulations according to the SSP2-4.5 and SSP5-8.5 scenarios. The results show that GCMs present different skills for capturing and reproducing the EMed temperature compared with ERA5. The ACCESS-ESM1-5, CMCC-CM2-SR5, INM-CM5-0, IPSL-CM6A-LR and MIROC6 show a better performance compared with other GCMs for capturing the temporal and distributional features of mean EMed temperature (0.4 ≤ KGE < 0.47). All model simulations show a warming of ~4 °C to 8 °C for the SSP5-8.5 scenario and ~2 °C to 4 °C for the SSP2-4.5 scenario till the end of the 21st century. The maximum changes are presented during the last period of the 21st century and they are identified over continental areas of the Balkan Peninsula and Turkey. Focusing on the end of the 21st century, the warming trend for the EMed fluctuates from 0.4 to 1.5 °C/decade for the SSP5-8.5 scenario and from 0.0 to 0.4 °C/decade for the SSP2-4.5 scenario. Finally, the seasonal temperature projections show that summer months show the maximum warming.