This study investigates the Mediterranean region, including parts of Southern Europe, North Africa, and the Middle East. The study region is defined as the area between 20.00 and 50.00° N latitude and 15.00° W and 50.00° E longitude. This spatial extent is selected to capture the full Mediterranean basin and its surrounding land–sea system, which is relevant for analyzing large-scale atmospheric variability and tropical edge dynamics. The Mediterranean basin is characterized by substantial environmental and geographical gradients, complex morphology with mountain chains, and significant land-sea contrasts. Its location in a transition zone between mid-latitude and subtropical atmospheric circulation systems contributes to its complex climate patterns (Lange, 2020). The Mediterranean hosts one of the world's richest biodiversity reserves with numerous endemic species. However, rising temperatures, habitat loss, and changing phenological patterns pose significant threats (Coll et al., 2010; Lionello et al., 2012).

We employ multiple datasets spanning January 1980 to December 2022. Monthly averaged temperature data on pressure levels from ERA5, the fifth-generation ECMWF reanalysis, are utilized to determine lapse rate tropopause (LRT) parameters (height and temperature). ERA5 provides data at 0.25° × 0.25° horizontal resolution (approximately 31 km × 31 km) with 37 vertical pressure levels ranging from 1000 to 1 hPa (Hersbach et al., 2019a). Monthly averaged surface temperature (Ts) and total precipitation data are also obtained from ERA5 at 0.25° × 0.25° resolution (Hersbach et al., 2019b) to explore the impacts of tropical belt expansion on temperature and precipitation patterns over the Mediterranean region. Additionally, monthly precipitation and potential evapotranspiration data at 0.50° × 0.50° resolution from the Climatic Research Unit Time-Series (Harris et al., 2020) are used to compute the Standardized Precipitation Evapotranspiration Index (SPEI) as a meteorological drought indicator, following the methodology of Vicente-Serrano et al. (2010) and Beguería et al. (2013).

We apply the LRT definition to determine tropopause parameters (height and temperature) using ERA5 temperature data on pressure levels. Temperature data are first vertically spline interpolated to a regular 100 m height resolution. Following the World Meteorological Organization (WMO) definition, the thermal LRT is identified as the lowest level at which the lapse rate decreases to 2 °C km⁻¹ or less, provided the average lapse rate between this level and all higher levels within 2 km does not exceed 2 °C km⁻¹ (WMO, 1957). Long-term trends of tropopause parameters over the Mediterranean region are examined using linear regression analysis at all grid points. The TEL, which indicates tropical belt width, is estimated using two tropopause height metrics. The first method to estimate the TEL is based on tropopause drop (TPD), and it defines the TEL as the latitude at which LRT height (LRT-H) drops 1.50 km below the tropical average (15.00° S–15.00° N) (Birner, 2010; Davis and Rosenlof, 2012). While the second method to determine the TEL relied on the tropopause gradient (TPG), it signifies the TEL as the latitude of the maximum LRT-H meridional gradient (Davis and Rosenlof, 2012; Davis and Birner, 2017).

We assess the relationships between TEL and key climatic variables using a suite of statistical techniques, including correlation analysis, linear regression, and singular value decomposition (SVD). The analysis focuses on how tropical expansion relates to tropopause characteristics, Ts (considered both as a potential driver and a response to tropical belt shifts), tropospheric temperature (Trp) (averaged from the surface to the LRT level), precipitation, and SPEI as an indicator of meteorological drought. For both the correlation and SVD analyses, the TEL dataset at each monthly time step consists of one latitude value at each longitude and therefore represents a one-dimensional, longitude-dependent field of TEL position. In contrast, the climatic variables are represented as two-dimensional latitude–longitude fields. To facilitate spatially consistent correlation and coupled-pattern analyses, as well as direct visualization of the resulting relationships, the longitude-dependent TEL values were replicated along the latitude dimension to construct a two-dimensional representation matching the geographical grid of the corresponding climatic field. This transformation provides a common spatial framework for examining the relationships between TEL variability and the climatic fields. The SVD analysis was applied to identify the dominant coupled modes of variability between the TEL field and the climatic fields. This method extracts pairs of spatial patterns (one for each field) and their respective temporal variations, each pair explaining a fraction of the shared covariance. To compute the SVD, we construct the temporal cross-covariance matrix between the two space-time fields, which must cover the same period of time (Bjornsson and Venegas, 1997; Darrag et al., 2024). Prior to the SVD analysis, all variables are detrended and deseasonalized to remove long-term trends and annual cycles. This preprocessing step ensures that the extracted SVD modes reflect genuine covariability between TEL and the climatic parameters, rather than artifacts arising from shared seasonal structure or long-term evolution.

Our analysis reveals that LRT-H has increased by approximately 80.58 ± 3.07 m per decade since January 1980 (Fig. 1a). Previous literature also shows an LRT-H increasing trend, but the values presented in this paper are greater. The larger values obtained in this study can be attributed to the fact that the Mediterranean region has experienced warming at rates exceeding the global average. Schmidt et al. (2008) indicated a global upward trend in LRT-H of 39 to 66 m per decade for the period from May 2001 to December 2007. This tropopause height increase is attributed to global warming driven by increasing atmospheric GHG concentrations (Meng et al., 2021). These findings in LRT-H also aligned with previous studies documenting global tropopause height increases from global navigation satellite systems radio occultation (GNSS-RO) data (Darrag et al., 2022), radiosonde observations (Seidel and Randel, 2006), and reanalysis datasets (Santer et al., 2004). LRT temperature (LRT-T) exhibits a downward trend of -0.12 ± 0.02 K per decade (Fig. 1b), showing a strong negative correlation of -0.86 with LRT-H. These results of LRT parameter characteristics are in agreement with those reported in earlier studies (Sausen and Santer, 2003; Schmidt et al., 2004; Seidel and Randel, 2006; Pisoft et al., 2021).

Tropopause parameters exhibit significant anomalies driven by atmospheric dynamics, climate change, and natural variability. The anomaly time series for tropopause parameters was created by subtracting the long-term monthly climatological means from the original observations, covering the period from 1980 to 2022. This deseasonalization procedure removes the mean annual cycle and highlights climate variability and long-term changes. LRT-H displays a maximum positive anomaly of approximately 0.41 km occurring in August 2020 and the greatest negative anomaly of approximately -0.46 km in June 1992. After November 1997, its anomaly time series shows mostly positive values, which means that the Mediterranean region has been getting warmer (Fig. 1c). These results are consistent with Seidel and Randel (2006), who reported tropopause monthly anomalies with standard deviations (SDs) ranging between 0.20 and 0.60 km for LRT-H. In the case of LRT-T, it exhibits a maximum positive anomaly of approximately 2.01 K in January 1993 and a greatest negative anomaly of approximately -2.08 K in January 1998. It shows a dominant negative anomaly after June 1993, reflecting a predominant warming pattern over the study area (Fig. 1d).

Linear regression of LRT-H shows a dominant upward trend evident across almost the entire study area (Fig. 1e). The eastern Mediterranean region at approximately 30.00° N exhibits the maximum LRT-H increasing trends, reaching 266.50 m per decade, attributed to the stronger influence of the South Asian monsoon system in this area (Tyrlis and Lelieveld, 2013). The zonal mean of the LRT-H regression map (right panel of Fig. 1e) indicates increasing trends at all latitudes, with the latitudinal band between 30.00 and 33.75° N showing maximum upward trends. For LRT-T (Fig. 1f), the majority of the study area displays decreasing trends with statistically significant values at p<0.05. The greatest LRT-T downward trend occurs in the eastern Mediterranean region around 30° N, with maximum decreasing trends of approximately -0.75 K per decade. The zonal mean of the LRT-T regression map (right panel of Fig. 1f) shows decreasing trends at all latitudes except the northern zone of the Mediterranean region between 43.50 and 50.00° N, with the latitudinal band from 20.00 to 32.25° N exhibiting maximum downward trends.

Figure 2 depicts TEL derived from ERA5 data based on both TPD and TPG methods. The TPD method shows an increase in tropical belt coverage of approximately 0.14 ± 0.05° per decade (Fig. 2a). The TPG method exhibits more poleward TEL occurrences than the TPD method. Tropical belt coverage expanded by approximately 0.27 ± 0.06° per decade using the TPG method (Fig. 2b). Meng et al. (2021) reported the latitudinal zone from 30.00 to 40.00° N exhibits the largest LRT-H trends, attributable to the poleward shift of the subtropical jet and tropical belt widening over the past 40 years (Staten et al., 2018). The TPD TEL demonstrates a strong positive association with LRT-H, with a linear regression model yielding a coefficient of determination (R2) of 0.93 (Fig. 2c). Similarly, the TPG TEL exhibits a strong positive association with LRT-H, with R2=0.87 (Fig. 2d). The longitudinal variation of TEL for both approaches over the Mediterranean region is depicted in Fig. 2e, f. Both methods display clear longitudinal variation across the study area, showing TEL in the eastern Mediterranean located more poleward than in the western Mediterranean. This pattern results from multiple factors, including the eastern Mediterranean region's exposure to strong natural climate variability associated with tropical–extratropical teleconnection processes. Furthermore, the Mediterranean region's dual climate influence (situated between the subtropical high-pressure systems of the Hadley cell and the mid-latitude westerlies) causes longitudinal differences in TEL through variations in these circulation patterns (Alpert et al., 2005; Trigo et al., 2006). The stronger land–sea thermal gradient in the eastern Mediterranean influences regional thermal structure and circulation, leading to higher TEL values in this region (Lionello et al., 2012). Additionally, SST variability influences longitudinal TEL variation, as its patterns affect atmospheric circulation and climate boundary positions (Meli et al., 2023). As shown in Fig. 2, TEL variability in the TPG method (Fig. 2f) is higher than in the TPD method (Fig. 2e), especially in the western Mediterranean region. Moreover, the TPG TEL occurrences are more poleward, with a maximum value of approximately 42.88° N, compared to the TPD TEL maximum value of approximately 42.78° N.

Figure 2g and h depict the seasonal variation of both TPD TEL and TPG TEL, respectively. In the case of TPD TEL, JJA and autumn (SON) show higher interannual variability than that in DJF and spring (MAM). The TPD TEL median values exhibit the anticipated seasonal pattern, being maximum in JJA at about 38.83° due to warming at the surface and in the troposphere. On the other hand, the TPD TEL median is minimum in DJF at about 29.20° because of the cooling of the surface and troposphere. All seasons show a multimodal distribution of the TPD TEL, which is clearly evident in JJA and SON. For TPG TEL, the seasonal variation follows the same pattern as that of TPD TEL, being of higher variability in JJA and SON than in DJF and MAM, displaying multimodal distribution of the TPG TEL values in all seasons. The TPG TEL median values are higher than those of TPD TEL, being maximum in JJA at about 41.00° and minimum in MAM at about 30.33°. Our results are in agreement with previous studies, which reported a pronounced seasonal cycle of the TEL, characterized by a poleward shift during JJA and an equatorward shift during DJF, with a seasonal migration amplitude of approximately 5–10° latitude (Davis and Rosenlof, 2012; Davis and Birner, 2013; Lucas et al., 2014; Luan et al., 2020).

To examine the relationship between TEL and Trp at different altitudes, Trp is divided into two distinct height bands. The first band corresponds to lower Trp (Trp-1) averaged from the surface to 5 km in height, and the second band corresponds to upper Trp (Trp-2) averaged from 5 km in height to LRT level. Figure 3 exhibits the spatial distribution of Pearson correlation coefficients (r) between the deseasonalized anomaly time series of TPD TEL and TPG TEL with climatic parameters (Ts, Trp-1, Trp-2, precipitation, and SPEI), which were computed independently at each grid point. Ts displays abundant positive correlations with TPD TEL across the study area, with the strongest positive correlation of approximately 0.34 in the eastern Mediterranean and a weak negative correlation, which is greatest at about -0.09 in the western Mediterranean (Fig. 3a). In addition, Ts shows positive correlations with TPG TEL in the eastern Mediterranean with a maximum value of about 0.35 and negative correlations in the western Mediterranean with a greatest value of about -0.15 (Fig. 3b). The predominance of positive correlations between Ts and TEL over the Mediterranean region indicates that warmer surface conditions tend to accompany a northward displacement of the TELs. Such a relationship is physically consistent with the thermodynamic expansion of the tropospheric column under warming, which can contribute to a rise in tropopause height. The east–west contrast reflects the strong regional heterogeneity of the Mediterranean climate, including land–sea thermal contrasts and differences in the influence of continental and maritime air masses.

Trp-1 exhibits a modest correlation with both TEL definitions over all grid cells (Fig. 3c, d), with correlation values ranging from -0.06 to 0.22 and -0.15 to 0.32 for TPD TEL and TPG TEL, respectively. In the case of Trp-2, the correlation with TEL is stronger than that in the case of Trp-1, ranging from -0.42 to 0.41 and -0.52 to 0.55 for TPD TEL and TPG TEL, respectively. Positive correlation values between Trp-2 and the TELs are predominantly observed in the southeastern and northwestern parts of the study area, whereas negative correlation values are predominantly found in the intervening region (Fig. 3e, f). The stronger relationship of TEL with Trp-2 signifies that variations in the upper troposphere thermal structure are more closely associated with TEL variability than changes confined to the lower troposphere. The alternating positive and negative correlation bands in Fig. 3e and f reflect regional differences in Trp-2 variability associated with changes in large-scale circulation and the meridional distribution of the subtropical thermal structure. The precipitation shows low correlations with both TEL definitions, with abundant negative values in the eastern Mediterranean and positive values in the western Mediterranean, varying from -0.27 to 0.19 for TPD TEL and from -0.32 to 0.25 for TPG TEL (Fig. 3g, h). The negative relation of TEL locations with precipitation is dynamically consistent with the poleward migration of the descending branch of the Hadley circulation, which displaces subtropical dry zones and modifies the positions of storm tracks. The SPEI shows weak correlations with TELs (Fig. 3i, j), ranging from -0.22 to 0.14 and -0.21 to 0.15 for TPD TEL and TPG TEL, respectively. The weak linkage between TEL positions and SPEI arises because SPEI represents an integrated hydroclimatic response that depends jointly on precipitation and atmospheric evaporative demand. Consequently, the response of SPEI to large-scale TEL variability can be partly obscured by regional variations in precipitation, temperature, soil moisture, and surface water availability. These results demonstrate that the correlation patterns of TEL from both definitions with climatic parameters are nearly identical and are in accordance with the findings of Darrag (2024).

For further investigation into the effects of poleward tropical belt expansion on climatic parameters in the Mediterranean region, SVD analysis is performed to reveal coupled patterns of covariability between TEL position and climatic parameters. In the case of the SVD analysis results, we consider only the first two dominant modes of covariance. Each SVD mode consists of a pair of spatial patterns representing the dominant coupled covariance structure between the TEL field and the corresponding climatic variable. SVD1 and SVD2 depict the spatial patterns for the first and second paired modes of covariability, respectively. whereas SC1 and SC2 represent the time series of the paired expansion coefficients for the same modes. For all parameters employed in the SVD examination, their SC' time series and related spatial variability maps are scaled to range from -1 to 1 to enable comparison and description of the paired modes of covariability (Darrag et al., 2024). Figure 4 shows the first leading mode of covariance between the TPD TEL and climatic variables over the Mediterranean region. The left and middle panels show the SVD1 of TPD TEL and the corresponding climatic field, respectively. The TPD TEL map represents the spatial structure of the TEL variability contributing to the dominant covariance mode, whereas the corresponding climatic-variable map represents the spatial structure that covaries with that TEL pattern. Whereas the right panels display their paired SC1. The squared covariance fraction (SCF) indicates the proportion of the total covariance captured by the first mode, while r quantifies the strength of the temporal relationship between the corresponding paired SC1. The first dominant mode explains the majority of the coupled covariance between TPD TEL and the investigated climatic parameters, accounting for 56.48 %–86.32 % of the total covariability. This signifies that the leading mode represents the dominant large-scale pattern linking tropical belt variability with the Mediterranean climate.

For the first dominant coupled mode of TPD TEL and Ts, the SCF explains 62.87 % of the total covariability of both fields. The TPD TEL SVD1 exhibits a pronounced east-west gradient across the Mediterranean (Fig. 4a), characterized by a northward displacement over the eastern Mediterranean and relatively southward displacement over the western Mediterranean. The corresponding Ts SVD1 shows an evident negative covariability relationship with TPD TEL. As it is characterized by positive covariance over western and northern regions of the study area, while it shows negative covariance over the southeastern Mediterranean and North Africa (Fig. 4b). The spatial structure demonstrates that the leading mode's covariance component does not account for the relationship between TEL variability and Ts. Because substantial TEL variability in the eastern Mediterranean is expected to be accompanied by positive Ts anomalies. The paired SC1 of TPD TEL and Ts displays a positive correlation of approximately 0.30. Furthermore, interannual oscillations dominate the SC1 for the temporal variability-paired mode of TPD TEL and Ts (Fig. 4c). In the case of the coupled fields of TPD TEL and Trp-1, the SCF of the first prevailing mode accounts for approximately 56.48 % of the total covariability of both fields. The TPD TEL SVD1 displays an east-west dipole (Fig. 4d), whereas the associated SVD1 of Trp-1 reveals enhanced positive covariance over the eastern Mediterranean and North Africa. On the other hand, it shows a negative covariance over the western Mediterranean (Fig. 4e). The first mode suggests that opposite-phase coupling between TPD TEL and Trp-1 is prevalent throughout the study area, implying that the thermal response of Trp-1 to TEL variability is not characterized in the fraction of covariance represented by SVD1. The paired SC1 of TPD TEL and Trp-1 depicts a synchronous fluctuation and has a correlation of about 0.29 (Fig. 4f).

The first coupled mode of TPD TEL and Trp-2 describes approximately 86.32 % of the total covariability of both fields, indicating that this mode dominates the shared variability. The TPD TEL SVD1 exhibits widespread negative covariance over the Mediterranean basin (Fig. 4g). While the Trp-2 SVD1 exhibits a distinct zonal positive fluctuation band that spans North Africa, South Europe, and the eastern Mediterranean (Fig. 4h). The paired SC1 of TPD TEL and Trp-2 displays strong temporal agreement throughout the analysis period (r=0.67), suggesting that variations in TPD TEL are closely linked to changes in the Trp-2 thermal structure (Fig. 4i). The tropopause and TEL are sensitive to Trp-2 variation, which can modify the meridional thermal structure and atmospheric thickness and can therefore influence the location of the subtropical tropopause and the associated transition between tropical and extratropical atmospheric regimes. In the case of the coupled fields of TPD TEL and precipitation, the SCF of the first leading mode explains approximately 68.59 % of the total covariability of both fields. The SVD1 reveals a pronounced longitudinal gradient in the TPD-TEL pattern (Fig. 4j), while the precipitation SVD1 exhibits predominantly negative fluctuation across much of the Mediterranean basin (Fig. 4k). The opposite-phase coupling between TPD TEL and precipitation over the eastern Mediterranean suggests that poleward widening of the tropical belt may be associated with expansion of subtropical dry zones and reduced precipitation. As depicted in Fig. 4l, there is a correlation between the paired SC1 of TPD TEL and precipitation of approximately 0.42, indicating that tropical belt variability is associated with changes in Mediterranean precipitation, which is controlled by regional circulation and thermodynamic processes. For the coupled fields of TPD TEL and SPEI, the SCF of the first dominant mode represents approximately 61.32 % of the total covariability of both fields. The TPD TEL SVD1 is characterized by widespread positive fluctuations across the Mediterranean (Fig. 4m), whereas the corresponding SPEI SVD1 is characterized by alternating positive and negative fluctuation zones across the Mediterranean (Fig. 4n), with a clear positive covariance in the eastern Mediterranean and negative covariance in the western Mediterranean, reflecting regional differences in drought response. The SC1 of both TPD TEL and SPEI has a correlation of about 0.37. Moreover, both show interannual and interdecadal oscillations (Fig. 4o). The heterogeneous SPEI spatial pattern shows that the hydroclimatic response to TEL variability is spatially nonuniform. This pattern is expected because SPEI integrates the effects of precipitation and atmospheric evaporative demand, both of which exhibit substantial regional variability.

Unlike the first leading mode of covariability, the second coupled mode explains a substantially smaller fraction of the total covariance, ranging from 11.05 % to 32.07 %. Although this mode contributes less to the total covariability, it is characterized by more localized spatial structures, as is evident in SVD2, and the paired SC2 remain positively correlated (r=0.27–0.58), indicating that the secondary mode still reflects meaningful interactions between tropical belt variability and Mediterranean climate (Fig. 5). For the coupled fields of TPD TEL and Ts, the SCF of the second dominant mode explains 26.77 % of the total covariability of both fields. The second coupled mode of spatial variability, SVD2, shows that TPD TEL is characterized by observed negative loadings over most of the study area (Fig. 5a), accompanied by Ts fluctuations of the same sign over the majority of the eastern Mediterranean and a limited zone of the western Mediterranean. On the other hand, the SVD2 of Ts shows positive fluctuations in the central Mediterranean. The second SVD mode exhibits a pronounced positive coupling between TPD TEL and Ts over the eastern Mediterranean, indicating that poleward displacement of the TEL is associated with enhanced surface warming in this region. The Ts spatial heterogeneity may reflect the strong land–sea thermal contrast of the Mediterranean region and differences in surface energy partitioning between continental and maritime environments (Fig. 5b). The paired SC2 of TPD TEL and Ts exhibits a correlation of about 0.29 (Fig. 5c). In the case of coupled fields of TPD TEL and Trp-1, the SCF of the second dominant mode accounts for 32.07 % of the total covariance of both fields, representing the largest secondary contribution among all variables. The TPD TEL SVD2 shows abundant negative fluctuations over the Mediterranean (Fig. 5d), which coincide with Trp-1 fluctuations of the same sign in most of the eastern Mediterranean and a restricted area of the western Mediterranean. Meanwhile, the SVD2 of Trp-1 displays a well-defined meridional dipole, with positive covariance centered over North Africa and South Europe (Fig. 5e). Both TPD TEL and Trp-1 show evident coupling over the eastern Mediterranean, covering a wider area than that in the case of TPD TEL and Ts, suggesting that the poleward expansion of the tropical belt is associated with an increase in Trp-1. The paired temporal coefficients for TPD TEL and Trp-1 exhibit a weak correlation of approximately 0.27 (Fig. 5f).

For the coupled fields of TPD TEL and Trp-2, the SCF of the second prevailing mode represents only 11.05 % of the total covariability of both fields, implying that most of the shared variability between these two variables is already captured by the first mode. Nevertheless, the paired SC2 of TPD TEL and Trp-2 remain strongly correlated (r=0.58), indicating that although this secondary pattern contributes little to the overall covariance, it still represents a physically meaningful mode of coupled variability (Fig. 5i). The SVD2 of the TPD TEL displays an observed east–west gradient over the Mediterranean region (Fig. 5g), which is accompanied by Trp-2 fluctuations of opposite sign in the form of a narrow zonal band of alternating positive and negative covariance centered near 30° N, extending across North Africa and the eastern Mediterranean (Fig. 5h). The negative coupling between TPD TEL and Trp-2 over this narrow band is consistent with the negative relationships of tropopause and near-tropopause temperatures with both tropopause height and TEL position. Although this secondary mode of covariability between TPD TEL and Trp-2 contributes relatively little to the total covariance, it may represent regional modulation of the dominant influence of upper-tropospheric thermal variability on TEL. In the case of the coupled fields of TPD TEL and precipitation, the SCF of the second leading mode describes approximately 20.88 % of the total covariability of both fields. As exhibited in Fig. 5l, there is a correlation between the SC2 of TPD TEL and precipitation of approximately 0.32. The TPD TEL SVD2 shows widespread negative fluctuations over the study area (Fig. 5j). While the precipitation SVD2 depicts fragmented positive and negative fluctuations distributed throughout the Mediterranean basin (Fig. 5k), it contrasts the broad-scale precipitation pattern identified in SVD1 (Fig. 4k). The second covariability mode between TPD TEL and precipitation exhibits pronounced negative coupling over the eastern Mediterranean and southern Europe, indicating that poleward displacement of the tropical belt edge is associated with reduced precipitation in these regions. For the coupled fields of TPD TEL and SPEI, the SCF of the second dominant mode accounts for approximately 31.90 % of the total covariability of both fields. As shown in Fig. 5o, the correlation between the SC2 of TPD TEL and SPEI is approximately 0.44. The TPD TEL SVD2 shows an obvious east-west gradient over the Mediterranean region (Fig. 5m), while SPEI SVD2 is characterized by opposite-sign variations relative to TPD TEL over the eastern Mediterranean and adjacent areas of North Africa (Fig. 5n), suggesting that poleward displacement of the TEL in this region is associated with drier conditions. The spatial heterogeneity in SPEI likely arises from the combined effects of precipitation variability and atmospheric evaporative demand, which jointly regulate regional hydroclimatic conditions.

The SVD analyses results of TPG TEL and climatic parameters over the Mediterranean region reveal a high degree of consistency with those of TPD TEL and climatic parameters. The first leading mode of covariability for the coupled fields of TPG TEL and climatic variables (Fig. 6) shows coupling patterns nearly identical to those for TPD TEL and climatic variables (Fig. 4). The SVD1 for coupled fields of TPG TEL and climatic parameters explains the largest SCF (61.13 %–82.13 %) and exhibits nearly identical large-scale spatial structures, indicating that the primary coupled signal is robust and largely independent of the TEL definition. Likewise, the paired SC1 of TPG TEL and climatic variables display similar temporal evolution, with the strongest r observed for Trp-2 of about 0.67. In the case of the second dominant mode of covariability, the coupled fields of TPG TEL and climatic parameters (Fig. 7) closely mirror those derived from the TPD TEL and climatic parameters (Fig. 5) with a smaller fraction of the total covariance (SCF = 13.58 %–25.29 %) and capture more localized regional variability, characterized by enhanced spatial patterns and weak but consistent correlation between paired SC2 (r=0.30–0.57). The differences between the SVD analyses results of both TPD TEL and TPG TEL with climatic parameters are limited to minor variations in the amplitude and local distribution of the spatial anomalies, and the correlation between paired SC2 is higher than that of paired SC1 in the case of coupled fields of TPG TEL with Ts and Trp-1, but the overall spatial and temporal covariability remains remarkably consistent.

The Mediterranean region's monthly average Ts over the study period has increased by approximately 0.48 ± 0.03 K per decade (Fig. 8a). This is in line with the results of Castellanos et al. (2021), who found land surface air temperature was significantly rising, with yearly amplitudes ranging from 0.006 to 0.027 °C for maximum temperatures and slightly greater rates for minimum temperatures. The primary drivers of Mediterranean region warming are the rapid increase of anthropogenic GHGs and the decline of aerosols and soil moisture (Kim et al., 2019; Urdiales-Flores et al., 2023). Linear regression analysis of Ts was performed at all grid points, with decadal trends plotted in Fig. 8b. Ts shows a dominant upward trend across the study area. The greatest temperature upward trends are evident in the eastern Mediterranean region, while temperature trends in the western Mediterranean are less pronounced. This Ts spatial pattern may arise from differences in land–sea thermal contrast, surface energy balance, soil-moisture conditions, aerosol loading, and regional atmospheric circulation. Moreover, the zonal mean of the temperature linear regression map (right panel of Fig. 8b) shows positive temperature trends throughout the entire study area with maximum trends in the northern zone of the Mediterranean from 46.00 to 50.00° N. Zittis et al. (2022) reported that the Eastern Mediterranean and the Middle East are warming approximately twice as rapidly as the global average, indicating the Mediterranean as a whole is experiencing dramatic warming.

Trp-1 shows increasing trends of approximately 0.36 ± 0.03 K per decade (Fig. 8c). The decadal trends of Trp-1 at all grid points over the study area (Fig. 8d) depict increasing trends caused mainly by global warming. Trp-1 trends in the eastern and northern Mediterranean are stronger than those in the southern and western Mediterranean, and the zonal mean of Trp-1 linear regression map (right panel of Fig. 8d) depicts dominant positive trends throughout the entire study area, with the strongest trends in the zone from 45.00 to 50.00° N. For the monthly time series of Trp-2 (Fig. 8e), the decadal trend is 0.09 ± 0.01 K per decade, which is less than that of Trp-1. Additionally, linear regression analysis of Trp-2 performed at all grid points shows clear variation in temperature trends. In the northern Mediterranean, positive Trp-2 trends are abundant, while in the southern Mediterranean, negative trends are prevalent (Fig. 8f). The zonal mean of Trp-2 linear regression map (right panel of Fig. 8f) displays clear negative trends around the latitude of 31.25° N. Our results are compatible with Karl et al. (2006), who reported that global-average Trp has increased significantly since the late 20th century. From 1958 to the present, the global average temperature has risen at a rate of approximately 0.12 °C per decade, accelerating to about 0.16 °C per decade since 1979.

The decadal trends of the monthly zonal mean temperature structure during 1980–2022 from the earth's surface to 20 km, including the troposphere and lower stratosphere, and the LRT-H zonal mean are shown in Fig. 8g. As evident, there are apparent upward temperature trends in the whole troposphere, with the greatest increasing trends over the latitudinal band from around 45.00 to 50.00° N, with a maximum value of 0.50 K per decade. As is clear from the zonal mean of linear regression trend maps of Ts, Trp-1, and Trp-2 (right panel of Fig. 8b, d, and f), there is an observed increase in temperature trends also over the latitudinal band from around 45.00 to 50.00° N. The warming in the troposphere is associated with apparent cooling trends in the stratosphere at a greatest rate of slightly over -0.50 K per decade while the latitudinal band 20.00 to 30.00° N is the zone of maximum stratospheric temperature declining trends. The phenomenon of tropospheric warming accompanied by stratospheric cooling is a signature fingerprint of human-induced climate change, principally driven by rising GHG emissions. These assumptions are corroborated by several studies on tropospheric warming and stratospheric cooling employing radiosonde, GNSS-RO, and satellite observations (Narayana Rao et al., 2007; Titchner et al., 2009; Haimberger et al., 2012; Steiner et al., 2020). The contrasting thermal responses of the troposphere and lower stratosphere also modify the vertical temperature gradient near the tropopause and may therefore affect the height and spatial distribution of the LRT.

Total precipitation over the Mediterranean exhibits decreasing trends of approximately -0.25 ± 0.05 mm per decade (Fig. 9a). This is in accordance with previous studies that reported a clear trend toward declining precipitation, particularly during DJF months, which poses serious implications for the region's ecosystems and water resources (Cos et al., 2022; Delworth et al., 2022; André et al., 2024). Linear regression analysis of total precipitation in Fig. 9b exhibits a dominant downward trend evident in eastern and western Europe. Such drying may be associated with changes in the large-scale circulation, including the poleward expansion of the subtropical circulation and associated modifications of the Mediterranean storm tracks, which can reduce the frequency or intensity of precipitation-producing weather systems over parts of the basin. The zonal mean of the precipitation linear regression map (right panel of Fig. 9b) exhibits evident positive trends at approximately latitude 41.00° N while depicting a strong negative trend between 47.00 and 50.00° N. SPEI is commonly used to monitor meteorological drought status; it shows a decrease of -0.070 ± 0.001 per decade (Fig. 9c). Recent studies have highlighted significant trends in drought occurrences and characteristics across the Mediterranean Basin (Essa et al., 2023), and historical analyses using SPEI from 1980 to 2014 reveal a significant upward trend in drought frequency (Choutri and Hussien, 2024; Kesgin et al., 2024). SPEI linear regression analysis exhibits evident decreasing trends, especially in the northern Mediterranean (Fig. 9d), with a zonal mean (on the right panel of Fig. 9d) showing a strong negative trend between 40.00 and 50.00° N. The observed negative SPEI trend likely reflects the combined influence of precipitation deficits and increasing evaporative demand associated with regional warming.