Ocean–atmosphere circulation coherences associated with temperature increase in Pakistan

In recent decades, temperature variations have significantly affected the ecosystem and human livelihood in Pakistan. The wavelet analysis is employed to identify the associations between regional temperature change and global teleconnections, i.e. Atlantic Multidecadal Oscillation (AMO), Arctic Oscillation (AO), North Atlantic Oscillation (NAO), El Niño–Southern Oscillation (ENSO), Pacific Decadal Oscillation (PDO), Tropical Northern Atlantic Index (TNA), Pacific/North American Index (PNA), North Pacific Pattern (NP), Western Pacific Index (WP), and Western Hemisphere Warm Pool (WHWP). This paper first uses monotonic modified Mann–Kendall and Sen’s slope estimator to compute the temperature changes in Pakistan and its homogenous climatic regions during 1960–2020. It is found that the temperature in Pakistan has increased significantly at 0.23 °C dec−1 in the last 62 years, which is also higher than the global average increase. This increase is more evident in region IV and V in spring at 0.63 and 0.43 °C dec−1 respectively, followed by summer and autumn. Temperature changes in Pakistan and its sub-regions are mainly associated with NP, WP and WHWP with higher mean significant coherences. Overall, temperature changes are significantly influenced by multiple interactions of global teleconnections, and these combinations indicate that the integrated influence of teleconnections can better explain the regional temperature changes. The teleconnections with broader significant influence over Pakistan are NP + WP + WHWP, followed by Pacific-based (ENSO, PDO, and PNA), and Atlantic-based (AMO, AO, and NAO) indices as revealed by the mean significant coherences of 0.82, 0.82 and 0.72 respectively. Annually, AMO, TNA and WHWP showed significant correlation with higher magnitudes of 0.44, 0.42 and 0.20 respectively, indicating the modulation effect of these teleconnections on temperature changes over Pakistan. The combined impacts from the various teleconnections represent a considerable advancement in the accuracy of significant temperature variations over the country. These findings have significant implications for policymakers in terms of better planning and policies in light of climate change as well as atmospheric circulation variability.


Introduction
The increase in Earth's surface temperature caused by human activities has countless effects on human health and ecosystems (Watts et al 2021), whereas understanding on ecosystem is vital for the regulation of global climate change . As one of the most populated regions in South Asia (Ullah et al 2020b), warming in Pakistan is anticipated to increase above the global mean (Eckstein et al 2018, Ali et al 2021). A huge portion of the country's economy is tied to agriculture (Almazroui et al 2020); however, this sector is highly vulnerable to climate change (Aryal et al 2020). In the past two decades, Pakistan has been significantly affected by extreme climate events (i.e. heatwaves, droughts, and floods), which resulted in significant socioeconomic losses . The increasing threat of climate change and its extremes in Pakistan has been widely recognized (Iqbal et al 2016, Hussain et al 2019, Saleem et al 2021, Ullah et al 2021; however, limited research has been conducted on understanding the underlying factors of climate variability in the country. As the extended and severe dry spells, as well as the increased evaporation brought on by higher temperatures, would have a considerable impact on water supplies (Hussain et al 2023a). Therefore, understanding the prominent associations between global teleconnections and their impact on interannual to multidecadal regional climate variability controls is imperative for disaster risk reduction and sustainable development in Pakistan.
Global teleconnections have received wide attention due to the climate-associated variabilities in Pakistan temperatures and teleconnections during spring, and the NAO has an influence during the monsoon and in August. Saleem et al (2021) reported strong links between La Nina (El Nino) events over the western Pacific Ocean and variations in extreme temperatures. The severity of La Nina episodes has a higher effect on the intensity of temperature extremes in Pakistan. Furthermore, Rashid et al (2022) found negative anomaly of surface air temperature accompanied by ENSO. Moreover, the upper-level anticyclonic circulation anomalies associated with the upper to middle troposphere favor clear skies with an increase in net shortwave radiation, resulting in extreme warming over Pakistan. Hussain et al (2021) observed inter-annual seasonal maximum and minimum temperature coherences with ENSO, NAO, PDO, and Indian Ocean Dipole (IOD) in northern Pakistan. Karim et al (2023) found a strong correlation between warm winter temperatures and the warming of the western and eastern Indian Oceans; however, insignificant relation to the Nino3.4 index. As observed by Dogar and Almazroui (2022), the Hadley Cell is strengthened by the El-Niño, but the La-Niña phase has the opposite effect; altering the circulation of the Inter-Tropical Convergence Zone, this ENSO-induced strengthening and weakening of the Hadley Cell has a significant impact across South Asia. Recently, Nawaz et al (2023) also highlighted the significant influence of the Pacific variabilities on the hydro-climatology over northern Pakistan. Increased warming is observed across South Asia from the mid-1990s onward (Song et al 2022). Therefore, under dense population and socioeconomic significance, it is important to comprehend the root causes of these shifts from a wider perspective.
Many techniques, including Pearson correlation, principal component analysis, multiple correlations, empirical orthogonal function, and wavelet analysis have been used to evaluate the relation between temperature variations and oceanic indicators (Athira et al 2022). However, it is important to note that the relationship between climate and oceanic indices is much more complex than the basic linear methods (He and Guan 2013, Hussain et al 2022b). Hence, to investigate the significant modes of variability, we applied wavelet transform coherence (WTC). The WTC swiftly determines the dominant modes of variation over time and space by converting time series into time-domain frequency and space (Torrence andCompo 1998, Hussain et al 2022c). It has been used in a wide range of climatological investigations (Labat 2010, Hu and Si 2016, Tan et al 2016, Su et al 2019, Rezaei and Gurdak 2020, Hussain et al 2021, Morales-Márquez et al 2021, Pérez-Ciria et al 2022. As mentioned above, few studies have focused on the temperature changes and its associated large-scale mechanism in relation to atmospheric circulations. However, the dynamic influence of ocean-atmospheric coherences and regional climatic changes over homogenous climatic regions of Pakistan needs more attention. Indices from the Pacific and Atlantic Ocean have diverse kind of influences over Pakistan. Like, the western disturbances from Atlantic have a significant influence on the winter moisture transport over northern Pakistan, whereas the ENSO and Indian Summer Monsoon Index (ISMI) from the Pacific and the Arabian Oceans respectively. Moreover, the temporal influence of these indices from monthly, seasonal, annual to interannual scales (NAO, ENSO, PNA), and then from decadal to interdecadal scales (i.e. PDO, and AMO). However, one of the major goal of this study to explore the influence of multiple indices through the time from the Pacific and Atlantic Oceans. To this end, the combined influence of indices from the Pacific Ocean (ENSO, PDO, and PNA) and the Atlantic Ocean (AMO, AO, and NAO) on regional climatic variations over Pakistan are not been thoroughly examined in the literature.
In this context, there are still many significant knowledge gap in the effects of the large-scale climate variability controls over the diverse topography/homogenous climatic regions of Pakistan. To understand the association of observed temperature variability over homogenous climatic regions of Pakistan between 1960 and 2020 with global teleconnections. We (1) first assessed the climatology and trends, (2) explored the relationship of observed sub-regional temperature variation concerning global average surface temperature anomalies (GASTAs), (3) quantified the sporadic interannual and inter-decadal coherences of monthly temperature over different region across Pakistan, (4) determine the coupled (two and three) influence of various teleconnections on temperature variability over Pakistan. Lastly, (5) spatial correlation of regional temperature with global sea surface temperature (SST) indices are also mapped to explore for more information. Large-scale climatic changes are mostly associated with various oceanatmosphere circulation at a time, and it can be understand better by assessing the combined influence of such climate indices.

Study area
Pakistan lies in South Asia, encompassing 23.6 • -38 • North latitude and 61 • -78 • East longitude . It has an area of 881 912 km 2 and an elevation range of 0-8611 m that decreases southward (figure 1) (Akhtar et al 2022). The country has a complex climatology and multifaceted topography, such as the Arabian Sea to the south and the Himalayas, Hindu Kush, and Karakoram mountains to the north . The country has four seasons (winter: December-February, spring: March-May, summer: June-August, autumn: September-November), having spatial and temporal climatic variation . The climate varies greatly depending on location, with one of the biggest non-polar glaciers in the north and a coastal region in the south (Khan et al 2023). A humid climate predominates in the northern and southern mountainous regions, with annual temperature variations between 0 • C and 20 • C. Similarly, the Indus plains, located in the center and east, have a tropical climate with mean annual temperatures ranging from 21 • C to 25 • C. In contrast, the southern coastal area has a coastal climate, and the winter, spring, summer, autumn and annual mean temperature for the whole country are 11.2 • C, 22.4 • C, 29.6 • C, 21.8 • C, and 21.3 • C respectively (Hussain et al 2023b). However, the whole country is divided into six homogenous climatic regions to understand the regional climatic variation (Hussain et al 2022a(Hussain et al , 2022b. Based on topography and climatology, R-I has a semi-arid climate, with maximum number of glaciers outside the polar region, also known as the upper Indus basin (UIB), and it stretches through the Hindu Kush-Karakoram-Himalaya and Tibetan plateau (TP). Here, maximum temperature below 1500 m varies between 20 • C and 30 • C, whereas it reaches below 0 • C over higher altitudes, primarily over Karakoram region >2800 m. Regions over 1500 m receive precipitation in the form of snow during late autumn, winter and sometimes in the advent of spring as well (Hussain et al 2021). R-II represents a humid climate, with mean annual precipitation of 1325 mm, with cold winter and hot summer (Hussain et al 2022a). In R-III, temperature fluctuates between 2 • C and 45 • C, but can reach as high as 47 • C in summers and as low as 5 • C in winters, whereas precipitation ranges from 100 mm in the south to 600 mm in the northwest, but frequently reach 1000 mm along the northeast . R-IV represents arid and shrub-land, sub-mountainous region with extreme cold winter and hot summer (Arshad et al 2021). R-VI has an arid climate, plane area and croplands, and warm winter and a scorching summer (Ullah et al 2019b). The locations in region I are Gilgit-Baltistan, Chitral UIB, region II includes Parachinar, Islamabad, and Jammu and Kashmir. Region III has most of Punjab province, central and eastern parts of Khyber Pakhtunkhwa, while region IV includes Zhob and Quetta region, southwestern parts of Punjab, and western Khyber Pakhtunkhwa. Moreover, region V includes almost major parts of Balochistan province, and region VI has Sindh province and southwestern parts of Balochistan, including Jiwani and Pasni (Hussain et al 2022b).

Data
The temperature variations in homogeneous climate zones of Pakistan (figure 1) and their association with global teleconnections during 1960-2020 are examined in this study. The study utilized observed mean temperature records of 50 climate stations acquired from the Pakistan Meteorological Department (www.pmd.gov.pk/en/). The mean temperature data underwent various quality control methods before application. Like, the stations were selected based on data length, completeness, and quality (i.e. consistency and homogeneity), and deviations attributed to climatic and non-climatic factors were checked (Kidd et al 2017, Ullah et al 2021, Hussain et al 2022a, 2022c, Rebi et al 2023. As, it is important to guarantee that observed temperature trends must represent natural variability instead of non-climatic factors, such as heat islands (El Kenawy et al 2012, Hussain et al 2023b). Moreover, the World Meteorological Organization recommends such techniques to ensure the quality, consistency, and uniformity (Hussain et al 2023b).
The global teleconnections/oceanic indices (i.e. AMO, AO, NAO, ENSO, PDO, TNA, PNA, North Pacific Pattern (NP), Western Pacific Index (WP), and Western Hemisphere Warm Pool (WHWP)) are downloaded from the National Oceanic and Atmospheric Administration data archive (https:// psl.noaa.gov/gcos_wgsp/Timeseries/). The details of each index are briefly explained in table 1. The number of meteorological stations in the sub-regions I to VI are 7, 9, 16, 2, 5, and 11, respectively (figure 1). The monthly temperature records from all stations were averaged in all sub-regions for regional means and the whole country for the country mean (Hussain et al 2022a, 2022b).

Cluster analysis
Clustering is one of the methods to explore the underlying structure/pattern of data to gain insights that might be missed through other analytical approaches. It is used to partition a dataset into similar clusters based on their similarities across data (MacQueen 1967). It is a subfield of multivariate statistical analysis and a method for unsupervised machine learning and pattern identification (Yang and Hussain 2023). Cluster analysis is employed in this study to determine and investigate the homogenous climatic regions over Pakistan based on seasonal and annual temperature datasets. We adopted the criteria of defining the homogeneous climatic regions based on the findings of (Ullah et al 2020a). The readers are referred to (Ullah et al 2020a, Sinaga et al 2021, Hussain et al 2022a, 2022b, Yang and Hussain 2023 for more details regarding cluster analysis. Cluster analysis is expressed as mentioned in equation (1): Here, a valid d-dimensional vector exists for the observations x 1 , x 2 , …, x n . The K-means clustering algorithm divides n observations into k sets (k ⩽ n) of E C = C 1 , C 2 , …, C k to reduce the total number of members squares inside the cluster µ i denotes the average of the points in C k .

Modified Mann-Kendall and Sen's slope
The nonparametric modified Mann-Kendall (mMK) and Sen's slope estimator (SSE) (Sen 1968, Hamed andRao 1998) techniques were used to investigate the significance and slope of regional temperature trend in the study area. Both the mMK and the SSE approaches are very successful against data oddity and missing values in a time series and do not need the normality of data (Ullah et al 2019a). The variance V(S) * in the mMK test is calculated using equations (2) and (3) as follows: where, r i denotes the i delayed autocorrelation coefficient and V(S) signifies the MK test. V(S) is substituted by V(S) * from the MK test to produce the Z statistic in the mMK test (Ahmadi et al 2018). Moreover, the slope estimates of N data pairs are obtained from equation (4) Here, 1 < I < j < n, and n is the length of the climate dataset , Hussain et al 2022a. A positive slope value shows a rising trend in the temperature time series, whereas a negative slope value indicates a descending trend (Hussain et al 2021).

WTC
The WTC was used to investigate the significant coherences of the dominant monthly temperature in the homogenous climatic regions of Pakistan with various global teleconnections/oceanic indices. This technique combines cross-spectrum analysis with wavelet transform. It has been utilized to quantify the covariance between hydro-meteorological variables and their drivers in the time-frequency domain (Irannezhad et al 2020). This method is more suitable due to the local phase-locked behavior (Grinsted et al 2004). The WTC between two-time series X and Y having transform wavelet W X i and W Y i is given by (Torrence and Compo 1998): Here S is the smoothing operator. It is important to consider the coherence as a localized correlation coefficient in time-frequency space since this description closely resembles that of a standard correlation coefficient (Hussain et

Multiple wavelet coherence
Multiple wavelet coherence (MWC) is further used to identify localized and scale-specific multivariate relationships, and it outperforms other common multivariate methods in such applications (Hu and Si 2016). The WTC calculates the wavelet power of dependent time series concerning two or three independent datasets at a specific time and frequency (Ng and Chan 2012). In this study, we employed the algorithm of Hu and Si (2016), which is a modified version of the Grinsted et al (2004) algorithm, to calculate the mean significant coherence between single and combined indices among clusters (Rezaei 2021). If there are several predictor variables X(X = {X 1 , X 2 , . . . .X q }) and a single response variable Y, then the MWC at scales s and location τ , ρ 2 m (s, τ ) can be expressed as: . (6) Here, ← → W Y,X (s, τ ) is the smoothed cross-wavelet power spectra between predictor variable X and response variable Y. Moreover, ← → W X,X (s, τ ) is smoothed matrix of auto-and cross-wavelet power spectra among various predictor variables X;  Figure 2 depicts the monthly seasonal and temporal temperature trends over various homogenous climatic regions of Pakistan. Results of mean monthly temperature show variation in temperature over various homogenous climatic regions ranging from 2 • C to 18 • C, with the maximum temperature observed in May, June, and July. Here, the lowest mean temperature was observed in R-IV (regions mentioned in figure 1), followed by R-II, R-V, and R-III, whereas the highest means were observed in R-VI to the southern Pakistan ( figure 2(a)) and in R-I to the northeastern parts of the country. Moreover, seasonal temperature variation ( figure 2(b)) depicts the same pattern observed in the mean monthly temperature over various homogenous climatic regions of Pakistan. In comparison to the monthly, seasonal, and annual temperature, the temporal variation over various regions during 1960-2020 explains the maximum variation in R-VI ranged from 25 • C to 27 • C, followed by R-III (23 • C-25 • C), R-V and the whole country (19 • C-22 • C), whereas minimum temporal variability was observed in R-I in the range of 14 • C-16 • C (figure 2(c)).

Climatology
The mean temperature in winter increased significantly in R-I, III, IV, VI, and the whole country at 0.18, 0.21, 0.10, 0.68, and 0.26 • C dec −1 , respectively (table 2). In contrast, summer temperature decreased in R-I, III, IV, and in the whole country at the rates of −0.28, −0.07, −0.05, and −0.05 • C dec −1 , respectively. Temperature has increased significantly in R-III, IV, V, VI, and the whole country at the rates of 0.11, 0.18, 0.28, 0.18, and 0.08 • C dec −1 , respectively, during autumn (table 2) and at 0.12, 0.30, 0.30, 0.18, and 0.12 • C dec −1 , respectively, over the entire year. Figure 3 shows that the temperature increase rate (0.23 • C dec −1 ) in Pakistan is more than that occurred on the global scale (0.22 • C dec −1 ) during 1960-2020. Moreover, the response to regional temperature changes concerning the GASTA was assessed using the WTC and MWC. The WTC analysis of temperature changes in regions R-II and IV as well as the whole country clearly indicates dominant cycles at 8-16 months. This significant mode is evident after the mid-1990s (figures 4(a)-(c)). Another dominant mode is evident at 32-64 months from nearly 2010 onward. These sporadic significant modes indicate that the temperature has increased in Pakistan substantially after the mid-1990s and doubled after 2015. Furthermore, the MWC findings in figures 4(d)-(f) indicate that regions R-II, III, and V combine to share the maximum cycles of temperature variations in Pakistan. These regions would affect more if the same patterns continue. According to the 5th and 6th assessment reports of IPCC, the regional climatic trends showed much higher deviations as compared to global scale. The IOD and the TP thermal forcing affect the regional climate of Asia, whereas the El Nino and La Nina also creates wet and dry conditions   (Hussain et al 2023a). During the wet conditions, the air composite shows negative air temperature anomalies and the whole cycle explains such deviations in the regional air temperature (Ullah et al 2021a).

Coherence between monthly temperature and oceanic indices
This section presents the WTC between the monthly temperature over Pakistan and each individual oceanatmosphere index. In WTCs, the blue color indicates less intensity within the cone of influence impacted by edge effects. In contrast, the red color shows higher coherence where red areas within the thick black lines are statistically significant at the 5% level against red noise (Rezaei 2021). The right-pointing arrows indicate that the two signals are in phase, while the left is for antiphase signals. The percent area of significant coherence compared to the overall region is evident by thick outlines, which are marked by red color patches, compared to the total areas. Figure 5 shows the WTCs between monthly temperature averaged over Pakistan and various oceanic indices (i.e. AMO, AO, NAO, TNA, PDO, PNA, NP, WP, and WHWP) during 1960-2020. Although there are some significant coherence between Pakistan's mean temperature and all the climate indices at annual periods (8-16 months), the NP, followed by WHWP and AMO, has the highest impact on the countries temperature. The coherence between NP, WHWP and AMO indices and temperature is almost positive (arrows pointing rightward). For the inter-annual cycles, there is a significant coherence at around 20-40 months between temperature and NAO, TNA, and AO concentrated over 1975-2000 as well as a significant coherence at 64-128 months between temperature and WP index. For longer than decadal periodicities, the WHWP index shows the most significant coherence with Pakistan's temperature. On a less than one-year timescale, temperature coherences displayed a robust but   sporadic coherence with AMO, AO, NAO, TNA, PDO, PNA, NP, WP, and WHWP. Overall, temperature coherences remained sporadic and significant, ranging from 8 to 128 months (0.6-10.6 years) (figures 5(a)-(i)), while a significant inter-decadal oscillation of ∼200 months (16.6 years) are evident in WHWP (figure 5(i)).
To quantitatively assess the temperature teleconnections, table 3 shows mean significant coherence (at 95% confidence level) between the temperature and the single, two-coupled, and three-coupled climate indices during 1960-2020. Only four major combinations of teleconnections are used on the basis of their significant findings. The results are presented separately for each region from R-I to R-VI as well as the whole country. On a monthly scale, the temperature variations in the six sub-regions are associated with various teleconnections with different coherence levels and large spatial heterogeneity from north to south over Pakistan. Concerning the coherence of single index analyses, the R-I's temperature is mainly influenced by AMO, PDO, PNA, NP, and WHWP (>0.40 coherences). In contrast, ENSO, PDO, NP, and WHWP dominate in regions R-II and III, with mean significant coherence levels of > 0.40. Moreover, AMO, PDO, PNA, TNA, and NP primarily influence the temperature in R-IV, while ENSO, PDO, WP, and WHWP (table 3) mostly affect R-V and VI. Temperature changes over the whole country are mainly associated with WP and WHWP with higher levels (>0.40 coherences) of mean significant coherences (table 3). Consistent with WTCs shown in figure 5, temperature changes in Pakistan and its sub-regions, overall, are mostly associated with NP, WP, and WHWP with higher mean significant coherences than the remaining indices (figure 5 and table 3). As an example, the coherence between WHWP and temperature is larger than 0.40 across the whole country except for region IV. Furthermore, the strong and broader coherence patterns for WP (figure 5(h)) and WHWP (figure 5(i)) prevail > 5 year bands for these two indices across the whole country.
To explore the combination effect of climate indices on Pakistan's temperature, we analyzed the MWC between temperature and the integrated effects of those indices show the highest coherences in singlebased analyses as well as the combined effect of Pacific-(ENSO + PDO + PNA) and the Atlanticbased (AMO + AO + NAO) indices. The combined coherence patterns of the teleconnections prevail higher mean significant bands across the sub-regions (table 3) and the whole country (table 3 and figure 6), compared to single index values. The sub-region coherence bands show higher mean significant coherences in all sub-regions with the Pacific Ocean and NP + WP + WHWP (table 3), mostly > 0.80%. The significant coherences bands of NP + WP + WHWP are at the cycles of 8-16 months (figure 6(d)). However, the Atlantic Ocean shared the coherences <0.80 for most sub-regions (II, III, and V) and the whole country (table 3). The MWC results of AMO + AO + NAO, PDO + PNA, and TNA + PNA prevailed in sporadic yet consistent coherence bands between 8 to 16 months during 1960-2020. However, the three combined MWCs shared significant coherences of ∼32-128 months (2.6-10.6 years) from 2005 onward, and if this pattern continues, it is obvious that this cycle will continue to influence the temperature changes over Pakistan in the next 5-10 years ( figure 6). Overall, the WP, WHWP, ENSO + PDO, NP + WP + WHWP, Pacific, and Atlantic Oceans, as revealed in coherences, mostly affect temperature changes in Pakistan.

Correlation between seasonal temperature and oceanic indices
Temperature variations on monthly scales differ from that in the seasonal and annual time scales, as various teleconnections have different intensities of influences over different seasons. Like, the large-scale modes in AO, NAO, ENSO, and PDO are the dominant forces that influence many parts of the world, but each acts on quite different time scales. For example, ENSO is the strongest in boreal winter and acts on seasonal and 2-5 year time scales, but PDO is a multidecadal phenomenon and acts on multidecadal time scales. Therefore, it is more important to look at and understand the teleconnection patterns of temperature changes during the different phases of the dominant large-scale modes.
For this purpose, we explored the variations in temperature on seasonal and annual time scales using the traditional Pearson correlation method (figure 7) and monthly associations to complement the observed findings of WTC and MWC (figures 5, 6 and table 3). Moreover, some indices were selected to explore the spatiotemporal correlation pattern over Pakistan for a more detailed picture (figure 8). These indices are selected according to their magnitude of correlation observed from correlation presented in figure 7. All the indices positively influence temperature changes over Pakistan in winter, except AO. Here, AMO, TNA, and WP show a modest impact over Pakistan at 0.43, 0.37, and 0.46, respectively. In spring, AMO, PNA, TNA, and WHWP modulate changes in temperature across Pakistan. The AMO and TNA show significant positive impacts, while AO, NAO, ENSO, PDO, NP, and WP negatively influence temperature change. In summer, AO, NAO, NP, and WP positively influenced temperature. At the same time, AMO, ENSO, PDO, PNA, TNA, and WHWP share a negative impact on the temperature in Pakistan. In autumn, AMO, AO, PNA, TNA, NP, WP, and WHWP have a positive impact on temperature variations; only WP shares a significant correlation, indicating of its profound impact on Pakistan's temperature. AMO, PNA, TNA, NP, WP, and WHWP share positive influences in annual time scales (figures 7 and 8). Here, AMO and TNA show significant association with higher magnitudes across the country (figure 8), indicating the influence of these teleconnections on temperature changes over Pakistan.

Discussion
The present study quantifies the observed temperature changes over homogenous climatic regions of Pakistan during 1960-2020 and their association with global teleconnections, i.e. AMO, AO, NAO, ENSO, PDO, PNA, TNA, NP, WP, and WHWP. These teleconnections influence the global climate system from interannual and decadal to interdecadal scales. In light of the published literature, there is still  more understanding needed on how observed temperature changes in relation to global teleconnections prevail over Pakistan. Moreover, it is crucial to comprehend the relevance of the causes in relation to global teleconnections in Pakistan and help the policymakers in formulating policy guidelines to prevent and minimize the long-term negative impacts. Therefore, we assessed the influences of these teleconnections on monthly, seasonal, annual and on decadal timescales. For this, the Pearson correlation in tandem with wavelet analyses (both WTC and MWC) are used for comprehensive and systematic analysis of regional temperature change and its global teleconnections.
Results show that temperature in Pakistan has increased at 0.23 • C dec −1 in the last 62 years, and this increase is more evident in R-IV, V, and VI (arid and semi-arid regions) in comparison to the elevated humid regions (R-II and I). Higher temperature increases are observed in spring, followed by summer and autumn in Pakistan. The highest temperature enhancement in spring affects significantly the hydrology as the dry season may onsets sooner, and in turn, the water demand increases. Available studies on climate variability in Pakistan are mostly in relation to the Pacific variability, like Saleem et al (2021), Rashid et al (2022), Ullah et al (2022b) and Karim et al (2023), whereas Del Rio et al (2013 and Hussain et al (2021) focused on NAO and other indices. The atmospheric circulation at sea level pressure and 500 hPa has a significant relationship with January temperature in Pakistan. The tropical Atlantic Ocean, northern Indian Ocean, Arabian Sea, western tropical Pacific basin, and warm pool influence the regional temperature changes. Such changes are observed at interannual to interdecadal scales; however, interannual variations prevailed more in comparison to interdecadal scales (Ahmad et al 2014). On intra-seasonal and interannual time scales, the Madden-Julian oscillation and ENSO have a significant impact on the climate of Pakistan (Hoell et al 2015). Among all indices, NP, WP, and WHWP show strong interannual teleconnections with the country's temperature while WHWP is most effective on interdecadal variability of temperature (figure 5). In general, various teleconnections have a wide and considerableimpact on temperature variations; however, the observed interactions suggest that the combined influence of teleconnections, particularly the triple combinations, can better explain regional temperature changes in Pakistan. In this regard, the observed coupled teleconnections with dominant influence in Pakistan are NP +WP + WHWP and Pacific-based indices of ENSO + PDO + PNA, as revealed by interannual to multidecadal coherences ( figure 6, table 3). Temperature changes demonstrated interannual coherence at 8-16 month bands with ENSO, NAO, IOD and PDO in northern Pakistan (Hussain et al 2021). Furthermore, Iqbal et al (2016) observed strong teleconnection patterns with temperature changes during pre-monsoon (March, April, and May). Higher positive relationship prevailed during winter and in March with the NAO. Whereas, the ENSO remained prominent during April and May, while the North Sea Caspian (NCP) Pattern prevailed influence with the maximum temperature in May, and with minimum temperature in the post-monsoon. According to Del Rio et al (2013), higher coherence between mean temperatures and teleconnection patterns with NAO, ENSO, and NCP, respectively, were seen in March, April, and May. The NAO may have an impact on monsoon season temperatures, especially in August. NAO and NCP may be able to regulate temperatures at the seasonal level during the pre-monsoon season.
It is essential to comprehend how oceanic and atmospheric processes interact and how that affects regional climate in order to build strong measurements that may aid institutional efforts to advance policy solutions (Hussain et al 2023a). Previous studies have highlighted that, the NAO, ENSO, SST, and multivariate El Nino-Southern Oscillation Index are the main causes of seasonal temperature variations in Pakistan. Whereas, the regional land use/cover change dynamics greatly influences various climatic factors (i.e. temperature, precipitation, and humidity) (Ishaq et al 2019, Akhtar et al 2020. Soil moisture is a crucial factor in climatic variability, due to its restrictions on evapotranspiration, surface energy fluxes, and diabatic heating, which preserves a land-ocean thermal gradient . The temperature increase is more evident after the mid-1990s (Riaz et al 2021), and such increase is more prominent in R-II, III, and V. However, temperature variations are more noticeable with global teleconnections with significant increases and wider coherences in the extremely arid, arid, and semiarid regions of southern Pakistan in contrast to the high-elevation humid regions (i.e. Region I and II). El Niño, the warm phase of ENSO, is associated with enhanced precipitation over South Asia, whereas La Niña, the cold phase, is associated with suppressed precipitation and drought (Barlow et al 2021). For instance, Rebi et al (2023) and Nawaz et al (2023) observed dominant interannual precipitation coherences in northern Pakistan. The population in southern Pakistan is highly dependent on agriculture to meet basic livelihood needs. The population will suffer more, if the current coherence patterns of interannual, decadal, and interdecadal patterns continue, which has shown wide spatial coherences with various global teleconnections and GASTA. Based on the study findings, the mid-latitude, semiarid, and tropical regions, where the vast majority of the country's population exists, are likely to suffer from extreme drought risks, and they should be careful under the future warming climate. The observed findings indicate that population in central and southern Pakistan is anticipated to experience more low-frequency climatic variations (interannual and decadal) compared to the preceding decade. This study focused on the global teleconnections and their relation with temperature changes in Pakistan. However, the processes are much more complex related to the temperature changes, so further studies are required to assess such changes large-scale atmospheric circulations, wind anomalies, anti-cyclonic circulation and moisture divergence.
The social infrastructure is becoming increasingly susceptible to climatic extremes (i.e. heat waves, floods, and droughts), have varied effects on the economy, society, ecology, and the environment as a result of regional and global climate change . As temperatures rise, more people contract infectious diseases like dengue and malaria in Pakistan. For instance, slight temperature increases may promote mosquito reproduction and affect human fever. According to Saeed et al (2022), the dengue transmission suitable days (DTSD), will expand throughout Pakistan, especially in places where there have never been dengue illnesses before. Climate change, however, is expected to result in a future decline in the DTSD in hotspot cities. Additionally, the region experiences a temporal change between the post-and pre-monsoon season, when freshwater offers ideal breeding conditions for dengue mosquitos. Climate change can have an immediate effect on human health, by modifying exposure to unfavorable outside temperatures (Gasparrini et al 2017). In general, high population density and fast urbanization are the major causes of the rising temperature events throughout South Asia in general and Pakistan in particular (Hussain et al 2023b). Therefore, in order to create a long-term climate adaptation plan in response to a changing climate, policymakers may find the research's findings useful. From the above finding of the coherence of large-scale climate variability, and its control in the temperature changes over the homogenous climatic regions of Pakistan during 1960-2020. We strongly recommend future research studies to explore the spatial and temporal dynamics, various circulations patterns, and their influence on seasonal and annual temperature in Pakistan.

Conclusion
The present study investigated the observed temperature variations over homogenous climatic regions of Pakistan between 1960 and 2020, and their associations with GASTA and global teleconnections at interannual, decadal, and inter-decadal timescales. The main findings from this study are as follows: 1. The mean temperature in winter has increased significantly in R-III, IV, and VI, but decreased in summer over regions R-I, III, IV, and in the whole country. Moreover, a significant increase prevails in R-III, IV, V, VI, and the whole country for annual and autumn. 2. Temperature increases in Pakistan at the rate of 0.23 • C yr −1 in comparison to the increase in GASTA (0.22 • C yr −1 ). The sporadic significant modes indicate that temperature increased in Pakistan more after the mid-1990s and doubled after 2015, and such variations are more noticeable in regions R-II, III, and V. 3. Temperature coherences remained sporadic and significant ranging from 8 to 128 months, while a significant inter-decadal oscillation of ∼200 months is evident in WHWP. Monthly coherence analysis shows a higher contribution of NP, WP, and WHWP with higher mean significant coherences as compared to the remaining indices. 4. The combined coherence patterns of the teleconnections prevail higher mean significant associations with WP, WHWP, ENSO + PDO, NP + WP + WHWP, Pacific, and Atlantic Oceans, and the combined coherences are higher in comparison to the single indices. 5. All the indices showed positive influences on temperature variations over Pakistan in winter, except AO, while AMO and TNA influenced temperature variations in spring. Moreover, the AO, NAO, NP, and WP positively influenced in summer, whereas the AMO, PNA, TNA, NP, WP, and WHWP shared positive influences on annual time scales.

Data availability statement
The temperature data used in this study can be officially acquire from the Pakistan Meteorological Department.