In the current study, two types of datasets are used, i.e., observed and reanalysis for 1981–2018 based on the in-situ precipitation data available from the Pakistan Meteorological Department (PMD). The observed precipitation data of fifty (50) meteorological stations for the monsoon season (June, July, August, and September months) are obtained from the Pakistan Meteorological Department (PMD). The station density of the current study is shown in Figure 1B, representing the dense network in the north and relatively sparse density in the south. The dominant precipitation patterns of the country have a relatively higher deviation in the north than in the south and are better represented by the prevailing station density (Hanif et al., 2013; Bhatti et al., 2020). Figure 1B further shows the selected four regions for an area average trend analysis based on the individual differences in land cover, elevation, and climate class. The rain-gauge data are indeed subjected to deviations attributed to climatic and non-climatic factors, which are removed from the primary data before application (Toreti et al., 2011; Tapiador et al., 2012; Kidd et al., 2017; Tapiador et al., 2018). In addition, the atmospheric reanalysis datasets were acquired from the European Centre for Medium-Range Weather Forecasts (ECMWF) and the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) data archives.

The ECMWF 5th generation atmospheric reanalysis (ERA5) of zonal and meridional wind components, relative humidity, air temperature, sea surface temperature, and geopotential height were used (Hersbach et al., 2020). The horizontal resolution of ERA5 is 31 km and has 137 vertical levels to 0.01 hPa (Dee et al., 2011). The advantages of the ERA5 include its improvements in precipitation and cloud schemes, model parametrizations, snow schemes, and improved radiative balance of the model (Shen et al., 2021; 2022b). For the current study, monthly mean data of the above variables were downloaded for the monsoon season with a grid size of 0.75°. The ERA5 atmospheric reanalysis data were downloaded from the following link: https://cds.climate.copernicus.eu/cdsapp#!/home.

The current study also used the NCEP/NCAR monthly mean outgoing longwave radiation (OLR) data (Liebmann and Smith, 1996). The resolution of radiance observation of OLR is 1.0 × 1.0°, derived from a high-resolution infrared radiation sounder (HIRS). The data was interpolated at a 2.5-degree spatial resolution of the initial National Oceanic and Atmospheric Administration (NOAA) data. The interpolated data provide a smooth and error-free overview of the tropical convective activities due to consistent quality control and conditional checks before interpolation. The OLR data used in this study were retrieved from https://psl.noaa.gov/.

(a) Extreme precipitation indices

For extreme precipitation, a set of seven precipitation indices were derived from the Experts Team on Climate Change Detection Indices (ETCCDI) of the World Meteorological Organization (WMO) (Zhang et al., 2011). Table 1 shows a detailed overview of the selected indices and their classification for the current study. Using these indices, we have considered two aspects of the precipitation extremes, i.e., the frequency of the days and their relative intensity for the monsoon months. It should be noted that the precipitation variability over Pakistan significantly varies from the east to the west, especially during the monsoon season (Hanif et al., 2013). Therefore, these indices are useful in characterizing monsoon precipitation variability in arid, semiarid, and humid climates. The choice of multiple absolute and percentile-based relative threshold indices could provide a tangible way of assessing trends in regional heavy and very heavy precipitation extremes.

(b) Trend analysis

For inferring trends in the extreme precipitation indices, a suite of non-parametric statistical techniques, namely, modified Mann-Kendall (mMK) (Hamed and Rao, 1998) and Sen’s slope estimator (SSE) were used (Sen, 1968). The mMK test was used to determine the significance of the trend with a 95% significance bound, and SSE was employed to estimate the magnitude of the monotonic trend (Ullah et al., 2019c; Ali G. et al., 2020). Both the original Mann-Kendall (MK) and its modified version (mMK) tests are often used to detect the significance of the trend in a time series; the MK test is sensitive to autocorrelation (Ullah et al., 2018b; Khan et al., 2021), which may affect its outcomes regarding significance. On the other hand, the mMK effectively nullifies the effect of autocorrelation before determining statistical significance (Salman et al., 2017; Xie et al., 2020; Hussain et al., 2022). Therefore, the present study preferred the mMK test over the MK to overcome autocorrelation in the datasets. Moreover, the mMK test is simple and robust against outliers, missing values, and normal distribution and is less sensitive to abrupt breaks in time series.

Similarly, the Sequential Mann-Kendall (SQMK) test (Sneyers, 1990) was further used to detect the abrupt changes (mutations) in the temporal trends of the sub-seasonal extreme precipitation indices. In addition, linear regression analyses were used to estimate the temporal trend across time and the elevation during the study period. More details about mMK, SSE, SQMK, and linear regression tests can be found in recent studies (Ullah et al., 2018a; Hussain et al., 2022). Finally, a non-parametric Probability Density Function (PDF) was employed to describe the distribution pattern of the extreme precipitation indices, allowing for the changes in the input to shape the distribution pattern and moments based on the mean, mode, median, kurtosis, and skewness. A Gaussian kernel with zero mean and unit standard deviation represents a normal distribution for a simple PDF. In contrast, a deviation on either the left or right side suggests a decrease or increase in the respective field’s magnitude, frequency, and/or intensity (McColl et al., 2017; Ullah et al., 2021b).

(c) Extreme precipitation indices relationship with atmospheric circulations

We further attempted to explore the extreme precipitation indices’ relationship with the atmospheric circulation strength in the region following previous studies (Rasmussen et al., 2015; Ullah et al., 2021b). To find such a relationship, we used the South Asian High (SAH) oscillations and the long-term interannual scale difference in the intensity of the atmospheric processes. The SAH oscillation and regional deviations were previously linked with Asian monsoon precipitation variability and changes (Ge et al., 2018; Zhang et al., 2020). The SAH index was calculated from the meridional shift of the 200 hPa geopotential height difference between the southwest (20—27.5°S, 50—80°W) and northwest (27.5–35°N, 50—80°E), as shown in a recent study (Wei et al., 2015). We calculated the relative mean difference by dividing the study period into two phases to estimate the long-term differences in the atmospheric processes, including divergent circulations (200 hPa), outgoing longwave radiation, relative humidity (850 hPa), and atmospheric temperature (850 hPa). The first phase spans from 1981 to 1990, and the second spans from 2009 to 2018 for the monsoon season. The first phase is then averaged and subtracted from the second phase. The purpose of this relative difference between the 2 decades relates to global warming and its impact on the circulation intensity of the region. Unlike a composite analysis of the selected extreme events, the relative difference may better state the probabilistic changes and relative differences in the atmospheric states.

Figure 3 shows the spatial trend of extreme precipitation indices derived from the sub-seasonal daily precipitation during 1981–2018. The stations with the significant trend at 95% confidence bounds are encircled. The frequency of wet days (R1 mm) showed a general increase in the country’s central, southern, and northern parts with a z-score value of >1.5 (Figure 3A). The core monsoon and adjacent mountainous regions experienced an erratic trend, with few stations showing a decrease (z-score = −5) and an increase in the mountainous parts. For the frequency of heavy precipitation days (R10 mm), a consistently increasing trend was explicit in the monsoon domain of the region (Figure 3B). The magnitude of the increasing trend was relatively high in the central and eastern parts (>2.5), which are considered key monsoon areas. On the other hand, a decrease in the frequency of heavy precipitation days (R10 mm) was evident in the northern mountainous (z-score <–2) (Figure 3B) and arid southwestern climate (z-score = −1.49).

As seen in Figure 3C, the frequency of very heavy precipitation days (R20 mm) increased in the eastern parts of the region (z-score > 4) and was erratic in the rest of the country with a moderate decrease of < –1.80 notable in northern and southern regions. The increase is relatively high in the eastern parts, followed by the northwestern region (z-score: 3.5). The intensity of wet days average precipitation (SDII) showed an increase in the central monsoon region (z-score= >2), including the eastern and central regions of the country (Figure 3D). In the rest of the country, the SDII trend is erratic, with a general increase in the monsoon regions and a decrease in the arid western stations. The maximum 1-day precipitation (Rx1 day) trend is consistent with SDII in the core monsoon regions and erratic in the rest of the country (Figure 3E). The country’s eastern and western foothills regions have shown an increasing trend (z-score = 2.50) and a decrease in the arid southern region (z-score = −0.69).

The intensity of the very wet days (R95p TOT) has decreased in the central arid plains, increasing in the Indus Basin eastern parts and southern coastal regions with a z-score of 1.5 (Figure 3F). The decrease in the intensity of very wet days is smaller, mostly in the country’s central arid regions, with a z-score of −1.49. However, the increase is dominant in the central and western foothills of the country, with a z-score of >1.5. The intensity of extremely wet days (R99p TOT) precipitation (Figure 3G) is generally erratic but shows an increase in the monsoon regions. The eastern, central, and western foothills increase is 0.71, stretching towards the southwestern and coastal regions.

Figure 4 shows the SSE of the extreme precipitation frequency and intensity from 1981 to 2018. As seen in Figure 4A, the frequency of wet days (R1 mm) increased in the eastern, northern, and northwestern parts (0.20 days), whereas a moderate decrease in the south and southwest is obvious (<–0.19 days). For heavy precipitation days (R10 mm) (Figure 4B), the increase is concentrated in the central monsoon region (>0.11 days). These parts are the main monsoon domain where the average increase amounts to >0.25 days. The frequency of very heavy precipitation days (R20 mm) has a moderate increase in northwestern parts (>0.13 days) and a moderate decrease (−0.02 days) in northern and southern parts of the country (Figure 4C). The mean precipitation on a wet day (SDII) (Figure 4D) exhibited an increase in the eastern, western parts of the region (>0.2 mm), alongside a decrease in the north and southeastern coastal regions (−0.14 mm). The maximum 1-day precipitation (Rx1 day) (Figure 4E) precipitation is erratic in the north and southwest (−0.27—0.17 mm) and increased in the eastern, northwestern, and central parts of the region (0.52 mm). The very wet days (R95p TOT) precipitation (Figure 4F) showed an obvious increase in the eastern and central parts of the region (0.11 mm), whereas the rest of the country, including the northern, western, and southern parts, showed a consistent decrease (−0.14 mm). The intensity of extremely wet days (R99p TOT) precipitation (Figure 4G) has a small decrease and thus can be referred to as unchanged during the study period. More detailed analysis with long time series of the precipitation data can better predict this trend.

The mMK and SSE of regional precipitation averaged for the selected stations in each region (Figure 1B) are further shown in Table 2. The regional average precipitation trend can better capture the regional precipitation behavior concerning the erratic nature of the spatial trend. The regional relative highest trend values for mMK and SSE magnitude are shown in bold. The relative increase in the wet days (R1 mm) showed an overall increase in all regions, with the highest values of mMK (3.71) and SSE (0.3) in region 2. For heavy precipitation days (R10 mm), the increase was moderate in all regions, with a consistent increase in region 2 (mMK: 2.92, SSE: 0.12) and a decrease in region 4 (mMK: –0.062, SSE: 0). For very heavy precipitation days (R20 mm), a similar increase in region 2 (mMK: 3.38, SSE: 0.078) was obvious, with a moderate increase in the rest of the regions. For SDII, a decrease in regions 1 and 4 (mMK: –1.45, −1.95, SSE: –0.035; −0.16) was obvious, and an increase in region 3. The 1-day maximum precipitation (Rx1 day) decreased in region 2 and 4, with a moderate increase in region 3 (mMK: 1.39, SSE: 2.32). For very wet days (R95p TOT), a decrease in regions 1 and 4 was obvious, along with an increase in region 2 (mMK: 2.46, SSE: 2.32). The extremely wet days (R99p TOT) precipitation has a moderate increase in all regions with a decrease in region 1.

Figure 5 shows the abrupt changes (mutations) in extreme precipitation indices’ temporal trends over Pakistan from 1981 to 2018. The UB statistics (blue dashed line) show the backward trend from 2018 towards 1981, while the UF statistics (orange line) show the forward trend (1981–2018); the green line shows the 95% significance limit for the trend in a time series. The points where UB and UF lines cross each other represent the abrupt shift in trend from decreasing to increasing state and vice versa. For the frequency of wet days (R1 mm) (Figure 5A), a constantly increasing trend is evident from 1981 with multiple mutations from 1990 to 2000. The catastrophic flooding and drought events in Pakistan are depicted in mutation points during the study period (Xie et al., 2016; Abbas et al., 2021). These events limited access to food and water, damaged the basic infrastructure, and affected the socioeconomic and sustainable development of the country (Rahman and Khan, 2011; Ullah et al., 2021b).

For the frequency of heavy precipitation (R10 mm) (Figure 5B), a consistent periodicity with multiple turning points inferred a continuous change in the trend. The drought periods of 1988–1990 and 1997–2002 and the flooding events of 1990 and 2010 showed a phase shift in the trend. For the frequency of very heavy precipitation days (R20 mm) (Figure 5C), three turning points were explicitly evident in early 1990, 2000, and 2015, inferring a shift in the frequency trend. The drought period of the late-1990 was very obvious, followed by a smooth increase afterward. The intensity of wet days mean precipitation (SDII) (Figure 5D) has different statistics and dynamical patterns evident from the respective frequency plot with a significant increase during 1985–1995, followed by a sharp decline during early 2000. The 1-day maximum (Rx1 day) precipitation (Figure 5E) showed the dry and wet episodes of the region with dynamic mutations in the early 1990s and a moderate increase in post-2010 with a decrease afterward. The intensity of the very wet days (R95p TOT) precipitation (Figure 5F) has shown four turning points in the trend linked with changes in the precipitation intensity. The mutation points reflect the major precipitation extremes, such as the floods in the early-1990s, the drought in 1997–2002, and the 2010 floods. The extremely wet days (R99p TOT) precipitation (Figure 5G) showed a moderate to very lower variation over the study period and rather dynamic mutations with no obvious trend visible. In conclusion, Figure 5 shows that the precipitation variability over Pakistan is moderately associated with and influenced by the large-scale atmospheric and oceanic forcing as previously explored. The forcing appears to influence the interannual monsoon precipitation extremes over Pakistan during their transition from one phase to another, such as during the 2010 flooding and drought phases (Webster et al., 2011; Xie et al., 2016).

Figure 6 shows the relationship between extreme precipitation frequency and intensity with the altitude of rain gauges. The mMK values for the frequency and intensity of the selected absolute and percentile-based precipitation indices are regressed against the elevation of the stations. The mMK values for the frequency (Figure 6A) of wet days (R1 mm) showed a decreasing trend with elevation with regression coefficients of −0.00039. The mMK values of heavy precipitation days (R10 mm) showed a similar decreasing trend with regression coefficients of −0.0011 (Figure 6B), with a similar decrease for very heavy precipitation days (R20 mm), with a regression coefficient of −0.00085 (Figure 6C), respectively. The mean precipitation on wet days (SDII) showed an overall decreasing trend with altitude (−0.00054); however, the trend was erratic in the lower and moderate altitudes (Figure 6D). The 1-day maximum precipitation (Rx1 day) also showed a decrease across altitude (−0.00045), but the station density in lower and moderate elevations showed an increase in the precipitation (Figure 6E). The very wet days (R95p TOT) intensity (Figure 6F) has decreased across altitude with regression coefficients of −0.00028, followed by a similar decrease for extremely wet days (R99p TOT) intensity with the highest decreasing trend compared to the intensity-based indices (Figure 6G).

In conclusion, Figure 6 shows that the frequency and intensity of precipitation extremes concerning altitude exhibited a decreasing trend during the monsoon season. The relationship showed that some stations have positive and negative residues in the lower and higher altitude regions, inferring that precipitation increases with altitude in the moderate altitudes. Still, the overall relationship is negative due to the higher altitude and density of the stations in the northern parts of the study region. The possible reason for such a weaker and negative relationship could be that the extreme monsoon precipitation is centered in the country’s central regions, mostly in the middle and lower altitude classes. Furthermore, the extreme precipitation indices have a diverse regional variability, with each regional index trend varying from the rest. The results also agree with the previous studies showing a decrease in monsoon mean precipitation and an increase in wet days across moderate and lower altitudes (Ullah et al., 2018a; Bhatti et al., 2020).

Figure 7 shows the probability density function (PDF) of the extreme precipitation frequency, intensity, and its moments of distribution (Table 3). The wet days (R1 mm) (Figure 7A) PDF shows an overall positive asymmetric distribution in monsoon months, indicating an inconsistent increase. This indicates that the onset and recession of the monsoon season are getting vibrant, while the mid-monsoon season is getting intense with more precipitation days. As seen in Table 3, the wet days’ precipitation (R1 mm) has shown a higher deviation towards maxima during August (4.54), followed by July (1.44), September (1.01), and June (0.94). Similarly, July and August have the highest mean wet days’ precipitation at 5.59 and 5.42, respectively.

The heavy precipitation days (R10 mm) (Figure 7B) exhibited a similar pattern to wet days (R1 mm), highlighting the high frequency (intensity) of heavy precipitation events (R10 mm). In terms of distribution statistics (Table 3), the months of June and September (July and August) showed the highest skewness and kurtosis (deviation and mean) values, affirming higher frequency (intensity) of heavy precipitation days (R10 mm) during these months. The skewness and kurtosis values for heavy precipitation days (R10 mm) were high in September (1.45 and 5.24), followed by June (1.21 and 4.66), while the higher degree of deviation and mean was evident during July (0.77 and 2.43), followed by August (0.68 and 2.28), respectively.

For very heavy precipitation days (R20 mm) (Figure 7C), the PDF followed the pattern of heavy precipitation days (R10 mm); however, slight changes are observed in the density and distribution of the selected months. June and September have the highest frequency with slight rightward movement during September, indicating frequent heavy precipitation events. The higher skewness and kurtosis during June and September infer a shift towards more extreme values and a high frequency of very heavy precipitation events (R20 mm) during these months (Table 3). On the other hand, the larger distribution towards the right side in July and August is explicitly conclusive of a relatively higher degree of change in the intensity of very heavy precipitation events (R20 mm) during these months.

The maximum 1-day precipitation (Rx1 day) is similar to other absolute indices (Figure 7D), with high density and positive distribution during the selected months. However, June and September have the highest density, while July, August, and September have more positive shifts. The PDF pattern indicates that the study region was predominated by regular daily maximum precipitation extremes (Rx1day) during June and September. The PDF pattern was confirmed by the related statistics (Table 3), showing the highest skewness, kurtosis, and standard deviation during July (1.86, 8.37, and 11.08) and September (1.92, 6.32, and 11.85). The higher degree of skewness, kurtosis, and deviation suggests asymmetrical and variable daily maximum precipitation (Rx1 day) during these months.

In mean precipitation on wet days’ (SDII) PDF (Figure 7E), the highest density was observed in June with a slight asymmetrical shift in the high precipitation distribution, suggesting a high frequency during the study period. However, positive shifts with the symmetrical pattern were found in the rest of the months, indicating high uniformity in the magnitude of wet days’ mean precipitation (SDII). Table 3 shows high skewness and kurtosis values were observed in June (1.26 and 4.85), indicating recurrent wet-day precipitation (SDII).

For very wet days’ precipitation (R95p TOT) (Figure 7F), an opposite pattern is found in the density and distribution of the selected months relative to the absolute indices. Interestingly, July and August (June and September) show a relatively significant increase in density (distribution), which contradicts the PDF patterns of absolute precipitation indices. These results suggested the frequent and intense occurrences of very wet events (R95p TOT) in Pakistan during July and June were relatively predominated by very wet day extremes (R95p TOT). In terms of PDF statistics (Table 3), the highest skewness and kurtosis values were found in July (1.82 and 8.96), which confirmed the occurrence of frequent and intense very wet extremes (R95p TOT).

The extremely wet days’ precipitation (R99p TOT) PDF followed the very wet days’ PDF pattern with slight variations in the density and distribution of the selected months (Figure 7G). The higher intensity of extremely wet days (R99p TOT) in the July and August months could be attributed to more precipitation, as these months are mid of monsoon season in Pakistan. In terms of PDF statistics (Table 3), the skewness and kurtosis values were high during July (1.19 and 4.99) and September (1.21 and 4.55), whereas the degree of deviation was high during June, followed by September and July at the rate of 28.82, 27.47, and 21.94, respectively.

In conclusion, from Figure 7, an obvious overall increase is observed in the frequency and intensity of extreme precipitation indices. The absolute indices have observed a relatively high frequency during June and September, while a strong intensity during July and September. In contrast, the percentile-based indices have shown an opposite pattern with high frequency in June and September and intensity in June and September. The high frequency of extreme precipitation indices during June and September indicates that the onset and recession of the monsoon season have been vibrant in Pakistan during the study period. Moreover, the increasing intensity of extreme precipitation indices during July and August suggests that the study region has received more precipitation during these 2 months and could be considered the peak monsoon months in Pakistan. Generally, the extreme precipitation increase in intensity is reported in the previous study over Pakistan (Bhatti et al., 2020).

This section has attempted to show a potential mechanism of the changes in the extreme precipitation indices and their association with the large-scale atmospheric phenomenon. For this purpose, we have used a two-step approach, including a long-term correlation of the extreme precipitation indices with South Asian High (SAH) meridional oscillation in the first step and a relative difference of the circulation intensity during the study period in the second step.

Figure 8 shows the correlation between the South Asian High (SAH) index and the extreme precipitation frequency and intensity during the study period. For wet days (R1 mm) precipitation, a positive correlation (>0.45) was visible in core monsoon regions and negative in the central parts of the country (Figure 8A). A consistent positive correlation in the northern and southern and a negative correlation in the central parts of the country was obvious for heavy and very heavy precipitation days (Figures 8 B, C). The relative strength of the correlation was stronger for very heavy precipitation days (>0.60) among the frequency-based indices (Figure 8C). The mean wet days’ precipitation (SDII) showed a consistent positive correlation in the north and southern parts alongside a negative correlation in the central parts of the region (Figure 8D). For maximum 1-day precipitation (Figure 8E), the strength of the correlation was positive in the north, and a rather weaker to moderate negative relationship was obvious in the rest of the country. The relatively strong relationship among the intensity of very wet days (R95p TOT) (Figure 8F) across the core monsoon regions of the country, whereas the extremely wet days (R99p TOT) precipitation (Figure 8G), was rather uniform across the eastern parts of the country from the north to southward. In conclusion, the SAH meridional oscillation across the northward and southward of the study region exhibited an erratic relationship with the frequency and intensity of the extreme. The correlation strength varies regionally among the indices, indicating that the large-scale shift toward northwestward may suppress the monsoon precipitation over Pakistan and vice versa (Wei et al., 2015).

Figure 9 shows the relative difference of (a) 200, (b) 500, and (c) 850 hPa wind components (vectors) and stream function (shaded) for the first (1981–1990) and last decade (2009–2018). From Figure 9A, the wind components and stream function showed the relative strength of the circulations and the center of the cyclonic/anticyclonic motion. Two high-pressure systems are evident; the first is located over the Tibetan Plateau in the north of Pakistan, while the second is over northwestern Pakistan, centered over Eurasia. A low-pressure cut-off can be seen between the two pressure systems, separating them. The mid-tropospheric circulations (500 hPa) have preserved the probabilistic Rossby Wave pattern with the leading ridge and Tibetan High favoring intensified ascent over Pakistan (Figure 9B). A strong southwesterly flow (850 hPa) over the Arabian Sea (20°N, 50°E-70°E) is discernable that after entering the Bay of Bengal region forms a southeasterly flow, favoring enhanced moisture transport from the Arabian Sea and the Bay of Bengal (Figure 9C). The two pressure systems provided a frontal system where warm air masses from the Bay of Bengal and warm pool region through southwesterlies and cold Eurasian air masses merge and probably drive the extreme precipitation during the monsoon season. The mechanism shown here suggests an increase in the intensity of the circulation pattern and its associated impact on the regional precipitation extreme indices. Further detailed studies are thus needed for a more definitive mechanism assessment.

The divergent wind and velocity potential (Figure 10A) further indicated that a relatively intensified vertical ascent in the study region has been evident in the recent decade. The divergent circulations’ center has exhibited a strong divergent motion with vectors indicating a strong heat source over the SA landmasses and the Bay of Bengal. The difference between the two periods indicates that more precipitation is received in the region than in the first phase, leading to extra latent heat release and stronger vertical ascent. The heat source of the SA monsoon is located over the Tibetan Plateau and has been extensively studied for its thermal controls over the Asian monsoon and exhibited a strong correlation with the precipitation variability over Pakistan (Ullah et al., 2021a; 2022b; 2023). The velocity potential further indicates a consistently intensified ascent of the wind, descending into the Indian Ocean, coastal regions of Somalia, and East Africa.

The outgoing longwave radiation has been used as a proxy to indicate strong convective activities over the study region (Ullah et al., 2021b). The difference in the outgoing longwave radiation (Figure 10B) showed an intensification of the convective activities over Pakistan and the adjacent Indian Ocean. In addition, a decrease is evident in northwestern India, the Bay of Bengal, and the Arabian Sea region. The study regions’ intensified convection indicates increased vertical motion, condensation aloft, and enhanced precipitation extremes. The decrease (negative anomalies) in OLR indicates an increase in cloud cover due to an increase in the convection leading to extreme precipitation accompanied by other synoptic weather patterns during the monsoon (Lim et al., 2011). For relative humidity at 850 hPa (Figure 10C), an increase in the magnitude was obvious over Pakistan’s eastern and southeastern monsoon regions. The increase accounted for >8% of the mean magnitude from the mean of the first phase, inferring an increase in the region’s lower troposphere’s moisture profile. The northern parts have featured a decrease in the relative humidity’s magnitude. Such a decrease can be disregarded because of terrain-induced biases. Hence, no valid data exist at 850 hPa in those regions. The air temperature at 850 hPa over Pakistan (Figure 10D) indicated a decrease in air temperature’s absolute magnitude from the first phase. Concerning previous figures, the increase in relative humidity and strengthening of the monsoon trough over the study region indicated lower temperatures due to more water vapors in the air, causing cooling through evapotranspiration. The air temperature over the adjacent Indian Ocean was relatively higher, showing a contrast pattern. Such a pattern indicates that over the ocean increase in temperature will also accelerate the convection. The evaporated water vapors are generally transported to the continental regions, resulting in precipitation and cooling the air temperature.

In conclusion, convection and monsoon trough over the study region has intensified, indicating relatively higher precipitation in the country than in the first phase. The global warming-induced changes have shuffled the global water cycle as indicated by the IPCC that the wet regions or seasons will become wetter and warmer at the end of the century (IPCC, 2022). Similar findings are reported in the previous study on a global scale, indicating that the future precipitation in the Asian monsoon domain will be more violent (Trenberth, 2011; Greve et al., 2014; Singh et al., 2014) due to the intensification of the monsoon trough and enhanced monsoon activities over the region. The high-pressure systems’ (South Asian High) movement into the mid-latitude regions is further linked with intensified downstream precipitation patterns due to intensified wind ascent (Wei et al., 2015; Ullah et al., 2021b).