Spatial and temporal variability of future extreme precipitation in Bangladesh using CMIP6 models Open Access

Due to recent climate change and increased human activities, extreme climate events, including extreme rainfall, drought, and heat waves, have increased (Rahman et al. 2024). These extreme climate events, especially precipitation extremes, have significant socioeconomic impacts, often leading to devastating climatic phenomena that impede agricultural production and result in human casualties (AghaKouchak et al. 2020; Yu et al. 2023). The impacts of such extreme events are more profound in highly climate-vulnerable countries like Bangladesh. Bangladesh is highly vulnerable to climate change; it ranks 7th out of 170 countries on the global climate risk index (Imran et al. 2023). Bangladesh's subtropical monsoon climate, characterized by high summer temperatures, heavy rainfall, elevated humidity, cold winter days, and pronounced seasonal variations (Caesar et al. 2015), makes it especially susceptible to global warming and associated sea-level rise. These climate impacts affect the country's socioeconomic status each year (Islam et al. 2023; Rahman et al. 2023). Therefore, accurate measurement and prediction of climate extremes across Bangladesh are of paramount importance for gaining insights into rainfall processes, extreme events, and their evolving patterns for minimizing socioeconomic impact planning.

Climate extreme events refer to unusual, severe, and often destructive weather phenomena that deviate significantly from a region's average or expected weather conditions. These events can encompass many phenomena, including heavy rainfall, droughts, heatwaves, hurricanes, cyclones, floods, and more (Das et al. 2023; Islam et al. 2024). Extreme climate events can have devastating consequences for both natural ecosystems and human society. They are distinguished by their rarity and the extent to which they deviate from the usual climate patterns. Climate change is primarily responsible for the increase in extreme climate events (Akter et al. 2024). The consequences may include displacement of communities, damage to infrastructure, food and water scarcity, and increased health risks (Sillmann et al. 2017; Ren et al. 2018; AghaKouchak et al. 2020). Precipitation-extreme events are particularly hazardous in tropical areas and countries with agricultural economies. Agricultural activities in these regions depend heavily on predictable and consistent rainfall patterns; any deviation from this pattern can result in crop failure, food shortages, and economic hardship. Predictions of extreme precipitation events can assist communities and governments in taking preventative measures and mitigating their effects (Fabian et al. 2023; Yin et al. 2023; Zhi & Wang 2023).

Bangladesh is a low-lying deltaic country in South Asia, ranking among the most vulnerable countries to climate change. Its unique topography and geography make it particularly susceptible to the adverse effects of heavy rain and flooding caused by extreme precipitation events (Rahman et al. 2023). Due to its densely populated coastal regions, extensive river systems, and significant agriculture sector, Bangladesh is particularly susceptible to climate change as a result of changing precipitation patterns (Nowreen et al. 2015; Kamruzzaman et al. 2019; Rahman & Islam 2019; Abdullah et al. 2020; Khan et al. 2020; Das et al. 2022a; Islam et al. 2022). The climate of the country is highly influenced by global atmospheric circulations, including the El Niño-Southern Oscillation (ENSO), which significantly impacts the country's weather patterns (Rahman & Islam 2019). During El Niño years, Bangladesh tends to experience drier conditions, leading to potential droughts, particularly in the northwestern regions, where rainfall can decrease by up to 20% (Ehsan et al. 2023). Conversely, La Niña events are associated with increased rainfall, increasing the risk of flooding and landslides, especially in the southeastern and coastal regions. The influence of the Indian Ocean Dipole and the monsoonal winds further modulates the seasonal distribution of rainfall (Wahiduzzaman et al. 2022). Extreme precipitation events, such as prolonged droughts or heavy monsoon rains, severely affect agricultural productivity (Hossain et al. 2020; Ahmed & Khan 2023), which is the backbone of the country's economy. A prolonged period of heavy rainfall can cause flooding and waterlogging, which can result in damage to crops and infrastructure. Conversely, prolonged periods of drought may result in crop failures and food shortages, affecting both food security and the livelihoods of millions of farmers. In this regard, studying precipitation extremes in Bangladesh is crucial to developing effective strategies for climate adaptation and mitigating adverse consequences. Furthermore, it is critical to develop resilient infrastructure and early warning systems to reduce the economic and human costs associated with extreme precipitation events in Bangladesh.

General circulation models (GCMs) are complex tools used to study how the climate system responds to changes in radiative forcing and the underlying processes (Taylor et al. 2012; Sharma & Goyal 2020). Despite improvements over the past two decades in GCM simulations of past, present, and future climates, significant regional biases and deficiencies persist due to inadequate representation of crucial regional-scale processes, suboptimal parameterizations, flawed initial conditions, and coarse resolutions (Wehner et al. 2014; Van Der Wiel et al. 2016; Diallo et al. 2019). These modeling shortcomings, combined with Earth's intrinsic variability, contribute to uncertainties in global to regional climate change estimates. Modern GCMs play a crucial role in the Coupled Model Intercomparison Project (CMIP), a coordinated initiative addressing these issues (Tang et al. 2021). CMIP employs multiple models to understand better past, current, and future climate changes driven by natural variability and radiative forcing changes. The Intergovernmental Panel on Climate Change assessment reports also rely on CMIP models (IPCC 2022). The accuracy of GCMs in the simulating present and historical climates affects our confidence in their future climate predictions (Eyring et al. 2015). The latest state-of-the-art climate models in CMIP Phase 6 (CMIP6) offer new opportunities to assess Earth's response to radiative forcings in the 21st century. Evaluations of CMIP6 simulations, including regions like South Asia, Africa, and the Arabian Peninsula, reveal differences from earlier CMIP studies (Almazroui et al. 2020). CMIP6 models are generally improved versions of those in earlier phases, featuring enhanced cloud microphysics parameterizations, better representations of Earth system processes, and finer resolutions compared to CMIP5 GCMs (Eyring et al. 2015). Past climate simulation comparisons have demonstrated that CMIP6 models excel, particularly in Bangladesh (Kamruzzaman et al. 2021a).

Several studies have explored the impacts of climate change on extreme weather events in Bangladesh using earlier CMIP models (Nowreen et al. 2015; Kamruzzaman et al. 2019; 2021a; b; Das et al. 2022b; Ezaz et al. 2022). For instance, research based on CMIP5 models, such as that by Nowreen et al. (2015) and Kamruzzaman et al. (2019), identified significant trends toward increasing extreme precipitation events, particularly in the eastern regions of Bangladesh. These studies underscored the growing vulnerability of Bangladesh's monsoon-dominated climate to future climate change, with the potential for more intense and frequent flooding events. However, the coarser resolution and less sophisticated atmospheric representations of CMIP5 models limited these analyses. The advent of CMIP6 has brought substantial improvements, enabling more accurate simulations of regional climates and extreme events (Das et al. 2022b; Rimi et al. 2022). Recently, Akter et al. (2024) utilized the 13 CMIP6 models to assess extreme temperature in Bangladesh. Studies using CMIP6 models, such as those by Das et al. (2022b), Rahman et al. (2023), and Fahad et al. (2024), confirm the increasing trends in extreme precipitation but with greater spatial detail and improved reliability. These models predict significant increases in extreme rainfall, particularly under high-emission scenarios like SSP5-8.5, which pose severe risks for the eastern and coastal regions of Bangladesh.

Despite these advancements, there remains a paucity of research examining the spatial and temporal variability of future extreme precipitation and temperature in Bangladesh using CMIP6 models (Akter et al. 2024). Furthermore, there has yet to be a thorough exploration of integrating shared socioeconomic pathways (SSPs) into these projections, especially at regional and national scales where the impacts of climate change are most acute. To address this gap, the present study assesses the spatial and temporal variability of future extreme precipitation events in Bangladesh using 13 bias-corrected CMIP6 models. This study focuses on projecting changes in precipitation-extreme indices for the near future (2021–2060) and far future (2061–2100) under two SSP scenarios: SSP2-4.5 and SSP5-8.5. This study improves the accuracy of GCM projections by using a quantile-based bias correction method, which was first suggested by Heo et al. (2019) and then utilized by Kamruzzaman et al. (2021b).

In addition, the study utilizes a multimodel ensemble (MME) approach to address uncertainties in future climate projections, providing a more reliable assessment of extreme precipitation patterns (Sillmann et al. 2013). The novelty of this research lies in its comprehensive approach to analyzing future precipitation extremes using the latest CMIP6 models and SSPs, providing critical insights for policymakers and stakeholders involved in climate adaptation and disaster risk management. The findings will contribute to a better understanding of how climate change will alter the frequency and intensity of extreme precipitation events in Bangladesh, guiding the development of effective strategies to mitigate these impacts and enhance the country's climate resilience.

The country's climate is tropical, with temperatures never falling below 0 °C. Historical records show temperature ranges between 15 and 34 °C. The country receives an average of 2,200 mm of precipitation per year, with annual total rainfall ranging between 1,500 and 5,000 mm (Dewan et al. 2022). The northeastern region has the largest average annual rainfall (4,300 mm), whereas the northwest has a lower average yearly rainfall (1,400 mm) (Shahid 2010). Seven climatic zones can be distinguished within the country based on entire climatic condition: (a) the southeast, (b) the northeast, (c) the northern part of the northern region, (d) the northwestern region, (e) the western region, (f) the southwestern region, and (g) the south-central region. There are three distinct seasons in Bangladesh: a cool dry season (November–February), a premonsoon hot season (March–May), and a rainy monsoon season (June–October). Most rain falls between June and September, with very little falling between November and February (Fattah & Morshed 2022).

Daily annual precipitation of 13 CMIP6 GCMs was adopted to project extreme precipitation across Bangladesh as presented in Table 1. Mishra et al. (2020) evaluated the performance of 13 GCM-CMIP6 for climate simulations over South Asia and stated that these 13 GCMs can be applicable for climate projections for the South Asian countries including Bangladesh. This research utilized the models' initial run (r1i1p1f1) simulations for an unbiased evaluation at a daily time scale (Eyring et al. 2016). GCM data for historical (1985–2014) and SSPs scenarios (2021–2100) were downloaded from: https://esgf-node.llnl.gov/search/cmip6/.

The two scenarios used in the CMIP6 coupled shared socioeconomic pathways (SSPs) and target radiative forcing levels at the end of the 21st century (Gidden et al. 2019). For instance, SSP2-4.5 indicates SSP-2 and target radiative forcing at the end of the 21st century 4.5 Watt/m². Thus, SSP2-4.5 is a moderate mitigation scenario, whereas SSP5-8.5 is based on the high-emission scenario considering SSP-5 and radiative forcing of 8.5 Watt/m² at the end of the 21st century (O'Neill et al. 2014). We estimated the historical and two future-period extreme precipitation indices, namely, period 1 (near future) (2021–2060) and period 2 (far future) (2061–2100). For each model, a 30-year reference period (1985–2014) of total annual precipitation was used for the study. Utilizing data from the 1985–2014 timeframe mitigates the impact of such constraints since contemporary data collection and analytical techniques provide a more precise foundation for model adjustment. The 1985–2014 era offers a high-quality, representative, and adequately extensive dataset that corresponds well with climate model simulations, allowing precise bias correction pertinent to contemporary and future climate research. This study used bias-corrected gridded precipitation datasets. It is noted that the earth's surface divides into a grid, with each grid cell representing a specific geographic area. The size of the grid cells can vary depending on the resolution of the dataset. In this study, the grid resolutions were 1° × 1°. Station data (meteorological station) were not utilized for this purpose. However, station data were only collected to check the quality and accuracy of the gridded dataset. To replicate the distribution of mean climate as well as annual extremes derived from the model results, bias correction methods are effective tools in climate change studies (Bürger et al. 2012; Cannon et al. 2015).

In this study, 13 CMIP6 GCMs of daily precipitation were employed to assess the extreme precipitation for the baseline and two future timeframes over the country. GCMs were regridded to 1° × 1° spatial resolution, close to the mean resolution of all GCMs at 21 grid points, using the empirical quantile mapping (Hamed et al. 2021), to eliminate biases from various spatial resolutions (Mishra et al. 2020).

The dataset is preprocessed in several steps, including the collection of the meteorological sites and gridded datasets, quality control, downscaling of reanalysis dataset and bias correction for precipitation dataset, and finally merging of the meteorological sites and bias-corrected gridded data (Mishra et al. 2020). Further details about the 13 CMIP6 GCMS are described by Mishra et al. (2020). The bias-corrected datasets outperform other available precipitation products in terms of total precipitation and other indicators (Montes et al. 2021). Codes utilized for bias correction of 13 CMIP6 GCMs can be found through the Github link: https://github.com/udit1408/cmip6_downscaling.

GCM models have a large spatial resolution gap and methodological flaws that limit their use in climate change evaluation. This research downscaled 21 grid points GCM simulations statistically. First, the raw GCM simulations were interpolated to observed points. Then, a simple quantile mapping (SQM) approach was used to correct GCM simulation bias based on gridded data distribution at all points. SQM is a tool that has grown acceptance in atmospheric science because of its effort to bias-correct the whole exposure distribution. The main advantage of this SQM is that it improves climate forecasts by thoroughly resembling the observed distribution, minimizing biases, and increasing dependability (Heo et al. 2019). SQM-based temperature downscaling in Bangladesh is also reliable (Akter et al. 2024). Kamruzzaman et al. (2019) describe the study's bias correction technique.

Bias correction is a statistical technique to modify the datasets for systematic differences between model results and observations to reduce systematic errors (biases) and make them more accurate, especially in climate modeling that predicts the future. Some advanced techniques for fixing bias, like SQM with large datasets, can be complex and take time and processing power. Bias correction methods may be effective under typical settings but often fail to modify severe events or infrequent occurrences appropriately. The SQM technique does not assume the form of the bias probability function, ensuring that the distribution of corrected values mirrors that of the actual data (Bhattacharjee et al. 2025). The SQM technique fails to preserve spatial or intervariable relationships. Understanding its drawback is vital for confirming that the bias corrections increase rather than compromise the integrity of the dataset.

This research evaluates change in the frequency and intensity of extreme precipitation using eight indices suggested by the Expert Team on Climate Change Detection and Indices (ETCCDI) and Expert Team on Sector-Specific Climate Indices (ET-SCI) (Table 2). Eight extreme precipitation indices were used in this research because these indices contribute to the acceleration of research and detection of extreme weather and climate phenomena to rainstorms, floods and droughts (Alexander et al. 2006; Islam et al. 2021). The projected changes in the extreme precipitation for the near future (2021–2060) and far future (2061–2100) were compared to the historical period (1985–2014) to determine the spatial variability in the estimated changes in the extreme precipitation. All the eight indices are on an annual basis and are defined in Table 2. All these indices used in this study are calculated and interpreted based on annual averages. ClimPACT2package of R software was employed for the computation of eight indices. This software's package can be found at https://infoasis.shinyapps.io/climpact/w82bbdb27/userguide/Climpactuserguide.htm.

Furthermore, trend analysis was conducted on all derived indices using the Mann–Kendall (MK) test (Mann 1945; Kendall et al. 1975; Praveen et al. 2020), with statistical significance at the 5% level. Pearson's correlation matrix was employed to show the time–frequency transformation association between the large-scale atmospheric circulation and 13 precipitation indices in Bangladesh, based on the contributing factors influencing the climatic system in the country (Endo et al. 2015; Ahmed et al. 2017). The detailed methodology of the MK test can be found elsewhere (Rahman & Islam 2019; Islam et al. 2024).

The probability density function (PDF) method was used to assess the expected outcomes of possible values for the various indices assessed. This method is nonparametric and hence straightforward to apply (Kim et al. 2019). Moreover, it gives a snapshot of the distribution of the indices. The distribution curves of the selected models (models with the least biases) are compared to the observation for the indices to ascertain models' capability to reproduce the distribution. Recent studies have provided details on computing PDFs and related distributions (Yin & Sun 2018; Islam et al. 2022).

The bias correction approach was used to reduce these errors. This research used an SQM method to adjust for bias. The MME estimates of CMIP6 GCMs were adjusted using bias correction methodologies based on the SQM technique, resulting in a closer alignment between the monthly average data and the observed values (Figure 2). The biases in the precipitation in the bias-corrected MME data decreased to less than 1.6 mm per month over the whole year.

If the increasing trend of global emissions persists (SSP5-8.5), Bangladesh is likely to receive higher rainfall in the near future (2021–2060) compared to the PRTOT recorded between 1985 and 2014. By the late 21st century (2061–2100), precipitation levels are projected to increase further across the country, with the northeastern region receiving the highest amounts, exceeding 5,953 mm/year. Districts in the central region are expected to receive PRTOT ranging from 2,448 to 3,377 mm per year during 2061–2100. These scenarios suggest that due to the continued increase in global Greenhouse Gas (GHG) emissions, Bangladesh will face flood risks and related challenges, particularly in areas already prone to heavy rainfall.

Under SSP2-4.5, the western region is expected to see little change in precipitation intensity during the near future, with the southwestern region maintaining levels between 18 and 22 mm/day, while the northeastern region is projected to exceed 40 mm/day. In contrast, under SSP5-8.5, the near future is projected to see precipitation intensities ranging from 37 to 40 mm/day, slightly higher than the historical average of 1985–2014. The projected simple daily intensity index (SDII) is expected to increase from a range of 17–41 mm/day during the historical period to 18–44 mm/day in the near future, and further to 18–48 mm/day in the far future. Significant regional differences are evident, particularly under SSP5-8.5, where the far future sees an increase in SDII, peaking at 39 mm/day, compared to the historical low of 20 mm/day. The regional differences in SDII under SSP5-8.5 are substantial, with a variance of 49 mm in the near future and 66 mm in the far future. The southwest region, however, is projected to experience a gradual decline in SDII, with more pronounced changes expected in the near future compared to the far future. Overall, the geographical distribution of SDII across Bangladesh is expected to become more homogeneous in the future, with increases in the eastern region offsetting decreases in other regions. This shift suggests a convergence in precipitation intensity patterns, driven by significant changes in climate dynamics across the country.

Projections suggest an overall rise in CWD across the country, with the most significant increase expected in the near future under the SSP2-4.5 pathway. Between 2021 and 2060, under SSP2-4.5, CWD across the district of Chittagong division and Satkhira district is expected to increase from 25–32 days to 34–38 days, while in the northeast region, it is projected to rise from 15–16 days to 24–34 days. By the late 21st century, CWD in the northwest region is anticipated to range from 24 to 32 days, while in Satkhira, Chittagong, Cox's Bazar, Bandarban, and Khagrachari districts, it is expected to be between 32 and 38 days. If the increasing trend of global GHG emissions persists (i.e., SSP5-8.5 persists), CWD across the country is projected to increase during 2021–2060 but may see a slight decrease in the subsequent decades. The estimated range for CWD under SSP5-8.5 is 15–36 days in the near future and 14–35 days in the far future. These projections suggest that while CWD will generally rise, there may be some regional variations and slight decreases over time depending on emission pathways.

Future projections under SSP2-4.5 show an overall increase in CDD across the country. In the near future (2021–2060), CDD is expected to range from 48 to 96 days, increasing further to 48–105 days in the far future. The northwest region, in particular, is projected to experience CDD exceeding 71 days per year during 2021–2060, rising over 81 days by 2061–2100. Significant spatial variations are predicted in the southern coastal districts under SSP2-4.5. Under SSP5-8.5, the northern region is projected to experience higher CDD compared to the southern region, with maximum durations exceeding 88 days per year in the Lalmonirhat and Nilphamari districts, and the shortest durations (55–62 days) in the Satkhira, Chittagong, Cox's Bazar, Cumilla, and Bandarban districts during 2021–2060. However, in the far future, the duration of average dry spells will reduce to 41–85 days. Coastal districts in Bangladesh are expected to experience the most significant decrease in dry spells, with CDD declining from 55 to 67 days during 2021–2060 to 41–53 days during 2061–2100.

Projections indicate an increasing trend in R95p under both SSP2-4.5 and SSP5-8.5, with the most significant increases expected in the near future timeframe under SSP2-4.5 and in the far future timeframe under SSP5-8.5. Projections under SSP2-4.5 suggest that the lowest R95p values will shift to the western region, and the highest values will remain concentrated in the eastern districts in both timeframes. This indicates that the western region of Bangladesh will likely receive less rainfall and experience a higher risk of drought. Under SSP2-4.5, Dhaka, the capital city, is projected to receive between 356 and 428 mm of rainfall on very wet days in the near future and between 350 and 379 mm in the far future. In contrast, under SSP5-8.5, Dhaka is expected to receive between 257 and 280 mm of rainfall in the near future, rising dramatically to between 586 and 785 mm of rainfall in the far future period. Additionally, R95p precipitation is projected to decrease in the northwest (244 to 277 mm) while increasing significantly in the southeast (554 to 829 mm). The highest projected R95p value in the far future under SSP5-8.5 is 1,643 mm, marking a substantial increase from the historical low of 44 mm. The future scenario under SSP5-8.5 is expected to exhibit significant regional variations, with differences reaching up to 1,449 mm, highlighting the growing disparities in precipitation patterns across the country.

The precipitation intensity trend analysis in Figure 11(b) shows that the change rate ranged between −2.86 and 2.94 mm/day under historical and both projected timeframes. The change rate of CDD varied from −1.84 to 2.47 day/year (Figure 11(c)), CWD from −2.84 to 2.66 day/year (Figure 11(d)), R95p from −0.9 to 3.32 mm/year (Figure 12(e)), R20 from −2.42 to 2.97 day/year (Figure 12(f)), Rx1day from −2.38 to 3.57 mm/year (Figure 12(g)), and Rx5day from −2.06 to 3.34 mm/year (Figure 12(h)) under historical and both SSP timeframes. Increasing trends during the historical timeframe were observed in 83% of the sites for precipitation intensity: 11% for CWD, 50% for R95p, 77% for R20, 6% for Rx1day, and 11% for Rx5day. The trends and significance of trends for different precipitation indices under the SSP2-4.5 and SSP5-8.5 showed distinct significant levels in different stations, indicating higher spatial variations of the rainfall pattern in the future. Under SSP2-4.5, significant decreasing trends were found in no sites, while decreasing trends were found in Cox's Bazar districts for CDD in the near future. Though several sites exhibited significant positive trends in the near future, the majority of the sites exhibited no trends in the far future under SSP2-4.5. Similar prediction results were also identified under SSP5-8.5 for all the precipitation indices.

Figure 13 shows that the mean of PRTOT was roughly 2,330 mm during 1985–2014, 25 mm/day for SDII, 64 days for CDD, 19 days for CWD, 322 mm for R95p, 52 days for R20, 88 mm for Rx1day, and 273 mm for Rx5days. Under SSP2-4.5 for the near future, PRTOT, CDD, R95p, and Rx1day exhibited a higher standard deviation (long curve) with a mean value of 2,751 mm, 66 days, 421 mm, and 421 mm, respectively. For the far future, a higher standard deviation value (short curve) was calculated for CDD, CWD, R95p, and Rx1day, with a mean value of 70 days, 20 days, 432 mm, and 432 mm, respectively. Under SSP5-8.5, low standard deviation was found for PRTOT and Rx1day during both timeframes, CDD, and R20 during 2061–2100, and SDII, CWD, R95p, and Rx5days during 2021–2060. PRTOT, CDD, and R20 mm exhibited lower variations during 1985–2014, while SDII, CWD, R95p, Rx1day, and Rx5day during 2021–2060 under SSP5-8.5.

The outcomes of Pearson's correlation analysis are presented in Table 3, which shows that all indices have the strongest relationship with each other except CDD and CWD, where they have a strong negative relationship between them (r = −0.529, p < 0.05). PRTOT exhibited a strong positive relationship with R20, R95p, Rx1day, SDII, and Rx5days (r = 0.917, p < 0.01; r = 0.931, p < 0.01; r = 0.922, p < 0.01; r = 0.736, p < 0.01; r = 0.956, p < 0.01). On the other hand, R20 showed a strong positive relationship with R95p, Rx1day, SDII, and Rx5days (r = 0.821, p < 0.01; r = 0.813, p < 0.01; r = 0.813, p < 0.01; r = 0.860, p < 0.01). However, R95p exhibited a strong positive association with Rx1day, SDII, and Rx5days (r = 0.909, p < 0.01; r = 0.723, p < 0.01; r = 0.931, p < 0.01). On the other hand, Rx1day showed a strong positive correlation with SDII and Rx5days (r = 0.863, p < 0.01; r = 0.992, p < 0.01). Also, SDII has a strong positive relationship with Rx5days (r = 0.850, p < 0.01).

Climate extreme events are becoming more frequent and severe across the world, especially in climate susceptible regions due to recent climatic changes. Bangladesh, with its low-lying topography and dense population, is highly vulnerable to the impacts of climate extreme events, such as floods, landslides, and storm surges, which can lead to widespread devastation, including loss of life, displacement of communities, and damage to infrastructure and agriculture. Understanding historical trends is critical to identifying areas that have been most affected in the past, while future projections will enable policymakers and planners to anticipate and prepare for potential risks. In this regard, this study attempted to assess the historical trends and changes in eight extreme precipitation indices over Bangladesh, also projected for the two future periods under two distinct SSPs (SSP2-4.5 and SSP5-8.5). In this study, 13 CMIP6 models and the extreme precipitation indices include PRTOT, SDII, R95p, R20, Rx1day, Rx5day, CDD, and CWD.

Findings show that northern plainlands, especially the northwestern regions, experienced less rainfall compared to mountainous and coastal areas. The study found that the eastern region of the country received a higher amount of total precipitation over the past few decades, which is consistent with other studies (Ezaz et al. 2022). Predictions of precipitation revealed an increasing trend of PRTOT across the country under both SSPs. The eastern region of the country will experience a higher increase in PRTOT in the coming decades. Wahiduzzamann & Luo (2020) and Ezaz et al. (2022) confirmed the rising trends of PRTOT in Bangladesh. A particularly pronounced increase in PRTOT was projected in both the near and far future timeframes under SSP5-8.5, suggesting that if global emissions continue to increase, the precipitation events and amount across Bangladesh will increase substantially. This aligns with Abdelmoaty & Papalexiou's (2023) global projections using CMIP6 data, which reported that due to increased GHG emissions, global precipitation could increase by 7.5–21% from historical averages. Rising emissions will lead to a rise in global temperatures, which further contributes to the intensification of the monsoon system in the country. Moreover, warmer temperatures will increase the evapotranspiration rates from the ocean and sea, which will lead to more rainfall events in Bangladesh, highlighting the profound impacts of global warming as well as climate change. The prediction of SDII further supports the rising trend of PRTOT, indicating that the increasing trend of precipitation will likely result in higher rainfall intensity, with large volumes of rainfall occurring over shorter periods. The eastern region of Bangladesh is projected to experience the highest SDII, with the most significant increases expected under SSP5-8.5. These findings of the spatial distribution and trends of PRTOT and SDII in Bangladesh are consistent with the studies by Imran et al. (2023) and Khan et al. (2019), suggesting the profound impacts of global climate change and shifting patterns of atmospheric circulation.

CDD in Bangladesh exhibited an inverse spatial distribution of PRTOT and CWD. Historical CDD distributions showed a higher duration of dry spells across the northwestern region during 1985–2014, while the southeastern region experienced shorter CDD. Conversely, CWD was higher in the southern region of the country and shorter in the northwestern region. Future projections suggest an increase in CDD across the country, especially under SSP2-4.5. This finding is consistent with prior studies in Bangladesh by Ahmed et al. (2017) and in a tropical context by Lin et al. (2019). However, projections for the far future under SSP5-8.5 suggest a decline in the CDD range across the country, suggesting that emissions will continue to rise significantly in the future, which will increase global warming, further lead to more frequent rainfall events, and reduce dry periods. This trend is particularly concerning for the northwestern and western regions of Bangladesh, where CDD currently exceeds 120 days and is expected to rise further under SSP2-4.5. The increase in CWD across the country under both SSP scenarios contrasts with the slight rise in CDD, highlighting the complex interplay between global climate drivers and local precipitation patterns. As global temperatures increase due to higher GHG emissions, the intensified hydrological cycle is expected to lead to more frequent and intense rainfall events, particularly in hilly and southeastern regions of Bangladesh. This pattern aligns with findings from the studies by Hu et al. (2017), who observed a decrease in PRTOT in low-altitude areas and an increase in hilly regions, and Islam et al. (2022), who reported rising rainfall trends in Bangladesh's eastern and southern regions between 1980 and 2017.

Future simulations under the SSP2-4.5 scenario suggest that Bangladesh will experience more frequent dry spells, particularly in the northern region, where a significant rise in CDD is projected. The northwestern region is expected to see the most pronounced increase, with CDD exceeding 120 days compared to the reference period, while the southern region is projected to see a more moderate increase of up to 41 days. This indicates that the northern areas may face longer and more severe drought episodes in the future. However, in the southwestern coastal regions, reductions in CDD are anticipated, potentially mitigating drought conditions in these areas. The spatial distribution of CDD across Bangladesh did not reveal a clear trend, which contrasts with the findings of Imran et al. (2023), who reported significant rising trends in CDD from 1981 to 2020. This discrepancy could be attributed to differences in the study periods and the influence of external climatic factors. For instance, the northern region's projected decrease in CDD aligns with the findings from Imran et al. (2023) and Khan et al. (2019), while the expected increase in CWD suggests an opposite trend, highlighting the complexity of climate dynamics in Bangladesh. The significant rise in CWD in the southwestern region suggests a higher risk of coastal flooding, a trend that has already been observed with multiple instances of flooding and cyclones in recent years. This increase in wet days can be linked to global climate change effects, particularly the influence of the ENSO and La Niña events, which are known to disrupt regional weather patterns. ENSO, particularly its La Niña phase, tends to bring increased rainfall to South Asia, which could exacerbate CWD in Bangladesh's coastal regions. The Zonal Walker Circulation, as described by Wang & Yang (2016), also plays a role in intensifying seasonality, causing dry periods to become drier and wet periods wetter, particularly in regions influenced by these large-scale atmospheric patterns. Additionally, SDII showed a decreasing trend in most parts of the country, except for the coastal southwest region, where it followed a similar pattern to CDD. This trend suggests that while the frequency of CWD may decline, the intensity of rainfall on those days may increase, leading to more extreme precipitation events.

The prediction of R95p shows the significant increasing trends of precipitation across the coastal districts in the near future, suggesting a heightened risk of coastal flooding across the country except for the northwest region. This trend is likely driven by the substantial rise in anthropogenic impacts, such as aerosol and greenhouse gas emissions, which have been shown to influence precipitation patterns in floodplain areas (Kamruzzaman et al. 2021a; Islam et al. 2022). The differing climatic nature of Bangladesh's floodplain areas, which are subtropical, and its coastal regions, which are tropical (Abdullah et al. 2020), further complicates these dynamics. Global trends indicate a general decline in precipitation across subtropical regions, particularly in South Asia, the Mediterranean, and Africa (Trenberth 2011; Huang et al. 2017), a phenomenon that could exacerbate the variability in precipitation distribution in Bangladesh.

Moreover, the increase in extreme consecutive precipitation events, as indicated by rising R95p trends, not only elevates the risk of flooding but also poses a significant threat of landslides, particularly in the hilly regions of southeast Bangladesh (Marengo et al. 2020a, b). This is a cause for concern, as Ahmed et al. (2020) have already predicted a higher likelihood of landslide events in the Chittagong hills due to similar trends. The upward trends in R20, which tracks the number of heavy rainfall days, were also observed across most parts of Bangladesh, except for the northeastern region, where no significant trend was noted. Projections for R20 during the period 2021–2060 under SSP2-4.5 revealed positive trends in districts such as Dhaka, Khulna, Jessore, Cumilla, Barishal, Chittagong, Khagrachari, and Bogura, while other areas showed no clear trend. Interestingly, projections for the far future under SSP2-4.5 did not reveal significant trends, and under SSP5-8.5, most districts outside the northeastern region showed no notable trends. However, a surprising pattern emerged during the 2021–2060 period under SSP5-8.5, where many stations exhibited significant negative trends in R20. This suggests a potential shift in the frequency of heavy rainfall events, likely influenced by global climate change and regional atmospheric dynamics, including the impacts of ENSO and La Niña reported by Ehsan et al. (2023). These phenomena are known to affect rainfall distribution in South Asia, with La Niña typically bringing increased rainfall (Sun et al. 2021; An et al. 2023). The observed trends in R95p and R20 indicate that Bangladesh may experience more erratic and intense precipitation patterns in the future.

The findings regarding Rx1day and Rx5day values, which are projected to be highest in the far future under SSP5-8.5 and lowest during the historical period, align with the broader trends of increasing extreme precipitation events in Bangladesh. These significant positive trends in Rx1day and Rx5day, particularly in the southwestern coastal region, can be explained by the critical role of the ocean-atmosphere interface, where internal climate variability, such as changes in sea surface temperatures in the Bay of Bengal, contributes to the rise in these extreme rainfall indices (Sarker 2021). This interaction between the ocean and atmosphere is a key driver of the increased frequency and intensity of precipitation, as highlighted by Roxy et al. (2015). These findings further reinforce the earlier discussion on the rising trends of R95p and R20, indicating a heightened risk of coastal flooding and landslides, particularly in the southeastern regions of Bangladesh.

The connection between the trends of these extreme precipitation indices and global climate change is evident (Rahman & Islam 2019). As global temperatures rise due to increased GHG emissions, the hydrological cycle intensifies, leading to more frequent and severe precipitation events. This is particularly relevant for regions like Bangladesh, where monsoonal patterns dominate. The warming of the Bay of Bengal, driven by global climate change, enhances moisture availability and contributes to more intense rainfall events in the coastal and hilly regions of the country. Moreover, atmospheric circulations, such as the Zonal Walker Circulation, play a significant role in modulating these extreme weather patterns. This circulation, influenced by the global climate system, intensifies seasonal differences, making dry periods drier and wet periods wetter, particularly in regions like Bangladesh that are highly susceptible to such changes. Furthermore, the increasing trends in heavy rainfall (R20), Rx1day, and Rx5day also suggest a higher likelihood of future landslide events in the hilly areas of southeast Bangladesh. The link between rising precipitation extremes and landslide risk is clear, especially in regions with steep terrain and a strong monsoon influence. Conversely, the descending trends in R20 and Rx1day in the western regions of the country point to an increasing potential for drought events in the far future, reflecting the broader global patterns of decreasing precipitation in subtropical regions due to climate change and shifts in atmospheric circulation patterns like ENSO.

However, the study found no significant long-term trends for Rx5day, which can be explained by the variability in short-term climatic events influenced by factors such as ENSO and El Niño/La Niña phases, which can cause fluctuations in precipitation patterns and lead to inconsistent trends over shorter periods. The observed high standard deviation (short curves) in indices like SDII and Rx1day under SSP2-4.5 suggests greater uncertainty and variability in extreme precipitation events in the future. This aligns with the influence of global climate drivers like ENSO on regional weather patterns, further complicating the predictability of these events (Tang et al. 2021). Overall findings imply that while the southwestern coastal and southeastern hilly regions of Bangladesh may experience increased risks of flooding and landslides due to rising precipitation extremes, the western regions could face more frequent droughts. The variability and uncertainty in the trends of precipitation-extreme indices underscore the complex interactions between global climate change, atmospheric circulations, and regional climate patterns, necessitating robust climate adaptation strategies to address the diverse and evolving risks posed by extreme weather events in Bangladesh.

The research has implications for disaster prevention, water resource management, and agricultural planning, all of which are important for the advancement of Bangladesh's agriculture and economy. Rainfall influences sociocultural and economic events in the country including agriculture, power generation, and land use (de Souza Dias et al. 2018). These extreme precipitation indices will change in the future, which could cause problems for agriculture and flooding in the northeastern and central regions of the country. Although the recent economy is mostly concentrated on industrialization, the agricultural sector plays a pivotal role in the economic backbone of the country. Agriculture will need more water for irrigation; thus, precipitation in Bangladesh's key agricultural sectors showed a declining trend in this present investigation. Hence, the declining trend in precipitation may have an effect on Bangladesh's Boro (winter) crops in the future (Islam et al. 2019). The study supports the development of strategies by decision-makers to lessen extreme precipitation events in Bangladesh, such as floods and droughts. The study advances our knowledge of the precipitation variability in Bangladesh. The study contributes to our understanding of climate change and upcoming disasters. The results could be used to assess the nation's infrastructure development and disaster management strategies, at both the policy and implementation levels. The adaptation to climate change and mitigation of climatological disasters will both benefit from this study. The adopted models provide satisfactory performance to predict future climatic events and useful information for future disaster prevention, water resource management, and projected trends of eight extreme precipitation indices in Bangladesh.

This research aims to explore the spatiotemporal trends of historical and future extreme precipitation events by integrating 13 GCMs under CMIP6 and two SSP narratives (SSP2-4.5 and SSP5-8.5) using eight indices. Results revealed that CWD decreased in the southeastern and northwest regions. In contrast, CDD showed an increasing trend in the southeast and mid-west regions and increased precipitation intensity in the northeast region. In the eastern region, between 62 and 114 days, none of the SSPs showed an increase or fall in R20. Rx1day and Rx5day will be higher over most of the country, whereas the lessening is comparatively small. Besides, R95p will rise at a higher rate, especially over the northeastern region. For all SSPs, the projected change in Rx1day did not reach an identical distribution. The future trends for Rx5day were comparable to those for Rx1day. Our results depicted a rise in extreme precipitation over most of the country's northeast and a decrease in the western and northwest regions. The results revealed a likely rise in extreme precipitation-induced risk like floods and landslides in the eastern and southeastern regions where it was previously very widespread. On the whole, the study suggests the elevated vulnerability of the country due to changes in extreme precipitation. The information generated in this research will be helpful for water resource management, weather, and severe disaster studies. This study did not assess the uncertainties and sensitivities among the models, which deserve further examination. Based on existing data, the severity and frequency of severe precipitation events are expected to increase in different regions of Bangladesh due to global warming. Hence, it is essential to thoroughly examine future changes in severe wind directions, relative humidity levels, sunshine hours, and mean sea-level pressure to understand regional climate patterns and their consequences. This study's findings could be applied to the design of hydraulic infrastructures and public safety measures. It can also serve as a foundation for planning mitigation and adaptation measures. Policymakers and stakeholders must comprehend expected changes to mitigate implications on agriculture, ecosystems, human health, and biodiversity. Future research is advised to consider additional GCM models and trustworthy datasets to quantitatively analyze the extreme precipitation in the country. The research assists in the formulation of policies for coping with climate change and opens room for further studies.

The authors highly acknowledge the Bangladesh Meteorological Department (BMD) and Mishra et al. (2020) for providing the required datasets. Output from the CMIP6 models from https://esgf-node.llnl.gov/projects/cmip6/ is highly recognized. We are thankful to the Department of Disaster Management, Begum Rokeya University, Rangpur (BRUR) for other sorts of supports during our study.

R.A.M. and A.R.M.T.I. designed, planned, conceptualized the study, prepared figures, and drafted the original manuscript; M.Y.A. was involved in statistical analysis and interpretation; J.M. contributed to instrumental setup, data analysis, and validation; M.K., M.A.S., and S.C.P. contributed to editing the manuscript, literature review, and proofreading; M.M.A., M.A.F., and A.R.M. T.I. were involved in proofreading during the manuscript drafting stage. All authors reviewed the manuscript.

The authors extend their appreciation to the Deanship of Research and Graduate Studies at King Khalid University for funding this work through Large Research Project under grant number RGP2/332/45.

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.