This study examines observed and projected changes in monsoon timing (onset/withdrawal/duration) and strength (total and extreme rainfall) in the Ganges-Brahmaputra-Meghna basin. While prior South Asia monsoon research has mainly used coarse-resolution models, we utilised CMIP6 HighResMIP models to assess whether improved horizontal resolution (HR-models: <0.25°) improves the simulation of the monsoon when compared to low resolution (LR-models: 0.35 to 1.2 °), evaluated against reference datasets (MSWEP and ERA5). Our findings indicate that HR-models generally outperform LR-models in capturing monsoon characteristics, with the bias in annual average rainfall higher in LR-models. Between 1979-2014, MSWEP (ERA5) tends towards an earlier (later) onset by around 7 (3) days, a later withdrawal by around 8 (12) days, resulting in a longer monsoon duration of approximately 15 (9) days, respectively; much larger changes than both HR- and LR- MODELS. The trends in duration are highest for MSWEP and lowest for LR-models. We find that HR-models better capture observed trends in total and extreme rainfall over 1979-2014 compared to LR-models. For the future climate (2015-2050), HR-(LR-) model ensembles project a delay of ~4 (3) days in monsoon onset under the SSP585 forcing scenario, with HR-models indicating a later onset compared to LR-models. However, HR-models project a shorter monsoon duration and an earlier withdrawal (~3 days) compared to LR-models. We find that HR-models project a significantly greater increase in rainfall than LR-models: these project increases in extreme monsoon rainfall (up to 5.55%/decade) and total monsoon rainfall (around 1.4%/decade) in the GBM basin, while LR-models project little change. Our findings highlight the large uncertainties in simulating monsoon characteristics from climate models, but show that HR-models can be helpful in studying changing monsoon dynamics over the complex topography of the GBM basin.
Research Article
High-resolution climate models improve simulation of monsoon rainfall changes in the Ganga-Brahmaputra-Meghna basin
https://doi.org/10.21203/rs.3.rs-4937815/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
The monsoon system plays a critical role in the hydrological cycle and particularly impacts precipitation in South Asia (Serreze and Barry, 2010). Among the most densely populated agricultural basins globally, the Ganges-Brahmaputra-Meghna (GBM) basin relies heavily on South Asian monsoon rainfall (Ali et al., 2023; Azad et al., 2022). Monsoon rainfall profoundly affects agricultural production, environmental sustainability, and water resource management in the basin (Rahman et al., 2017; Gadgil and Gadgil, 2003). Variations in monsoon timing, intensity, and duration significantly impact agricultural productivity, food security, hydroelectric production, forest vegetation, water resources, and regional ecology (Turner and Annamalai, 2012; Jain et al., 2013). Therefore, a detailed analysis of monsoon rainfall characteristics, including timing, total and extreme rainfall amounts, and trends, is essential for understanding the implications for water resources and the economy in this basin (Mandal et al., 2021; Rahman et al., 2017).
Extensive research has focused on the timing of the South Asian monsoon, considering both regional and large-scale patterns and trends (Azad et al., 2022; Bombardi et al., 2020; Misra et al., 2018; Montes et al., 2019). These studies have utilized different criteria and atmospheric variables to analyze variations in monsoon timing and associated forcing mechanisms, including the assessments of long-term trends. Monsoon onset and retreat are influenced by a combination of local and regional factors, leading to multiple proposed explanations (Wang et al., 2017). In addition to fundamental large-scale factors such as continental heating and meridional wind shifts, mechanisms such as intraseasonal oscillations and forcing from convection over the oceans, especially in the Bay of Bengal, play a significant role (Fasullo and Webster, 2003; Karmakar and Misra, 2019). Furthermore, sea surface temperature anomalies in the Indian and Pacific Oceans, along with El Niño/La Niña events, contribute to variations in monsoon onset timing, impacting the GBM basin (Sun et al., 2017; Xavier et al., 2007).
Global climate models (GCMs) help us to understand the changes in monsoon rainfall by attempting to reproduce its past changes and make projections of its future (Zhu et al., 2020). However, it is challenging to simulate monsoon precipitation at regional scales with these models as they do not adequately represent many of the governing thermodynamic and dynamic processes, leading to systematic and longstanding model biases compared to observations (Haarsma et al., 2016; Roberts et al., 2019). This brings into question model reliability and limits our confidence in future climate projections. To overcome these limitations, there is a growing interest in enhancing the horizontal resolution of climate models, as higher resolution enables the model to more accurately simulate the small-scale regional structures of synoptic and mesoscale systems (Roberts et al., 2018; Xin et al., 2021).
Increasing spatial resolution traditionally involves using Regional Climate Models (RCMs) to downscale Earth System Model (ESM) outputs for finer climate data in specific regions (Avila-Diaz, et al., 2023; Ban et al., 2021). However, while RCMs offer detailed representations of topography and land-ocean differences, they introduce new uncertainties (Giorgi, 2005), such as boundary condition closure issues (Ambrizzi et al., 2019). To address this, high-resolution ESMs are being developed, aiming to provide comprehensive regional and global climate data while incorporating more climate processes compared to RCMs (Demory et al., 2020). Some of the support for this idea comes from previous comparisons of different climate models in projects like the Coupled Model Intercomparison Project (CMIP) (Roberts et al., 2019; Meehl et al., 2007; Taylor et al., 2012).
It is uncertain whether downscaling improves climate projections of the Indian summer monsoon, but there is high confidence in projections of precipitation changes in complex orographic regions due to consistent improvements in these areas, as evident in several dynamical downscaling studies including the IPCC AR6 Chapter 10; Doblas-Reyes et al., 2021). Johnson et al. (2016) discussed that higher resolution models can enhance the representation of precipitation processes over complex orographic regions such as the Western Ghats but they do not address the underlying issue of the dry bias over South Asia and the wet bias over the Indian Ocean. Moreover, Bock et al. (2020) argued that CMIP6 models show no significant improvement over CMIP5 or CMIP3 models in terms of annual mean rainfall biases in the tropics, and HighResMIP models do not significantly reduce the overall bias at the large scale compared to lower-resolution models. We argue that despite these limitations, high-resolution models improve precipitation process representation in regions of complex orography like the GBM basin and can be useful to conduct such studies.
The HighResMIP, endorsed by CMIP6, introduces a novel multi-model approach to systematically explore the effects of horizontal resolution for the first time (Haarsma et al., 2016). These simulations vary in resolution from typical CMIP6 values (~250 km in the atmosphere and 100 km in the ocean) to significantly higher resolutions (25 km in the atmosphere and 8 to 25 km in the ocean). There have already been some relevant analyses. For instance, Fahad et al. (2021) found that the low-resolution simulations from HighResMIP show poor spatial variability of precipitation and a dry bias across Bangladesh; however, the high-resolution coupled simulations have a better representation of topography, which improves the simulation of moisture convergence at the foothills of the Himalaya and reduces precipitation biases.
Here, we examine observed and projected changes in the timing (onset/withdrawal/duration) and strength (total and extreme) of monsoon rainfall across the GBM basin using ensembles of both high- and low-resolution HighResMIP models and reference reanalyses datasets. We address the following key science questions:
- How good are the HighResMIP models in simulating the timing and strength of monsoon rainfall compared to observed reference datasets?
- What are the observed and projected changes in the timing and strength of monsoon rainfall?
- Does an increase in the horizontal resolution of the HighResMIP models improve their performance in simulating monsoon rainfall characteristics for the GBM?
In Section 2, we describe the study region, data, and the definitions of the rainfall indices used in this study. Section 3 presents our results, while Section 4 concludes our findings.
2.1 Study region
Our study is focused on the Ganges-Brahmaputra-Meghna (the GBM hereafter) basin. The GBM is a densely-populated basin with large low-lying deltas and is found between latitude 38 21°25′N to 25°50′N and longitude 87°75′E to 91°75′E, covering Bangladesh and some parts of east India (Ali et al., 2023). It is one of the most populated river basins in the world with approximately 630 million people living in the catchment area of around 1.72 million km2 (Sharma et al., 2021). The GBM is one of the most complex river systems in the world, having a diverse range of topographical and morphological features (Mirza, 2002), an intricate river network and varied elevation (from 1 m a.s.l. in the South to 33 m in the North). Figure 1 highlights the complex orography of the basin and implies the necessity of using high-resolution models to capture important local-scale processes.
To delineate these basins we used boundaries from the HydroSHEDS website through the link https://hydrosheds.org/downloads.
2.2 Climate Reference Datasets
The reference climate datasets used in this study are the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5; Hersbach et al., 2020) and the Multi-source weighted-Ensemble Precipitation (MSWEP; Beck et al., 2017a).
ERA5 is a global atmospheric reanalysis product created by ECMWF using the 4D-Var data assimilation techniques in cycle 41r2 (Karl and Michela, 2019). Precipitation in ERA5 is obtained from a combination of data analysis and forecasting and consists of two surface-level parameters: rainfall and snow. Large-scale precipitation in ERA5 is produced by the cloud scheme, while convective precipitation is derived from the convection scheme. ERA5 data are available from 1950 to the present with a temporal resolution of 1 hour and a horizontal 0.25 degree spatial resolution (Hersbach et al., 2020). ERA5 is found to perform better than ERA-Interim due to the increased spatial resolution, however, uncertainties remain in tropical regions due to the limited observational data available for the evaluation (Ali et al., 2021; Hersbach et al., 2020). Mahto and Mishra (2019) found that ERA5 outperforms other reanalysis products (MERRA2, CFSR, ERA-Interim & JRA-55) for monsoon season precipitation across India.
MSWEP is a new fully-global historic precipitation dataset covering the period from 1979 to 2020. It offers a spatial resolution of 0.25° and a temporal resolution of 3 hours. The long-term mean background of MSWEP is derived from the CHPclim dataset and is supplemented with more accurate regional datasets where available (Beck et al., 2017a). MSWEP takes advantage of two gauge datasets (CPC Unified and GPCC), three satellite products (CMORPH, GSMaP-MVK, and TMPA 3B42RT), and two reanalyses (ERA-Interim and JRA-55) to provide reliable precipitation estimates globally. Therefore, MSWEP isn’t strictly a reanalysis dataset. Previously, Ali et al. (2019) used MSWEP to study multiday flooding events in the Indian subcontinent. More details about the MSWEP dataset can be found at http://www.gloh2o.org/.
2.3 Model Simulations and Projections:
The HighResMIP experiments are categorized into three tiers: atmosphere-only (Tier 1), coupled atmosphere-ocean (Tier 2), spanning from 1950 to 2050, and forced-atmosphere (Tier 3) with potential extension to 2100, alongside additional targeted experiments. Tier 1 experiments, named HighResSST-present, involve historically-forced atmosphere runs from 1950 to 2014 (ForcedAtmos) using the HadISST2.2.0.0 1/4 degree sea-surface temperature (SST) and sea-ice forcing dataset, with fixed land use following the HighResMIP protocol (Haarsma et al., 2016). Tier 2 consists of simulations: a) conducted over 100 years using forcing conditions from the 1950s – “control-1950”; b) spanning from 1950 to 2014 using historical forcing conditions including greenhouse gas, aerosols, land use-land cover, SST and sea ice, natural and anthropogenic forcings –“hist-1950”; and c) scenario projections from 2015 to 2050 using the SSP585 forcing scenario – “highres-future”. The target resolution for Tier 1 & 2 is set at 25 to 50 km, significantly higher than the typical CMIP6 resolution of 100 km. The data can be accessed from https://hrcm.ceda.ac.uk/research/cmip6-highresmip/. For more detailed information on the experimental design, see Haarsma et al. (2016).
We used data from 9 models from the Tier 2 experiments as the Tier 1 experiments can be significantly affected by the lack of atmosphere-ocean coupling. Moreover, atmosphere-ocean coupling helps in producing a realistic simulation of the key teleconnectionss that govern the interannual variability of the monsoon, such as to El Niño (Xavier et al., 2007), and is crucial for any seasonal prediction system (Krishna Kumar et al., 2005). This study considered the first ensemble member of all models (i.e., r1i1p1f1). The details of the selected models are given in Table S1. Models with spatial resolution up to 0.25 degrees are classified as HR-models and models coarser than this resolution are classified as LR-models.
2.4 Rainfall Indices
2.4.1 Timing of the monsoon:
We used the Liebmann et al. (2012) method to determine the onset/withdrawal of the monsoon season which has been previously used by Wainwright et al. (2019) for Africa and a variation thereof by Sperber and Annamalai (2014) for the Indian subcontinent. This accumulation method (method1) uses a timeseries of daily sums of precipitation to calculate the cumulative daily rainfall anomaly C(d), given by
where i ranges from 1 January to 31 December for each year, is the daily rainfall on the ith day and is the annual average daily rainfall. The day of the minimum of C(d) marks the beginning of the monsoon season and the day of the maximum marks the retreat (withdrawal hereafter) (Fig. S2). The time period between these two days is the duration of the monsoon season. The results presented in Figures 3-6 are averaged over the GBM basin. We checked our estimation of the timing of the monsoon using a different fractional accumulation approach (method2) from LinHo and Wang (2002), as discussed in the Supplementary Information. These methods provide dates for each grid point, and the average of these dates was calculated to determine the timing for the entire basin.
Both methods focus on local monsoon characteristics, providing insights into the onset and withdrawal of the rainy season within a small region (Bombardi et al., 2020). Moron and Robertson (2014) stated that local onset definitions can effectively capture large-scale interannual monsoon variability, especially with regional synchronization. Additionally, Bombardi et al. (2020) suggested that although statistical methods (such as multivariate regression models with predictors such as ENSO) and dynamical approaches (using climate models) may differ in defining monsoon onset and withdrawal locally, spatial data aggregation could potentially improve predictability by reducing noise and enhancing the regional monsoonal signal. Therefore, defining the monsoon’s timing in a small area could potentially represent the basin-scale timing.
2.4.2 Strength of the monsoon:
We estimated the changes in the strength of monsoon rainfall using three ETCCDI (Karl et al., 1999) indices:
a) Total Rainfall (PRCPTOT): the cumulative sum of the 6-hourly rainfall during the monsoon season.
b) Annual Maximum Rainfall (Rx6hr): the maximum of 6-hourly rainfall during the monsoon season.
c) 95th percentile of Rainfall (R95p): the 95th quantile of 6-hourly rainfall for wet days (R>=1mm) during the monsoon season.
We calculated the normalised rainfall for these indices by dividing the magnitude of the rainfall within a year by its mean annual rainfall. This normalization allows for a fair comparison of the trends in rainfall indices among the different datasets as it attempts to remove the effect of model bias.
3.1 Evaluation of rainfall datasets
We first assessed the performance of the CMIP6 HighResMIP models, three with horizontal resolutions of up to 0.25 degrees (HR-models) and six coarser LR-models, in simulating average annual rainfall for the historical period 1979-2014 (Fig. 2) against reference datasets (MSWEP and ERA5). MSWEP shows higher rainfall in the eastern half of the basin, with an average of 6.33 mm/day over the basin (Fig. 2a). However, this east-west contrast is not evident in ERA5, which appears to overestimate average rainfall with a wet bias of 2.37 mm/day compared to MSWEP (Fig. 2b). We found that all climate models show a dry bias, with the LR-models showing a higher dry bias (up to -5.3 mm/day) in average annual rainfall against MSWEP, compared to the HR models that show a dry bias of up to -3.02 mm/day. Overall, while the HighResMIP models fail to capture the spatial pattern of rainfall accurately, they do show a rainfall contrast between the ocean and land. Given the basin's size and complex topography, the coarser resolution of LR-models may not adequately capture the local processes driving rainfall, highlighting the importance of using finer resolution models as a better alternative to conduct similar studies.
There is a substantial body of literature evaluating global precipitation products against gauge data, but uncertainty remains due to the lack of ground observations, the selection of datasets, and the durations studied. This makes it challenging to evaluate reference datasets before assessing the HighResMIP models. Some confidence comes from global studies using both reanalyses for hydrological applications. For instance, Beck et al. (2017) evaluated 22 precipitation products on a global scale using rain gauges and hydrological modeling, identifying the MSWEP product as one of the top performers. Recently, Xiang et al. (2021) evaluated eight global gridded precipitation products, including MSWEP and ERA5, across 1382 catchments in China, Europe, and North America, finding that MSWEP outperformed ERA5. On the other hand, Baudouin et al. (2020) cross-validated 20 gridded precipitation datatsets in the Indus basin and found precipitation estimates from the ERA5 closest to observations.
3.2 Timing of the Monsoon
We next assessed the simulation of the timing of onset/withdrawal/duration of monsoon rainfall (as an average across the GBM basin) in the CMIP6 HighResMIP models compared to the reference datasets for 1979-2014 (Fig. 3). Comparison of the two reference datasets shows a strong interannual correlation with each other (r=0.84) and indicates a relatively early onset of monsoon rainfall, typically occurring in May, compared to the CMIP6 HighResMIP models, which show a later onset, on average, in June (Fig. 3a). Across the 1979-2014 period, ERA5 and LR-models display a slight positive trend in mean onset day (averaged across the basin) which means a delay in onset, with variations of up to 3 days (calculated by multiplying the regression slope of the onset with the duration), while MSWEP and the HR-models show a declining trend, and therefore a shift to an earlier onset by 7 days and 2 days respectively (Fig. 3a, d). The interannual variability in the onset timing is larger for the ensembles of LR- than HR-models. Additionally, the reference datasets (r=0.81 between MSWEP and ERA5) and LR-models show a relatively late withdrawal, towards the end of September for the reference datasets and towards the beginning of September for the LR-models. However, the HR-models typically show an average withdrawal in August (Fig. 3b), much too early. All datasets (observations and models) show an increasing trend in the withdrawal of the monsoon during 1979-2014, indicating a delay in the monsoon withdrawal by the end of 2014. The withdrawal date is delayed by up to 12 days in the reference datasets, with ERA5 displaying a greater change (12 days) compared to MSWEP, which shows a delay of up to 8 days. For LR- and HR-models, the delay is up to 3 days (Fig. 3d). Therefore, all datasets consistently show an increasing trend (highest for MSWEP) in the duration of the monsoon for the period 1979-2014. The average duration increases by up to 15 days for MSWEP, 10 days for ERA5, 4 days for HR-models, and 3 days for LR-models, respectively, across the basin (Fig. 3c, d). Method2 gave similar results for the onset across models but consistently indicated a delay in the withdrawal compared to method1 across all models (Fig. S2). The differences between accumulation and fractional accumulation approaches might come from a limitation in method2, possibly due to a threshold set too low for estimating the withdrawal. A proper threshold in method2 is crucial, as it is influenced significantly by winter rainfall – which may itself be biased – and might show notable delays in monsoon withdrawal.
We then examined the long-term trend in monsoon timing between the HIST period (using hist-1950 simulations from 1950-2014) and the FUTURE period (using highres-future simulations from 2015-2050) (Fig. 4). Since the lengths of these periods differ, we estimated the regression slope (%) per decade for the indices during these periods to ensure a fair comparison. For the onset, we observe rising trends (regression slope, RS) for both periods and both categories of models, indicating a delay in the start of the monsoon. The delay is more prominent in the FUTURE period (RS: 2.22%/decade for HR-models and 1.8%/decade for LR-models) compared to the HIST period (RS: 0.94%/decade for HR-models and 0.3%/decade for LR-models) (Fig. 4a, b and e). For the withdrawal, there is a rising trend (delay, RS: around 0.6%/decade for models) in the HIST period and a declining trend (early, RS: around -0.4%/decade for models) in the FUTURE period (Fig. 4e). Consequently, the monsoon duration decreases more in the FUTURE period (up to -2.8%/decade for HR-models) compared to the HIST period (around -0.48%/decade for HR-models). Monsoon duration is generally longer in LR-models than in HR-models for both periods (Fig. 4c-e).
The uncertainty in results from across the coupled models arises from their limitations in capturing various aspects of the monsoon, largely due to inaccuracies in representing physical processes like convection and SSTs, which are common model biases (Bollasina and Ming, 2013; Sperber et al., 2013). Coupled CMIP-class models typically have cold biases in the Arabian Sea, which leads to reduced evaporation and moisture fluxes reaching the monsoon during summer (Levine et al., 2012, 2013). Consequently, the coupling and associated cold SST biases over the Arabian Sea significantly contribute to the delayed mean onset in these coupled models compared to reference datasets (Levine et al., 2013; Menon et al., 2018). The limitation in simulating accurate SST can be partly addressed by increasing the horizontal resolution of the models. For instance, Bhattacharya et al. (2022) showed that CMIP6 high-resolution models produce more accurate Arabian SSTs with reduced cold bias compared to lower-resolution models.
Our results showing trends in observed timing of the monsoon rainfall using the ensemble of HR-models are comparable to the findings of Montes et al. (2021). The observed delay in the onset and withdrawal of the monsoon in the GBM basin, as well as the increase in its duration, can be attributed to a complex mix of factors including climate change, oceanic changes, land use chnages, and atmospheric pollution (Dong et al., 2016; Montes et al., 2021; Sun et al., 2023; Sun et al., 2017). Morevover, HR-models project an early onset and early withdrawal (and consequently a shortened monsoon duration). Global warming might weaken the upper tropospheric land-sea thermal contrast due to increased tropical diabatic heating, which could overshadow the enhanced lower tropospheric contrast, leading to a weaker monsoon and possibly delayed onset (Sun et al. 2010). However, global warming might also slow down or shift the tropical circulation (Vecchi and Soden 2007), weakening monsoon circulation and delaying onset (e.g., Zhang et al. 2013). The IPCC AR6 (in particular Chapters 8 and 10) suggest medium confidence in the projected weakening of the South Asian monsoon circulation, potentially leading to changes in the spatial distribution and timing of monsoon rainfall, including potential delays in monsoon onset and changes in withdrawal patterns (Douville et al., 2021; Doblas-Reyes et al., 2021).
3.3 Strength of the Monsoon
We also analyzed changes to the strength of monsoon rainfall, focusing on total (PRCPTOT) and extreme (Rx6HR & R95p) rainfall indices, using the reference datasets (MSWEP & ERA5) and ensembles of HR- and LR-models for the historical period 1979-2014 (Fig. 5). Since we observed a large bias in average annual rainfall among the models (Fig. 2), for a fair comparison, we calculated the trends in normalized rainfall averaged across the GBM basin for these indices. Our findings show a relatively similar linear trend in the change of PRCPTOT between HR-models (6%) and the reference datasets (up to 10%) over the historic period (Fig. 5a, d). In contrast, LR-models show a decline (2%) in PRCPTOT, capturing higher annual variability (Fig. 3). For the Rx6HR index, LR-models and MSWEP display a decline (~2%), while HR-models and ERA5 exhibit an increase of 10% and 2% respectively (Fig. 3b, d). Importantly, all datasets show an increasing trend in R95p (up to 5%) during the historic period.
We further assessed the projected changes in rainfall indices between the HIST and FUTURE periods (Fig. 6). All models show increasing trends in all indices, with more increases during the FUTURE period compared to the HIST period. Specifically, HR-models show an average increase of ~1.4% (~0.15%), ~3.8% (0.4%), and ~5.5% (0.15%) per decade for PRCPTOT, Rx6HR, and R95p, respectively, for the FUTURE (HIST) period. The LR-models show lower increases and a higher range (mean ± standard deviation) for the FUTURE period. Our results reveal a larger projected increase in extreme monsoon rainfall compared to total monsoon rainfall which is particularly prominent in the more realistic HR-models. The discrepancy in trends simulated by the HR-models and LR-models, as highlighted by Bador et al. (2020), underscores the significant rise in rainfall extremes over the tropics, which is underestimated by the LR-models. Furthermore, our findings are consistent with earlier studies focussing on GBM basin regions (Bhattacharjee et al., 2023; Kamruzzaman et al., 2023; Das et al., 2022) that project higher monsoon rainfall over Bangladesh and eastern India across all RCP scenarios. For example, Almazroui et al. (2020) reported a projected monsoon rainfall increase ranging from 7.5% to 36.9% (for SSP-8.5) by the end of the 21st century across Bangladesh, which covers a significant portion of the GBM delta.
There is a much discussion on monsoon rainfall trends as recent observational studies show mixed trends in South Asian monsoon rainfall over the past century, indicating significant interannual and spatial variability, with a weakening trend in overall monsoon rainfall since the 1950s (Kulkarni et al., 2012; Jamshadali et al., 2021). While average rainfall might not show a significant increasing trend, the frequency and intensity of heavy rainfall events have risen (Ali et al., 2019; Goswami et al., 2006; Shahid, 2011). Moreover, CMIP5 model projections suggest an increase in heavy rainfall events due to higher atmospheric moisture content in the future (Sooraj et al., 2015). The IPCC AR6 Chapter10 highlights that global warming is likely to increase the frequency and intensity of intense precipitation events in the monsoon regions where extreme rainfall events have become more common (Doblas-Reyes et al., 2021). This increase is attributed to the increase in atmospheric moisture content due to warming, which will be a significant factor driving intense monsoon rainfall. The IPCC AR6 Chapter 8 (Douville et al., 2021) also suggests high confidence that rainfall extremes in the Indian monsoon region will increase due to global warming.
Previous studies have debated whether increasing horizontal resolution, such as in the HighResMIP models, improves model performance. For instance, Xin et al. (2021) found that the multi-model mean of higher-resolution models (30–50 km) outperformed their lower-resolution counterparts (70–140 km) in capturing rainfall patterns over northwest and southwest China. This improvement was found to be primarily due to the higher-resolution models’ ability to reproduce topographical rainfall and local vertical circulation over complex terrain. Moreover, Liang et al. (2021) found that HighResMIP models with higher horizontal and vertical resolutions showed an improved performance in simulating total rainfall, capturing the observed annual cycle and spatial rainfall patterns, and representing the relationship between precipitation and monsoon intensity across different monsoon seasons from 2001 to 2014 in peninsular Malaysia when compared to coarser-resolution simulations and observed datasets. In contrast, Avial-Diaz et al. (2022) found no strong relationship between an increase in resolution and improved performance of the HighResMIP models in simulating rainfall extremes across Latin America and the Caribbean. We emphasise, therefore, that the HR-models within the HighResMIP framework offer some improvement in reliability in projecting potential future changes in rainfall under a warming climate, although their performance may vary based on the specific study region and phenomena of interest.
In this study, we have examined the changes to the timing and strength of monsoon rainfall in the GBM basin using reference datasets (MSWEP & ERA5) and the ensemble of HR- and LR-models from the CMIP6 HighResMIP. Our key findings are as follows:
a) Monsoon Timing:
The CMIP6 HighResMIP models show a delayed onset of monsoon rainfall, typically in June, compared to the reference datasets (ERA5 and MSWEP), which exhibit a strong interannual correlation (r=0.84) and indicate an earlier onset in May. Between 1979-2014, ERA5 and LR-models display a slight positive trend, indicating a delay in onset by up to 3 days, while MSWEP and HR-models show a shift to an earlier onset by 7 days and 2 days, respectively. All datasets indicate a delay in monsoon withdrawal by the end of 2014, with ERA5 and MSWEP showing delays of up to 12 and 8 days, respectively, and an increased monsoon duration by up to 15 days for MSWEP, 10 days for ERA5, 4 days for HR-models, and 3 days for LR-models.
We also estimated the projected change in monsoon timing between the HIST period (1950-2014) and the FUTURE period (2015-2050). For the onset, both periods and model categories show rising trends, indicating a delay to the onset, more prominent in the FUTURE period (RS: 8% for HR-models and 6.5% for LR-models) compared to the HIST period (RS: 6% for HR-models and 2% for LR-models). For monsoon withdrawal, there is a delay in the HIST period (RS: ~4% for models) and a trend towards earlier withdrawal in the FUTURE period (RS: ~-1.5% for models), which leads to a prominent decrease in monsoon duration in the FUTURE period compared to the HIST period, with longer durations generally observed in LR-models for both periods.
b) Strength of the Monsoon:
From 1979-2014, we found similar linear trends in PRCPTOT between HR-models (6%) and the reference datasets (up to 10%). In contrast, LR-models show a decline (2%) in PRCPTOT with higher annual variability. For the Rx6HR index, both LR-models and MSWEP observations display a decline (~2%), while HR-models and ERA5 reanalysis exhibit increases of 10% and 2%, respectively. Additionally, all datasets show an increasing trend in R95p (up to 5%) during this period.
All models show projected increases in all strength indices, with greater increases during the FUTURE (2015-2050) period compared to the HIST (1950-2014) period. Specifically, HR-models show average increases of ~5% (~1%), ~15% (1.5%), and ~20% (0.5%) for PRCPTOT, Rx6HR, and R-95, respectively, for the FUTURE (HIST) period. The LR-models, which showed a higher increase than HR-models in the HIST period, display lesser increases and a higher range (mean ± standard deviation) for the FUTURE period.
The projected rise in monsoon rainfall can be attributed to intensified thermodynamic conditions caused by global warming (Meehl et al., 2003). The Clasius-Clapeyron relationship explains that for every 1 °C increase in temperature, the atmosphere can hold approximately 7% more moisture which is evident in several studies using observations, and can be higher than 7% for sub-daily extreme rainfall (Ali et al., 2021b; 2022). The additional moisture from global warming, particularly during heavy rainfall, results in an increased long-term rainfall rate (Ali et al., 2017). Moreover, studies (IPCC AR6; Shahi et al., 2023) have utilised the CMIP6 models to study high-impact rainfall events across India and found a substantial connection between global warming and the projected increase in the frequency and intensity of extreme rainfall events. In the GBM basin, the Bay of Bengal acts as a crucial moisture source for thunderstorms. The rising sea surface temperatures may lead to stronger and more consistent winds over the Bay of Bengal, potentially contributing to further increases in rainfall in Bangladesh (Bhattacharjee et al., 2023). While the rising trend in total monsoon rainfall may benefit crop irrigation, the projected increase in extreme rainfall poses significant risks to the GBM delta, potentially making it more vulnerable to severe flash flooding, leading to flood hazards, crop damage, and soil erosion.
Overall, our analysis suggests that there is uncertainty in monsoon timing, duration, and strength across different climate models and reference observational datasets for the Ganga-Brahmaputra-Meghna basin. The HR-models generally performed better in capturing present-day monsoon characteristics compared to the LR-models, with notable differences in onset, duration, and rainfall trends. The local-scale definitions used to define the monsoon timing in our study could be further refined by considering factors such as rainfall event duration and dry periods during the monsoon season to improve accuracy. In addition to rainfall, Li et al. (2014) recently reviewed eight indices based on meridional wind (e.g., Monsoon Hadley Circulation Index), atmospheric temperature (e.g., Tropospheric Temperature Gradient), outgoing longwave radiation (e.g., Convection Index), and the hydrological cycle to study the timing of the monsoon.
Acknowledgement
Haider Ali and Hayley J. Fowler were supported by the Living Deltas project (UKRI/GCRF funded: Grant no: NE/S008926/1). Andrew G. Turner was supported by the National Centre for Atmospheric Science through the NERC National Capability International Programmes Award (NE/X006263/1). The CMIP6 HighResMIP models data was downloaded from the JASMIN (/badc/cmip6/data/CMIP6/HighResMIP/).
Conflict of Interest
The authors declare no conflict of interest.
Data Availability Statement
This article draws on data that will be made available via Newcastle University’s Research Repository (https://data.ncl.ac.uk/). The data will be available from March 2025 onwards, as part of the data generated by the GCRF UKRI-funded Living Deltas Hub (2019-2024) under Grant Reference NE/S008926/1. https://doi.org/10.25405/data.ncl.c.6288033.v1
Author contributions
Haider Ali: Conceptualization; Data curation; Methodology; Formal Analysis; Visualization; Writing – Original Draft Preparation; Writing – Review & Editing
Hayley J. Fowler: Conceptualization; Funding Acquisition; Writing – Original Draft Preparation; Writing – Review & Editing
Andrew G. Turner: Conceptualization; Writing – Original Draft Preparation; Writing – Review & Editing
- Ali, H., Fowler, H. J., Vanniere, B., & Roberts, M. J. (2023). Fewer, but more intense, future Tropical Storms over the Ganges and Mekong basins. Geophysical Research Letters, 50(17), e2023GL104973.
- Ali, H., Fowler, H. J., Lenderink, G., Lewis, E., & Pritchard, D. (2021b). Consistent large-scale response of hourly extreme precipitation to temperature variation over land. Geophysical Research Letters, 48, e2020GL090317.
- Ali, H., Fowler, H. J., & Mishra, V. (2018). Global observational evidence of strong linkage between dew point temperature and precipitation extremes. Geophysical Research Letters, 45(22), 12–320.
- Ali, H., & Mishra, V. (2017). Contrasting response of rainfall extremes to increase in surface air and dewpoint temperatures at urban locations in India. Scientific Reports, 7(1), 1–15.
- Ali, H., Fowler, H. J., Pritchard, D., Lenderink, G., Blenkinsop, S., & Lewis, E. (2022). Towards quantifying the uncertainty in estimating observed scaling rates. Geophysical Research Letters, 49(12), e2022GL099138.
- Ali, H., Peleg, N., & Fowler, H. J. (2021a). Global scaling of rainfall with dewpoint temperature reveals considerable ocean‐land difference. Geophysical Research Letters, 48(15), e2021GL093798.
- Ali, H., Modi, P., & Mishra, V. (2019). Increased flood risk in Indian sub-continent under the warming climate. Weather and Climate Extremes, 25, 100212.
- Almazroui, M., Saeed, S., Saeed, F., Islam, M. N., & Ismail, M. (2020). Projections of precipitation and temperature over the South Asian countries in CMIP6. Earth Systems and Environment, 4, 297-320.
- Ambrizzi T, Reboita M, da Rocha R, Llopart M (2019) The state-ofthe-art and fundamental aspects of regional climate modeling in South America. Ann NY Acad Sci 1436:98–120. https://doi.org/ 10.1111/nyas.13932.
- Annamalai H, Hamilton K, Sperber KR (2007) The south Asian summer monsoon and its relationship with ENSO in the IPCC AR4 simulations. J Clim 20:1071–1092Avila-Diaz, A., Torres, R. R., Zuluaga, C. F., Cerón, W. L., Oliveira, L., Benezoli, V., ... & Medeiros, F. (2023). Current and future climate extremes over Latin America and Caribbean: assessing earth system models from high resolution model intercomparison project (HighResMIP). Earth Systems and Environment, 7(1), 99-130.
- Azad, M. A. K., Islam, A. R. M. T., Ayen, K., Rahman, M. S., Shahid, S., & Mallick, J. (2022). Changes in monsoon precipitation patterns over Bangladesh and its teleconnections with global climate. Theoretical and Applied Climatology, 148(3), 1261-1278.
- Bador, M., Boé, J., Terray, L., Alexander, L. V., Baker, A., Bellucci, A., ... & Vanniere, B. (2020). Impact of higher spatial atmospheric resolution on precipitation extremes over land in global climate models. Journal of Geophysical Research: Atmospheres, 125(13), e2019JD032184.
- Ban, N., Caillaud, C., Coppola, E., Pichelli, E., Sobolowski, S., Adinolfi, M., ... & Zander, M. J. (2021). The first multi-model ensemble of regional climate simulations at kilometer-scale resolution, part I: evaluation of precipitation. Climate Dynamics, 57, 275-302.
- Baudouin, J. P., Herzog, M., & Petrie, C. A. (2020). Cross-validating precipitation datasets in the Indus River basin. Hydrology and Earth System Sciences, 24(1), 427-450.
- Bhattacharya, B., Mohanty, S., & Singh, C. (2022). Assessment of the potential of CMIP6 models in simulating the sea surface temperature variability over the tropical Indian Ocean. Theoretical and Applied Climatology, 148(1), 585-602.
- Bhattacharjee, A., Hassan, S. Q., Hazra, P., Kormoker, T., Islam, S., Alam, E., ... & Towfiqul Islam, A. R. M. (2023). Future changes of summer monsoon rainfall and temperature over Bangladesh using 27 CMIP6 models. Geocarto International, 38(1), 2285342.
- Bombardi, R. J., Moron, V., & Goodnight, J. S. (2020). Detection, variability, and predictability of monsoon onset and withdrawal dates: A review. International Journal of Climatology, 40(2), 641-667.
- Das, S., Kamruzzaman, M., & Islam, A. R. M. T. (2022). Assessment of characteristic changes of regional estimation of extreme rainfall under climate change: A case study in a tropical monsoon region with the climate projections from CMIP6 model. Journal of hydrology, 610, 128002.
- Demory, M. E., Berthou, S., Fernández, J., Sørland, S. L., Brogli, R., Roberts, M. J., ... & Vautard, R. (2020). European daily precipitation according to EURO-CORDEX regional climate models (RCMs) and high-resolution global climate models (GCMs) from the High-Resolution Model Intercomparison Project (HighResMIP). Geoscientific Model Development, 13(11), 5485-5506.
- Doblas-Reyes, F. J., Sörensson, A. A., Almazroui, M., Dosio, A., Gutowski, W. J., Haarsma, R., et al. (2021). Linking global to regional climate change. In V. Masson-Delmotte, P. Zhai, A. Pirani, S. L. Connors, C. Péan, S. Berger, et al. (Eds.), Climate change 2021: The physical science basis. Contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change. Cambridge University Press.
- Douville, H., K. Raghavan, J. Renwick, R.P. Allan, P.A. Arias, M. Barlow, R. Cerezo-Mota, A. Cherchi, T.Y. Gan, J. Gergis, D. Jiang, A. Khan, W. Pokam Mba, D. Rosenfeld, J. Tierney, and O. Zolina, 2021: Water Cycle Changes. In Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Masson-Delmotte, V., P. Zhai, A. Pirani, S.L. Connors, C. Péan, S. Berger, N. Caud, Y. Chen, L. Goldfarb, M.I. Gomis, M. Huang, K. Leitzell, E. Lonnoy, J.B.R. Matthews, T.K. Maycock, T. Waterfield, O. Yelekçi, R. Yu, and B. Zhou (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, pp. 1055–1210, doi: 10.1017/9781009157896.010.
- Dong, G., Zhang, H., Moise, A., Hanson, L., Liang, P., & Ye, H. (2016). CMIP5 model-simulated onset, duration and intensity of the Asian summer monsoon in current and future climate. Climate Dynamics, 46, 355-382.Fahad, A. A., Singh, B., Kamal, M., Ahmed, T., Kibria, M., & Chowdhury, N. R. (2022). The role of local topography and sea surface temperature on summer monsoon precipitation over Bangladesh and n ortheast India. International Journal of Climatology, 42(9), 4564-4579.
- Fasullo, J., & Webster, P. J. (2003). A hydrological definition of Indian monsoon onset and withdrawal. Journal of Climate, 16(19), 3200-3211.
- Gadgil, S., & Gadgil, S. (2006). The Indian monsoon, GDP and agriculture. Economic and political weekly, 4887-4895.Giorgi F, Francisco R (2001) Uncertainties in the prediction of regional climate change. In: Global change and protected areas. pp 127–139.
- Goswami, B. N., Venugopal, V., Sengupta, D., Madhusoodanan, M. S., & Xavier, P. K. (2006). Increasing trend of extreme rain events over India in a warming environment. Science, 314(5804), 1442-1445.
- Haarsma, R. J., Roberts, M. J., Vidale, P. L., Senior, C. A., Bellucci, A., Bao, Q., ... & von Storch, J. S. (2016). High resolution model intercomparison project (HighResMIP v1. 0) for CMIP6. Geoscientific Model Development, 9(11), 4185-4208.
- Hersbach, H., Bell, B., Berrisford, P., Hirahara, S., Horányi, A., Muñoz‐Sabater, J., ... & Thépaut, J. N. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999-2049.
- Jain SK, Kumar V, Saharia M (2013) Analysis of rainfall and temperature trends in Northeast India. Int J Climatol 33(4):968–978.
- Jamshadali, V. H., Reji, M. J. K., Varikoden, H., & Vishnu, R. (2021). Spatial variability of south Asian summer monsoon extreme rainfall events and their association with global climate indices. Journal of Atmospheric and Solar-Terrestrial Physics, 221, 105708.
- Johnson, S. J., Levine, R. C., Turner, A. G., Martin, G. M., Woolnough, S. J., Schiemann, R., ... & Strachan, J. (2016). The resolution sensitivity of the South Asian monsoon and Indo-Pacific in a global 0.35 AGCM. Climate Dynamics, 46, 807-831.
- Ju J, Slingo JM (1995) The Asian summer monsoon and ENSO. Q J R Meteorol Soc 121:1133–1168.
- Kamruzzaman, M., Wahid, S., Shahid, S., Alam, E., Mainuddin, M., Islam, H. T., ... & Thorp, K. R. (2023). Predicted changes in future precipitation and air temperature across Bangladesh using CMIP6 GCMs. Heliyon, 9(5).
- Karl, T.R., N. Nicholls, and A. Ghazi, 1999: CLIVAR/GCOS/WMO workshop on indices and indicators for climate extremes: Workshop summary. Climatic Change, 42, 3-7.Karmakar, N., & Misra, V. (2019). The relation of intraseasonal variations with local onset and demise of the Indian summer monsoon. Journal of Geophysical Research: Atmospheres, 124(5), 2483-2506.
- Khandu, J., Awange, L., Kuhn, M., Anyah, R. & Forootan, E. (2017) Changes and variability of precipitation and temperature in the Ganges–Brahmaputra–Meghna River basin based on global high-resolution reanalyses. International Journal of Climatology, 37, 2141–2159.
- Krishna Kumar, K., Hoerling, M., & Rajagopalan, B. (2005). Advancing dynamical prediction of Indian monsoon rainfall. Geophysical Research Letters, 32(8).
- Kulkarni, A. (2012). Weakening of Indian summer monsoon rainfall in warming environment. Theoretical and Applied Climatology, 109, 447-459.
- Lau, K.-M. & Yang, S. (1997) Climatology and interannual variability of the southeast Asian summer monsoon. Advances in Atmospheric Sciences, 14(2), 141–162.
- Levine RC, Turner AG (2012) Dependence of Indian monsoon rainfall on moisture fluxes across the Arabian Sea and the impact of coupled model sea surface temperature biases. Clim Dyn 38:2167–2190. doi:10.1007/s00382-011-1096-z.
- Levine, R. C., Turner, A. G., Marathayil, D., & Martin, G. M. (2013). The role of northern Arabian Sea surface temperature biases in CMIP5 model simulations and future projections of Indian summer monsoon rainfall. Climate Dynamics, 41, 155-172.
- Li, X., Wang, L., Zhong, S., & Liu, L. (2024). Comparative analysis of indices in capturing the onset and withdrawal of the South Asian Summer Monsoon. Environmental Research Communications, 6(3), 031007.
- Liang, J., Tan, M. L., Hawcroft, M., Catto, J. L., Hodges, K. I., & Haywood, J. M. (2022). Monsoonal precipitation over Peninsular Malaysia in the CMIP6 HighResMIP experiments: the role of model resolution. Climate Dynamics, 58(9), 2783-2805.
- LinHo, & Wang, B. (2002). The time–space structure of the Asian–Pacific summer monsoon: a fast annual cycle view. Journal of Climate, 15(15), 2001-2019.
- Mahto, S. S., & Mishra, V. (2019). Does ERA‐5 outperform other reanalysis products for hydrologic applications in India?. Journal of Geophysical Research: Atmospheres, 124(16), 9423-9441.
- Mandal, T., Das, J., Rahman, A. S., & Saha, P. (2021). Rainfall insight in Bangladesh and India: climate change and environmental perspective. Habitat, Ecology and Ekistics: Case Studies of Human-Environment Interactions in India, 53-74.
- Masson-Delmotte, V., Zhai, P., Pirani, A., Connors, S. L., Péan, C., Berger, S., ... & Zhou, B. (2021). Climate change 2021: the physical science basis. Contribution of working group I to the sixth assessment report of the intergovernmental panel on climate change, 2(1), 2391.
- Meehl, G. A., & Arblaster, J. M. (2003). Mechanisms for projected future changes in south Asian monsoon precipitation. Climate Dynamics, 21, 659-675.
- Meehl, G. A., Covey, C., Delworth, T. L., Latif, M., McAveney, B., Mitchell, J. F. B., Stouffer, R. J., and Taylor, K. E. (2007): The WCRP CMIP3 multimodel dataset: A new era in climate change research, B. Am. Meteorol. Soc., 88, 1383–1394, https://doi.org/10.1175/BAMS-88-9-1383.
- Menon, A., Turner, A. G., Martin, G. M., & MacLachlan, C. (2018). Modelling the moistening of the free troposphere during the northwestward progression of Indian monsoon onset. Quarterly Journal of the Royal Meteorological Society, 144(713), 1152-1168.
- Mirza, M. M. Q. (2002). Global warming and changes in the probability of occurrence of floods in Bangladesh and implications. Global environmental change, 12(2), 127-138.
- Misra, V., Bhardwaj, A., & Mishra, A. (2018). Local onset and demise of the Indian summer monsoon. Climate Dynamics, 51, 1609-1622.
- Montes, C., Acharya, N., Stiller‐Reeve, M. A., Kelley, C., & Hassan, S. Q. (2021). Interannual variability of monsoon onset and withdrawal in Bangladesh. Atmospheric Science Letters, 22(11), e1069.
- Prodhomme, C., Terray, P., Masson, S., Boschat, G., & Izumo, T. (2015). Oceanic factors controlling the Indian summer monsoon onset in a coupled model. Climate Dynamics, 44, 977-1002.
- Rahman MA, Yunsheng L, Sultana N (2017) Analysis and prediction of rainfall trends over Bangladesh using Mann-Kendall, Spearman’s rho tests and ARIMA model. Meteorol Atmos Phys 129:409–424.
- Roberts, M. J., Baker, A., Blockley, E. W., Calvert, D., Coward, A., Hewitt, H. T., ... & Vidale, P. L. (2019). Description of the resolution hierarchy of the global coupled HadGEM3-GC3. 1 model as used in CMIP6 HighResMIP experiments. Geoscientific Model Development, 12(12), 4999-5028.
- Serreze, M.C., Barry, R.G.,: 2010 Climate change. In: Barry, R.G., Chorley, R.J. (Eds.), Atmosphere, Weather and Climate. Routledge, Oxon.
- Shahi, N. K., Rai, S., Verma, S., & Bhatla, R. (2023). Assessment of future changes in high-impact precipitation events for India using CMIP6 models. Theoretical and Applied Climatology, 151(1), 843-857.
- Shahid, S. (2011). Trends in extreme rainfall events of Bangladesh. Theoretical and applied climatology, 104, 489-499.
- Sharma, C., Shukla, A. K., & Zhang, Y. (2021). Climate change detection and attribution in the Ganga-Brahmaputra-Meghna river basins. Geoscience Frontiers, 12(5), 101186.
- Sooraj, K. P., Terray, P., & Mujumdar, M. (2015). Global warming and the weakening of the Asian summer monsoon circulation: assessments from the CMIP5 models. Climate Dynamics, 45, 233-252.
- Sperber, K. R., & Annamalai, H. (2014). The use of fractional accumulated precipitation for the evaluation of the annual cycle of monsoons. Climate dynamics, 43, 3219-3244.
- Sun, W., Wu, G., Liu, Y., Mao, J., Zhuang, M., & Liu, X. (2023). Delayed response of the onset of the summer monsoon over the Bay of Bengal to land–sea thermal contrast. Journal of Climate, 36(12), 4051-4070.
- Sun Y, Ding Y, Dai A (2010) Changing links between South Asian summer monsoon circulation and tropospheric land–sea thermal contrasts under a warming scenario. Geophys Res Lett 37:L02704.
- Sun, S., Ren, R., & Wu, G. (2017). Onset of the Bay of Bengal summer monsoon and the seasonal timing of ENSO's decay phase. International Journal of Climatology, 37(14), 4938-4948.
- Turner AG, Annamalai H (2012) Climate change and the South Asian summer monsoon. Nat Climate Change 2:587–595.
- Taylor, K. E., Stouffer, R. J., and Meehl, G. A. (2012): An overview of CMIP5 and the experimental design, B. Am. Meteorol. Soc., 93, 485–498.
- Vecchi GA, Soden BJ (2007) Global warming and the weakening of the tropical circulation. J Clim 20:4316–4340.Wang, B., Li, J., & He, Q. (2017). Variable and robust East Asian monsoon rainfall response to El Niño over the past 60 years (1957–2016). Advances in Atmospheric Sciences, 34, 1235-1248.
- Wainwright, C. M., Marsham, J. H., Keane, R. J., Rowell, D. P., Finney, D. L., Black, E., & Allan, R. P. (2019). ‘Eastern African Paradox’rainfall decline due to shorter not less intense Long Rains. npj Climate and Atmospheric Science, 2(1), 34.
- Xavier, P. K., Marzin, C., & Goswami, B. N. (2007). An objective definition of the Indian summer monsoon season and a new perspective on the ENSO–monsoon relationship. Quarterly Journal of the Royal Meteorological Society: A journal of the atmospheric sciences, applied meteorology and physical oceanography, 133(624), 749-764.
- Xiang, Y., Chen, J., Li, L., Peng, T., & Yin, Z. (2021). Evaluation of eight global precipitation datasets in hydrological modeling. Remote Sensing, 13(14), 2831.
- Xin, X.; Wu, T.; Jie, W.; Zhang, J. (2021) Impact of Higher Resolution on Precipitation over China in CMIP6 HighResMIP Models. Atmosphere, 12, 762. https://doi.org/ 10.3390/atmos12060762.
- Zhang H, Moise A, Liang P, Hanson L (2013) The response of summer monsoon onset/retreat in Sumatra-Java and tropical Australia region to global warming in CMIP3 models. Clim Dyn 40:377–399.Zhu, H., Jiang, Z., Li, J., Li, W., Sun, C., & Li, L. (2020). Does CMIP6 inspire more confidence in simulating climate extremes over China?. Advances in Atmospheric Sciences, 37, 1119-1132.