Assessment of climate change impact on hydrological components of Ponnaiyar river basin, Tamil Nadu using CMIP6 models

Increasing temperatures, changing rainfall patterns, melting glaciers and rising sea level are evidence that the changing climate has impacted natural systems over the past few decades (van Vuuren et al. 2011; UNDESA 2014). In addition, it causes extreme events such as floods and droughts worldwide, which alters the hydrologic cycle components (Gurung et al. 2022). Therefore, more focus is needed to evaluate the climate change impacts on water availability at the basin scale to meet the future water needs of humans, agriculture and industries (Haleem et al. 2022; IPCC 2022a). Furthermore, since topography, land use, soil type and climate conditions vary from one river basin to another, the impact of climate change on available water resources could also change according to the river basins (Abeysingha et al. 2020; Ich & Sok 2022). Hence, it is crucial to evaluate the climate change effects exclusively on each river basin to develop future management practices and adaptive strategies.

The global circulation models (GCMs) developed by various institutes from different countries simulate the future projections of climate parameters, such as rainfall, temperature, humidity, radiation and wind speed (Iranmanesh et al. 2021). These parameters help to assess the future climate change impacts on the hydrological system (Carmin et al. 2012; UNDESA 2012). The applications of GCMs have increased enormously due to the efforts of coupled model inter-comparison project (CMIP). Recently, the Intergovernmental Panel for Climate Change (IPCC) released the sixth assessment report (AR6) based on the latest CMIP6 models (Eyring et al. 2016; Swathi et al. 2018; Yue et al. 2021; IPCC 2022b). The significant improvement of CMIP6 compared with CMIP5 is the inclusion of socioeconomic development factors with GHG emissions scenarios (representative concentration pathways (RCPs)) (Gidden et al. 2019; Lovino et al. 2021). Here, future projection of climate variables has been established based on the Shared Socioeconomic Pathways (SSP) under low-, medium- and high-emission scenarios (SSP1-2.6, SSP2-4.5, SSP3-8.5) (Lovino et al. 2021; Yue et al. 2021; IPCC 2022b).

The spatial scale of these GCMs is generally hundreds of kilometres to represent the earth system components, including land, ocean and atmosphere. However, capturing local scale details should be improved, giving promising results for the users working on regional scale studies (Li et al. 2010; Ougahi 2022). Downscaling and bias correction is further needed to overcome these problem associated with uncertainties (Serban 1990; Golmohammadi et al. 2014; Mishra et al. 2020). Mishra et al. (2020) have developed a daily bias-corrected projection dataset of rainfall and maximum and minimum temperature from CMIP6-GCMs at the spatial resolution of 25° for the South Asia continent, including India, Pakistan, Sri Lanka, Nepal, Bhutan and Bangladesh. This dataset is available for the historical (1951–2014) and future periods (2015–2100) over the South Asia domain.

Hydrological models represent the system that simulates the processes involved in the hydrological cycle. Applications of hydrological models are real-time forecasting and water resources system operation, planning and management (Serban 1990; Ghosh & Misra 2010). Numerous models are freely or commercially available all over the world and are used widely depending on the requirement of problems and needs of the users (Golmohammadi et al. 2014; Sood & Smakhtin 2015; Daniel & Abate 2022). Soil Water Assessment Tool (SWAT) is a hydrological model developed by the USDA (United States Department of Agriculture) Agricultural Research Service (ARS). It is a semi-distributed model widely adopted for large and complex river basins with different land use conditions, soil types and topography over a long period. The SWAT model is physically based, computationally efficient and capable of modelling long-term continuous simulation. One major advantage of the SWAT model is that it does not require as much calibration as other models. Hence the SWAT model was selected for this study (Jayakrishnan et al. 2005; Neitsch et al. 2009; Neitsch et al. 2011; Srinivasan et al. 2012).

Gurung et al. (2022) performed hydrological parameter characterisation for current and future climate scenarios with the help of the SWAT of the Myanmar river basins. Abeysingha et al. (2020) analysed the climate change impacts on future water availability and flow regimes in the Gomti river basin using the SWAT model under climate change scenarios from various GCMs. The SWAT model has been used in several studies to assess the impacts of climate change on hydrology, water availability and streamflow of the river basins (Aawar & Khare 2020; Abeysingha et al. 2020; Iranmanesh et al. 2021; Daniel & Abate 2022; Gurung et al. 2022). SWAT Calibration and Uncertainty programs (SWAT-CUP) is an auto-calibration software used for SWAT model calibration, validation and sensitivity analysis (Arnold et al. 2012; Singh & Saravanan 2022b).

Ponnaiyar river, which flows from Karnataka to Tamil Nadu, is the second largest river in Tamil Nadu, India. Since the main livelihood of the people in this river basin is agriculture, the amount of rainfall and its distribution plays a vital role in cropping patterns (CGWB 2017). It is necessary to investigate the change in streamflow and water availability of the Ponnaiyar river basin due to spatial and temporal variations of rainfall and temperature under future climate projections. This study aims to assess the impact of climate change on future water availability in the Ponnaiyar river basin using the SWAT model.

GCMs are used to understand the changes in rainfall patterns, temperature increase and sea level rise. Coupled Model Inter-comparison Project Phase 6 (CMIP6) GCMs dataset was used in this study and can be downloaded from https://esgf-node.llnl.gov/search/cmip6/. It has four shared socioeconomic pathway scenarios (SSP126, SSP245, SSP370 and SSP585) and target forcing at the end of the 21st century (Gidden et al. 2019). SSP126 resembles the low emission scenario which seems unrealistic and hence not considered. SSP245 resembles the low-mid emission scenario of RCP4.5, and SSP585 resembles the high-emission scenario of RCP8.5. Both SSP245 and SSP585 scenarios could cover the entire range from low-mid to high-emission scenarios and hence were selected for this study. 13 CMIP6-GCMs with bias-corrected rainfall, maximum and minimum temperature data were available for the historical (1850–2014) and future period (2015–2100) for the whole of India (Mishra et al. 2020). Sources and resolutions of the selected 13 CMIP6-GCMs are shown in Table 2.

When assessed at regional and local levels, many uncertainties are associated with the GCMs. Choosing suitable GCMs can help reduce the uncertainty in future climate projections. For selecting suitable GCMs, it is essential to evaluate their performance based on their agreement with the observed data. Selection of the best GCMs increases the confidence of GCMs used for impact assessment studies (Srinivasa Raju et al. 2017; Loganathan & Mahindrakar 2020; Raju & Kumar 2020). The historical simulations of 13 GCMs to reproduce the observed annual and seasonal time series of rainfall of the Ponnaiyar river basin are evaluated. The performance of the GCMs is evaluated using the performance metrics such as normalized root mean squared error (NRMSE), skill score (SS) and correlation coefficient (CC). Equal weights have been assigned to the performance metrics.

GCMs are the primary tool to estimate future climate patterns. There are a large number of GCMs available to predict future climate. The models are generally selected according to the availability, resolution or their applicability. In most cases, a single model is not sufficient to represent the climate characteristics. Hence, considering multiple GCMs could reduce the uncertainty associated with it. Many studies have selected the GCMs based on their availability Srinivasa Raju & Nagesh Kumar (2015) evaluated precipitation and temperature for 11 GCMs of CMIP3 repository for India as well as Krishna and Mahanadi basins. Panjwani et al. (2019) evaluated 12 CMIP5-GCMs for India for precipitation, minimum temperature and maximum temperature. Ahmed et al. (2019) evaluated 20 CMIP5-GCMs for precipitation over Pakistan. Hence, based on the availability of bias-corrected GCM data, 13 GCMs from CMIP6 were selected for this study. The GCMs were evaluated based on the historical annual and seasonal rainfall over the entire Ponnaiyar river basin. The historical simulations of the GCMs were compared with the IMD observed data for the period 1978–2015. There are many metrics developed for model evaluation based on different methods and thus, their effectiveness is different for different types and distributions of data. Therefore, multiple metrics are often used to measure the same property of a model (Iqbal et al. 2021). The performance of the GCMs is evaluated using the performance metrics CC, NRMS and SS. The performance evaluation of 13 GCMs using three indicators for annual rainfall is presented in Table 3. The 13 GCMs were ranked for each metric separately. CanESM5, MPI-ESM1-2-HR and NorESM2-MM models showed good performance for annual rainfall according to CC. Based on NRMSE and SS, BCC-CSM2-MR, EC-Earth3-Veg and MPI-ESM1-2-HR were the models with good performance. The cumulative ranks were calculated by considering equal weightage to all three metrics which is considered as a limitation of this study. To overcome this limitation in the future studies, an intensive evaluation of the performance metrics is required and weights can be assigned based on the significance of each metrics. Further validated research is required for assigning different weights based on the strength and importance of the metrics. This may be considered as future research work. Based on the annual rainfall, MPI-ESM1-2-HR, BCC-CSM2-MR and CanESM5 models were ranked first, second and third, respectively.

Simultaneously, the 13 GCMs were evaluated and ranked based on the seasonal rainfall over the study area. Four seasons, namely January–February (JF), March–April–May (MAM), June–July–August–September (JJAS) and October–November–December (OND), were considered in this study. The cumulative ranks for each season were calculated using the performance metrics for each season separately. MPI-ESM1-2-HR and NorESM2-MM models performed well for the JF season. The MPI-ESM1-2-LR model ranked first for the MAM season and second for the OND season. EC-Earth3-Veg ranked first for the JJAS season. The EC-Earth3 model performed well in MAM season (second rank), JJAS season (second rank) and OND season (first rank). Overall, EC-Earth3, MPI-ESM1-2-LR and MPI-ESM1-2-HR models ranked first, second and third, respectively. To increase the confidence over the future projections, more than one GCM can be used. Hence, the best three GCMs were selected for the impact assessment over the Ponnaiyar river basin. Selecting the optimal number of GCMs is yet another study which is beyond the scope of the work and can be considered in future work. The cumulative levels of the GCMs for annual and seasonal rainfall with overall ranks are listed in Table 4.

The calibration and validation of the SWAT model were carried out using observed river discharge data at two stream gauge stations, namely Gummanur and Vazhavachanur. SWAT-CUP under SUFI-2 optimization algorithm software was used to perform sensitivity analysis to determine the most sensitive parameters. Out of 25 tested parameters, 10 were selected based on t-stat and p-value, providing measures and sensitivity significance. The selected sensitive parameters based on their rank with their fitted value are shown in Table 5. These 10 parameters were further used to calibrate and validate the SWAT model. The results showed CN2, ALPHA_BF, SOL_K, GW_DELAY, GWQMN, CH_N2, CH_K2, ESCO, REVAPMN AND SOL_AWC are the most sensitive parameters associated with the hydrological processes and physical characteristics of the watershed. The curve number (CN2) is listed as the top-ranked parameter, representing the surface runoff and is mainly controlled by basin characteristics such as soil type, land use and management practices.

The calibration and validation results showed that the SWAT model well mimicked the stream discharge and provided satisfying outputs. The performance of the SWAT model is evaluated using statistical indices such as R², NSE, PBIAS and RSR. With the higher values of R², NSE and a lower percentage of PBIAS, the simulated discharge of the watershed has a good association with the observed values. Table 6 displays the calibration and validation results of two locations, namely Gummanur and Vazhavachanur.

SWAT model simulations were performed using the selected 3 GCMs, namely EC-Earth3, MPI-ESM-2-HR and MPI-ESM1-2-LR, for the future periods of the 2020s, 2050s and 2080s under SSP245 and SSP585 emission scenarios. Future simulation results were compared with the baseline period (1978–2015) to quantify the changes that would occur due to climate change. The results show a significant difference between the SSP245 and SSP585. The results (Table 7) indicated a change in streamflow for both selected locations within the basin for all three GCMs. The average annual streamflow increase at Vazhavachanur was estimated to be very high compared to the streamflow at Gummanur.

The changes in annual surface runoff are significant due to the changes in rainfall and temperature in the river basin. Even though the increase in rainfall will happen in the future, some models show that there may be a chance to get decreased surface runoff for some future period. The percentage changes in water balance components have been shown in Table 8. The annual surface runoff varied from −11.13 to 20.125%, −17.329 to −14.90% and −2.55 to 14.12% for all GCMs under SSP245 for the periods of the 2020s, 2050s and 2080s, respectively. Under the SSP585 scenario, the surface runoff varied from −20.41 to −15.46%, −10.51 to 18.34% and 73.88 to 134.56%. Hence, it is expected that the basin surface runoff will maximally increase up to 136.56% during the 2080s. For the future 2020s, the water yield varied from −7.02 to 11.36% and −1.41 to 6.15% for SSP245 and SSP585. During the 2050s and 2080s, the basin's water yield increased (7.89–21.18% and 36.12–115.25%) under SSP245 and SSP585 future climate scenarios. Changes in ET under SSP245 are predicted between −1.26 and 18.57%, −0.51 and 25.03% and 1.4 and 26.11%, while the increase in ET under SSP585 is expected to vary from 2.31 to 15.91%, 4.91 to 27.6% and 13.4 to 44.49% during the 2020s, 2050s and 2080s for the selected 3 GCMs. Potential ET shows variation between −0.63 and 0.72% and −0.32 and 27.55% during the 2020s and is also expected to increase up 0.154 to 2.08% and 0.2 to 3.08% during the 2050s and 2080s under both SSP245 and SSP585 scenarios.

This study evaluates the climate change impacts on water balance components of the Ponnaiyar river basin using the bias-corrected CMIP6 dataset. This is the first study to use the recently released CMIP6 projections to assess the impacts of climate change on water availability in Ponnaiyar river basin. This study examines the impacts of climate change on streamflow and water availability using the SWAT model. SWAT calibration and validation results have shown that the model performed well and reproduced the observed river discharge. NSE, R², PBIAS and RSR values were under acceptable ranges (NSE and R2 > 0.7, PBIAS > ± 25 and RSR < 0.6). We have used multisite single objective calibration for this study to improve the results. Further improvements could be achieved by multi-objective calibration. This is the limitation of this study and can be considered in the future research. The performance of the available 13 GCMs was evaluated with IMD observed data using the metrics CC, NRMS and SS. EC-Earth3, MPI-ESMI-2-LR and MPI-ESMI-2-HR are the three GCMs selected for climate change projections under SSP245 and SSP585 scenarios. The rainfall, maximum and minimum temperature data from the selected GCMs were used in the SWAT model for hydrological simulations for the 2020s, 2050s and 2080s, respectively. The maximum and minimum temperatures are expected to increase to 1.8 and 3.2 °C under future climate scenarios. According to the future rainfall projection, the maximum rainfall will likely occur during the 2080s and rise to 88% under SSP585. Similar to annual rainfall, the annual streamflow has been estimated to increase based on the simulation of selected GCMs under both scenarios. Though some future periods predicted a decreased annual streamflow, most of the periods are predicted to increase the annual streamflow at Gummanur and Vazhavachanur stations. The simulated increase in annual streamflow is varied from −3.60 to 28.53, −0.79 to 59.88 and 16.94 to 136.52% under different SSPs at Gummanur station; likewise, the annual streamflow changes are expected to increase from 13.16 to 25.60, 19.66 to 79 and 64.43 to 238.04% at Vazhavachanur station under future climate scenarios. Changes in rainfall and maximum and minimum temperature significantly impact the water balance components such as surface runoff, water yield, ET and PET. Based on the results, annual surface runoff is predicted to decrease by −14.90% during the 2050s under SSP245. Except for this period, all other future periods estimated the increased annual surface runoff up to 20.125% under SSP245. Given SSP585, the predicted annual surface runoff could be very high in the future and is expected to increase by nearly 135% during the 2080s. Like surface runoff, the annual water yield could rise to 115% under future climate scenarios. The annual ET is also expected to increase as much as 18.57, 25.03 and 26.11% under SSP245 and 15.91, 27.6 and 44.49% during the 2020s, 2050s and 2080s, respectively. The simulated annual PET has shown fewer changes in future climate scenarios during all future periods for the selected GCMs.

This study uses the CMIP6 dataset to simulate the water availability of the Ponnaiyar river basin. The change in future streamflow patterns and water balance components shows the potential impacts of climate change over the Ponnaiyar river basin. These results contribute to the policymakers, stakeholders, government officials and farmers for taking precautionary measures like proper planning in water management and agricultural practices and also developing adaptive strategies based on the future climate change impacts on the river basin. However, this study does not consider the future land use change. Hence, the future scope of the study is to assess the combined effects of future climate and land use change in the Ponnaiyar basin.

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

The authors declare there is no conflict.