Development of flood inundation maps for the Chaliyar Basin, Kerala under climate change scenarios

A flood is a high volume of water flow that rises and drowns the land which is otherwise not normally inundated. It is one of the most prevalent and widespread natural disasters in tropical nations like India, causing havoc on civil infrastructure and people's livelihoods (Mishra et al. 2018). It is widely reported that the climate is changing and that this change is already having an impact on every part of the world (Eum et al. 2011). As a result of climate change, the global average temperature is increasing, and the rainfall intensity and frequency have also increased. Consequently, the runoff and peak discharge have also increased, leading to flooding in many places. The impact of climate change on floods is severe in Indian river basins, especially in the low-lying floodplains, due to an increase in extreme rainfall events (Ramachandran et al. 2019). To identify the flood risk zones and to formulate effective flood mitigation and management strategies, it is crucial to map the possible flood inundation regions for a wide range of extreme rainfall events under climate change scenarios (Sahoo & Sreeja 2017; Hamdan et al. 2021; Kalra et al. 2021; Mohanty & Simonovic 2021).

The outputs of the general circulation models (GCMs) project future climate for different emission scenarios and are widely used in hydrological modelling to assess climate change's impact. Several studies have used the statistically downscaled GCM projections to simulate hydrological models and analyse the potential implications of climate change to a basin or sub-basin (Aich et al. 2016; Shrestha & Lohpaisankrit 2017; Nyaupane et al. 2018; Reshma & Arunkumar 2023). Similarly, flood inundation maps can also be developed by giving these GCM projections as input to an appropriate hydrological or hydraulic model. Hydrological models are simplified replications of the hydrologic cycle (Motovilov et al. 1999). Hydrological modelling at the river basin scale is essential to simulate the stream flows and predict floods under various climate change scenarios (Mohammed et al. 2018; Siddique & Palmer 2021; Wen et al. 2021). A few studies have been reported for mapping floodplains for future climate change scenarios (Shrestha & Lohpaisankrit 2017; Roy et al. 2021; Ukumo et al. 2022). El Alfy (2016) estimated the peak flows and abstraction losses by integrating the Hydrologic Engineering Center-Hydrologic Modelling System (HEC-HMS) model with geographic information system (GIS) to assess the flash floods. A similar study was conducted by El-Naqa & Jaber (2018) integrating the Hydrologic Engineering Center-River Analysis System (HEC-RAS) model and GIS to map the inundated areas along the main Wadi of Al-Ghadaf watershed for various return periods. It was reported the water level in some places in the inundated areas reached as high as 5 m. Abdessamed & Abderrazak (2019) used both HEC-HMS and HEC-RAS to assess the flood risk for different return periods. It was reported that the presence of retaining walls reduced the area from flooding, however, extremely low-lying regions were at risk of flooding. Using the MIKE 11 model, Rahman et al. (2011) investigated the flood flows and associated stages for various return periods for the Teesta sub-catchment in Bangladesh. The model results were used to create a stage–discharge relationship that was used to calculate the flood stages for 25-, 50-, and 100-year return periods.

Climate change has significantly altered the rainfall pattern and increased flood events across the world, especially in India. Therefore, it has become essential to map the potential flood inundation regions for various future extreme rainfall events and identify the flood risk zones for developing appropriate flood mitigation and management strategies. Madhuri et al. (2021) studied the flood depth, building risk and the effectiveness of various flood adaption measures for urban floods under climate change scenarios. For the Sebeya catchment, Assoumpta & Aja (2021) used quantitative precipitation forecasts to assess flood predictions. Roy et al. (2021) used open-source mathematical models to map the flood inundation of the Arial Khan River for the projected climate change scenario RCP 8.5. Ukumo et al. (2022) prepared a flood hazard map for the Woybo River catchment under climate change scenarios. The floods for future periods under different climatic scenarios were simulated using the HEC-RAS model. It was reported that 25.68% of the Woybo River catchment fell into the very high-hazard category, while 28.56% fell into the high-hazard category. From these studies, it is evident that flood mapping is mostly done by integrating hydrological and hydraulic models. Streamflows are simulated using a hydrologic model, whereas flood inundation areas and inundation depth are estimated using a hydraulic model. During the 2018 monsoon, Kerala experienced unusually heavy rainfall. As a result, 13 out of 14 districts experienced severe floods. The Chaliyar basin is one of the severely affected basins in Kerala due to extreme rainfall events and floods. Therefore, the main objectives of the study are to simulate the future streamflow of the basin for different climate change scenarios using a hydrological model and to develop flood inundation maps for different return periods under climate change scenarios using a hydraulic model. This study effectively combines the hydrologic and hydraulic modelling for mapping the flood inundation areas.

The study area and data are explained in the next section, followed by the methodology adopted for hydrologic modelling, extraction of future precipitation and hydraulic model development. Then, the results of various methods regarding streamflow simulation, performance indices, and flood maps for various return periods are explained. Finally, the conclusions from the results are explained in the last section.

The average annual rainfall in the basin is about 3,012 mm. The southwest (June–September) and northeast (October–November) monsoons are the two main rainy seasons. The main four soil types in the river basin are gravelly clay (60.73%), clay (24.56%), gravelly loam (9.85%), and loam (4.86%). Generally, the land use of the basin includes urban, rocky, and water bodies with agricultural land making up approximately 74.26% of the total area and 14.21% forest area. There is a hydrological observation station at Kuniyil and two meteorological stations in the basin located at Nilambur and Manjeri as shown in Figure 1.

The data collected for the study includes the observed daily precipitation for a period of 20 years from 1995 to 2014 from the Indian Meteorological Department (IMD). The digital elevation model (DEM) was obtained from the National Aeronautics and Space Administration (NASA) Earth data Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER), and observed daily streamflow (1995–2014) was obtained from Central Water Commission (CWC) through India-WRIS. The land use land cover (LULC) is obtained from United States Geological Survey (USGS) EarthExplorer. The precipitation and temperature projections for the future climate change scenarios were obtained from Mishra et al. (2020a). The statistical analysis of historical meteorological and streamflow data from 1995 to 2014 is given in Table 1. As seen from the table, a maximum of 571 mm was observed at Manjeri, whereas the difference in the mean of the two stations is less.

In the second phase, the bias-corrected climate projections developed by Mishra et al. (2020b) from Coupled Model Intercomparison Project-6 (CMIP6) were used in this study for future climate change scenarios. Among 13 GCMs, MPI-ESM-2-LR was selected based on the statistical analysis of the projections with the observed historical data. It has been reported that MPI-ESM-2-LR model precipitation gives good results for Indian climate conditions (Raju & Kumar 2020). To understand the variations in the streamflow, three future time periods, near future (2031–2040), mid-future (2051–2060), and far future (2071–2080) were considered. Flood frequency analysis was used to relate the magnitude of floods to their frequency of occurrence. The Gumbel extreme value distribution was used to develop the extreme flood events for 25-, 50-, and 100-year return periods (Ukumo et al. 2022).

In this study, the impact of climate change on streamflow and its subsequent flooding were assessed for various scenarios. The projected precipitation for different emission scenarios was used for the assessment. The HEC-HMS was used to simulate the streamflow for the near future (2031–2040), mid-future (2051–2060), and far future (2071–2080) under SSP2-4.5 and SSP5-8.5 scenarios. The simulated streamflow was then fed into the HEC-RAS model to create the flood maps for various return periods. The results of various models and assessments are discussed in the following sections.

The values of the model parameter were modified until the observed streamflow and computed streamflow were relatively close to each other. Both manual and automatic calibrations were performed to fine-tune the simulations of the HEC-HMS model. For automatic calibration, simplex and univariate methods were used. The final calibrated values of the model parameters are presented in Table 2. It is observed that the initial storage in the canopy loss, time of concentration and Muskingum parameters are significant streamflow simulations.

The extreme flood frequency was calculated using the Gumbel distribution. The return periods of 25-, 50-, and 100-year of streamflow were taken into consideration for the flood inundation modelling under SSP2-4.5 and SSP5-8.5 scenarios during the near future (2031–2040), mid-future (2051–2060) and far future (2071–2080) as shown in Table 4. The maximum flood is recorded during mid-future, SSP2-4.5 for a 100-year return period. The 100-year maximum flood is increased by 41.2% for the mid-future for the SSP2-4.5 scenario. The maximum flood shows an increasing trend for the mid-future with respect to the baseline period. Similar results were also reported by Kim et al. (2023) for the Hangang and Geumgang basins in South Korea.

In RAS Mapper, river banks, lines, and 2D mesh are first digitalized by using DEM. Cross-sections were created for every 1 km along the 32 km of the stretch. Following that, the geometric data were imported to HEC-RAS. Both the downstream and upstream boundary conditions are necessary for a mixed flow regime. By gradually changing Manning's coefficient from 0.03 to 0.08, the model was calibrated until the differences between the observed and the simulated values were within the acceptable ranges. When creating the flood inundation maps, the HEC-RAS model was set up to produce water levels and streamflow for 25-, 50-, and 100-year return periods. A total of 18 simulations were performed for different return periods and future periods. The output of the HEC-RAS model was viewed in the RAS Mapper before being exported to GIS to create flood inundation maps. The maximum water surface elevations (WSE) are tabulated in Table 5. SSP5-8.5 shows an increased WSE of 0.68 m than SSP2-4.5 during the near future for 100-year return period. SSP5-8.5 shows an increased WSE of 2.81 m than SSP2-4.5 during mid-future for a 100-year return period. Similarly, SSP5-8.5 shows an increased WSE of 2.77 m than SSP2-4.5 during the far future, 100-year return period.

In this study, the floodplains of the Chaliyar River basin are mapped using a hydrologic and hydraulic model for various climate change scenarios. Initially, a hydrologic model, HEC-HMS was developed to simulate the streamflow for various climate scenarios and time periods. Then, the frequency analyses were done to estimate the peak flows for different return periods. These peak flows were given as input to the HEC-RAS model to demarcate the floodplain regions. From the results, it is noticed that the HEC-HMS model has simulated the streamflow close to the observed values. The performance of the model is very good during calibration and good during validation. During the calibration of the HEC-HMS model, it is observed that time of concentration, initial deficit, maximum storage and constant rate are highly sensitive parameters as small change in the parameters affects the streamflow. For the near future and far future, simulated streamflow is higher for SSP5-8.5 than SSP2-4.5. But in the case of mid-future, simulated streamflow is higher for SSP2-4.5 than SSP5-8.5.HEC-RAS also has given good performance with R² at 0.63 and NSE at 0.61 during calibration. Vazhakkad, Kodiyathur, Kizhupparamba, Mavoor, Vilayil, and Cheekode are vulnerable areas to flooding. A maximum of 19.52 m WSE occurred at Kizhupparamba during mid-future for SSP2-4.5 and 18.38 m occurred at Cheekode during the near future for SSP5-8.5 for a 100-year return period.

Authors thankfully acknowledge the U.S. Geological Survey, India's Geo platform of ISRO (Bhuvan), Central Water Commission (CWC) and Indian Meteorological Society (IMD) for providing the necessary data required for this study.

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

The authors declare there is no conflict.