2.1 Study area
Significant morphological changes occur throughout the season at the mouth of the Arakawa River in Niigata, Japan (Figure 1). As a general topographic trend in the Arakawa River mouth, the river mouth opens in Baiu from spring to summer because of high discharge events and closes in the winter season due to intermittent high waves (Ohizumi et al. 2021). Moreover, the opened river mouth affects the brackish environment due to saltwater intrusion, whereas the closed river mouth increases the flood risk since the sandbar causes the water level to rise. Consequently, maintenance of the river mouth must be flexible and adaptable to topographical features. Therefore, the Ministry of Land, Infrastructure, Transport, and Tourism (MLIT) has been carrying out maintenance, including regular monitoring and excavation of sandbars before the flood period to reduce the water level. Hence, it is important to understand the future characteristics of sandbars under climate change conditions in this area.
This study focused on the flood term caused by the rapid increase in discharge rate at the Arakawa River mouth in 2018 (Figure 1 (c)). In particular, the topography of the river mouth sandbar was differed from April 27 to May 21, which is illustrated by aerial photogrammetry. The maximum discharge rate of 2063.27 m3/s at 20:00 UTC on May 17 was observed at Tsudurayama stream gauging station in the Water Information System (WIS) of MLIT. This was because a heavy rainfall event occurred from May 17–19, which was caused by the Baiu front. Therefore, the MLIT conducted field surveys before and after a high discharge event at the mouth of the Arakawa River. In addition, land area was measured on April 27 and May 21 and sea area was measured on May 1, 2, 8 and May 29, 30, respectively. Field survey data were used as the initial boundary conditions for numerical simulations, and the numerical simulation term was set to approximately one month to fully cover the field survey period.
2.2 Model descriptions
Figure 2 shows the configuration of the numerical modelling system. In this study, we used the following three numerical simulation models: Weather Research and Forecasting model Hydrological modelling system (WRF-Hydro) (Gochis et al. 2015), Simulating WAves Nearshore (SWAN) (Booji et al. 1999) and eXtreme Beach behavior (XBeach) (Roelvink et al. 2009). In addition, these configurations do not consider interactions between WRF-Hydro, SWAN, and XBeach.
WRF-Hydro is a hydrological model expanded from the Weather Research and Forecasting (WRF) model (Skamarock et al. 2008), and simulates the physical processes of the atmosphere, land surface, and hydrology. SWAN is a third-generation wave model based on the wave action balance equation, which simulates the significant wave height (Hs) and mean wave period (TM01). XBeach is a process-based nearshore morphology model that considers hydrodynamic and morphodynamic processes. As a hydrodynamic process, XBeach has two representative modes: surf beat mode and nonhydrostatic mode. The surf beat mode was solved using the wave action balance equation and nonlinear shallow water equations. In contrast, morphodynamic processes were solved by the advection-diffusion equation for sediment transport and formulation for bottom updating.
Numerical simulations under the present climate were conducted, and is illustrated by the black arrows in Figure 2. First, we developed the initial boundary conditions using WRF Preprocessing System (WPS) based on GFS-FNL provided by the National Centers for Environmental Prediction (NCEP), and WRF-Hydro simulates hydro-meteorological conditions including wind velocity, precipitation, and runoff employed for XBeach modelling. Moreover, the elevation boundary condition is the topographical data obtained from the Japan Flow Direction Map (J-FlwDir; Yamazaki et al. 2018). Second, SWAN simulates the wave spectral density using the wind velocity forecasted by WRF-Hydro and bathymetry data obtained from GEBCO2019 (Sandwell et al. 2014). Finally, XBeach simulates the morphological changes at the Arakawa River mouth using the wave spectral density forecast by SWAN, observed data available to the public, and field survey data of MLIT. The discharge boundary condition was the discharge rate data of the Tsudurayama stream gauging station provided by the WIS. In addition, the tide boundary is the astronomical tide data of the Nezugaseki tide station located approximately 50 km northeast, which was provided by the Japan Meteorological Agency (JMA).
Numerical simulations under climate change were conducted as indicated by the red arrows in Figure 2. Although the initial boundary conditions are different, the procedure is the same as that for the hindcast simulation. Future simulations will calculate the external forces under climate change using the PGW method. First, we created the initial boundary conditions based on the climate change scenario, namely, the CMIP6 and CMIP5 multi-model ensembles. The PGW fields of surface temperature (ST), atmospheric temperature (AT), relative humidity (RH), and geopotential height (GPH) are used to calculate the meteorological data which are the boundary conditions for numerical simulations by WRF-Hydro. In addition, the sea level change was considered for the boundary conditions of the numerical simulations by SWAN and XBeach. Here, sea level change is assessed by the local sterodynamic sea level change based on ocean dynamic sea level change and global mean thermosteric sea level rise (Gregory et al. 2019). Second, discharge is calculated using the runoff obtained under the present and future climates. Finally, the morphological change under climate change at the Arakawa River mouth was evaluated by numerical simulations using XBeach, with the boundary conditions of tidal level, waves, and discharge estimated by future scenarios.
2.3 Boundary conditions under climate change
We conducted climate change impact assessments using CMIP6 models under the Shared Socioeconomic Pathways (SSPs) scenario. CMIP6 models were used because they have been reported to perform better than CMIP5 models (Kim et al. 2020; Fan et al. 2020). Furthermore, we also conducted climate change impact assessments using CMIP5 models under the Representative Concentration Pathways (RCPs) scenario. This was because RCP scenarios have been used in many previous studies. Each scenario was selected as SSP5-8.5 (SSP585) and RCP8.5 (RCP85), which are the most extreme scenarios. In previous studies, PGW sensitivity experiments under RCP8.5 which used the CMIP5 model were used for climate change impact assessments (Nakamura et al. 2016, 2020; Mäll et al. 2020; Nakamura and Mäll 2021). In the present study, numerical simulations under climate change conditions were conducted using the PGW method with 25 and 18 multi-model ensembles of CMIP6 and CMIP5, respectively. The PGW fields were constructed using the monthly mean value difference between the future (2081–2100) and the present period (2015–2024). The construction method of the PGW fields is same as that of Nakamura and Mäll (2021). Tables S1 and S2 show the CMIP6 and CMIP5 models used by each multi-model ensemble.
Figure 3 shows PGW fields of ST, AT, RH, and GPH in May. ST generally rises around Japan, and especially over land. AT, RH, and GPH evaluate at 500 hPa height. Generally, AT increases, RH decreases, and GPH increases throughout Japan. Specifically, as the mean of a plane distribution, AT increased by 4.591 and 4.033 °C in SSP585 and RCP85, respectively. Further, RH decreased by 1.983 and 1.540 % in SSP585 and RCP85, respectively. Meanwhile, GPH increased by 92.525 and 76.688 m in SSP585 and RCP85, respectively. Therefore, the planar trend of PGW fields between SSP585 and RCP85 is similar, although SSP585 shows more pronounced global warming than RCP85.
Sea level changes around the Sea of Japan were evaluated from sterodynamic sea level changes (Gregory et al. 2019). Formulation of the sterodynamic sea level change is as follows:
where ∆Z is the hemodynamic sea level change, ∆Z is the ocean dynamic sea level change, and is the global mean thermosteric sea level rise. This was obtained from the CMIP6 and CMIP5 data variables zos and zostoga. Furthermore, the evaluation range of the sterodynamic sea level change was same as the numerical simulation range of SWAN.
Figure 4 shows sea level change of each GCMs from 2015–2100 and difference between future (2081–2100) and present (2015–2024) periods in May. According to results of each multi-model ensemble, sterodynamic sea level around the Sea of Japan is exhibiting a rising trend. In addition, the planar trend of sea level change between scenarios also is similar as that of the PGW fields although the model bias of RCP85 is larger than that of SSP585. Further, the mean value of sea level rise is 0.33 and 0.39 m in SSP585 and RCP85, respectively, within the target area.
2.4 Setting numerical models
Sensitivity analysis was performed using different microphysics schemes in the physical models of WRF-Hydro. This is because many sensitivity studies on precipitation using WRF have illustrated the sensitivity of microphysics schemes to precipitation simulations (Liu et al. 2011; Avolio and Federico 2018; Mohan et al. 2018; Jeworrek et al. 2021). In this study, the microphysics schemes selected were WRF Single-Moment 5-class (WSM5; Hong et al. 2004), WRF Single-Moment 6-class (WSM6; Hong et al. 2006), and Thompson scheme (Thom; Thompson et al. 2008). Furthermore, hydrological simulation in WRF-Hydro is parameterized for various hydrological processes. Thus, this sophisticated study uses the parameter estimation tool (PEST) to automatically determine the parameter settings (Wang et al. 2019). However, the parameters were uniform in all cases since our study aims to elucidate changes in the present and future climate. Other physical models were simulated in a unified manner in the numerical simulations of WRF-Hydro, SWAN, and XBeach.
As the main setting of the numerical model, WRF-Hydro performs hydrological calculations in domain 3, which includes the Arakawa River basin. Furthermore, WRF-Hydro does not activate the baseflow model in order to explicitly show the runoff impact. XBeach activates the discharge option shown in the manual (XBeach Documentation, 2017). This allows for discharge from the river boundary in the XBeach model. In addition, this study did not consider sediment supply from the upper reaches of the river. For the simulation of XBeach, the grain sizes of D50 and D90 were set to 0.370 and 0.830 mm, respectively, which were obtained from a field survey (Ohizumi et al. 2021).
Table 1 lists the results of numerical simulations. In this study, cases with physical fields based on different microphysics schemes are denoted as WSM5, WSM6, and Thom. In addition, cases based on climate change scenarios are denoted as SSP585 and RCP85. Here, the numerical simulation of XBeach under future climate conditions evaluated the morphological responses based on tide level, waves, and discharge under climate change. Therefore, for case studies of XBeach, we conducted cases based on sea level rise (SLR), waves under climate change (WC), and discharge under climate change (DC). Details of the DC scenarios are described in the results under future climate conditions in Section 3.2.